WhisPER v1.9.0
User Documentation
Authors: Simon Ciba, Fabian Brinkmann, and Alexander Lindau.
2
This document was originally created by Simon Ciba while being a student at the Audio
Communication Group, TU Berlin, and completed in its first issue in November 17th, 2008.
It is part of the software package WhisPER which has been developed by Andr´e Wlodarski
and Simon Ciba within the scope of their master’s theses conducted at TU Berlin.
Currently, this document is maintained by Fabian Brinkmann and Alexander Lindau (Audio
Communication Group, TU Berlin).
Additions to both this document and the software package WhisPER have been contributed
by both Fabian Brinkmann, and Alexander Lindau, as well as by master students Matthias
Herder, Andreas Rotter, Johannes Blickensdorff, and Martina Vrhovnik.
Note: The authors does not assume any liability for the correctness of this document’s content
and further takes no responsibility for any harm or damage resulting from its application.
Date of document version: Feb. 5th, 2015.
WhisPER v1.9.0 – User Documentation
Contents 3
Contents
Listofabbreviations................................... 5
1 Introduction 7
1.1 What is WhisPER?................................ 7
1.2 History, Versions and Availability . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3 FoldersandFiles .................................. 7
1.3.1 The top-level folder ‘whisper X.Y.Z ’ . . . . . . . . . . . . . . . . . . . 7
1.3.2 The folder ‘1 literature’ . . . . . . . . . . . . . . . . . . . . . . . . . . 8
1.3.3 Thefolder‘2mfiles’ ............................ 8
1.3.4 Thefolder‘3doc’ ............................. 8
1.4 LicenseTerms.................................... 8
1.5 Warranty and Safety Information . . . . . . . . . . . . . . . . . . . . . . . . . 8
2 A Short Characterization of the Program’s Features 9
2.1 BasicConcepts ................................... 9
2.2 Implemented Test Procedures . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.3 System and Software Requirements . . . . . . . . . . . . . . . . . . . . . . . . 11
3 Running the Program 13
4 The Main Window 14
4.1 MenuBar...................................... 14
4.2 Desktop ....................................... 15
5 Setting up a test series 17
5.1 First Step: Creation of a New Test Series . . . . . . . . . . . . . . . . . . . . 17
5.2 Second Step: Definition of Stimuli . . . . . . . . . . . . . . . . . . . . . . . . 17
5.3 Third Step: Configuration of Session Setup . . . . . . . . . . . . . . . . . . . 21
5.4 Fourth Step: Specification of Test Sections . . . . . . . . . . . . . . . . . . . . 21
5.5 Fifth Step: Setting up Procedure Properties . . . . . . . . . . . . . . . . . . . 22
6 Performing a Test Run 23
6.1 Initializing and Starting a Test Run . . . . . . . . . . . . . . . . . . . . . . . 23
6.2 The Test Subject’s GUI System . . . . . . . . . . . . . . . . . . . . . . . . . . 23
7 Test Procedures – Configuration and Execution 26
7.1 Adaptive Psychophysical Threshold Procedures (AP) . . . . . . . . . . . . . . 26
7.1.1 TheTestRun................................ 26
7.1.2 Setting up Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
7.1.2.1 Staircase ............................. 30
7.1.2.2 PEST............................... 32
7.1.2.3 ML/Bayesian procedure . . . . . . . . . . . . . . . . . . . . . 34
7.2 Repertory Grid Technique (RGT) . . . . . . . . . . . . . . . . . . . . . . . . . 40
WhisPER v1.9.0 – User Documentation
4Contents
7.2.1 TheTestRun................................ 40
7.2.2 Setting up Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
7.3 Semantic Differential (SD) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
7.3.1 TheTestRun................................ 48
7.3.2 Setting up Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
7.4 ABX Detection Test (ABX) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
7.4.1 TheTestRun................................ 54
7.4.2 Setting up Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
7.5 Spatial Audio Quality Inventory (SAQI) . . . . . . . . . . . . . . . . . . . . . 58
7.5.1 TheTestRun................................ 58
7.5.2 Setting up Properties (Fig. 7.31) ...................... 64
7.5.3 How to: Implementing SAQI for a New Language . . . . . . . . . . . . 68
7.6 ITU-R Rec. BS.1116-1 (ABC/HR) & ITU-R Rec. BS.1534-1 (MUSHRA) . . 70
7.6.1 TheTestRun................................ 70
7.6.2 Setting up Properties (Fig. 7.38) ...................... 71
8 Test procedures - Default Values’ Presets 73
9 Program Setup 78
9.1 BasicAudioSetup ................................. 78
9.2 BasicNetworkSetup................................ 78
10 Data Handling 79
10.1TheTestSeriesFolder............................... 79
10.1.1 The File ‘testseries.info’ . . . . . . . . . . . . . . . . . . . . . . . . . . 79
10.1.2 The Files ‘TSD.mat’ and ‘TSP.mat’ . . . . . . . . . . . . . . . . . . . 80
10.1.3 The Subfolder ‘audio’ . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
10.1.4 The Subfolder ‘export’ . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
10.1.5 The Subfolder ‘log files’ . . . . . . . . . . . . . . . . . . . . . . . . . . 88
10.1.6 The Subfolder ‘plots’ . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
10.1.7 The File ‘PS.mat’ (“program settings”) . . . . . . . . . . . . . . . . . 88
Bibliography 89
List of Figures 92
WhisPER v1.9.0 – User Documentation
List of abbreviations 5
List of Abbreviations
AP adaptive psychophysical procedure
DW dialog window (used in a generalized meaning, cp. annot. 3 on p. 14)
GUI graphical user interface
ID identification
LAN local area network
ML maximum likelihood
OSC open sound control
PEST parameter estimation by sequential testing
QUEST quick estimation by sequential testing
RGT repertory grid technique
SD semantic differential
ZEST zippy estimation by sequential testing
SAQI Spatial Audio Quality Inventory
WhisPER v1.9.0 – User Documentation
6List of abbreviations
WhisPER v1.9.0 – User Documentation
7
1 Introduction
1.1 What is WhisPER?
WhisPER is a package of scripts running under MATLABrfor performing listening tests
in the field of auditory perceptual research, thereby, WhisPER is designed for maximum
usability. Hence, most of the handling can be done by usage of a coherent graphical user
interface (GUI) system. The aim of WhisPER is to provide experimenters in the auditory
field with a selection of the most popular listening test procedures while allowing for a con-
venient automation with respect to user interactions and stimulus playback. For an overview
of the currently implemented listening test procedures see sec. 2.2.
1.2 History, Versions and Availability
WhisPER was developed by Andr´e Wlodarski and Simon Ciba within the scope of their
master’s theses (Ciba [2008], Wlodarski [2008]) at the Audio Communication Group of the
TU Berlin, in 2008. Today, WhisPER is continuously developed by the staff of the Audio
Communication group. So far, updates included feature upgrades and the extension by mul-
tiple new test methods. Information about the history of software updates and the latest
version can be read from the file release notes.txt (see sec. 1.3.1).
Since its version 1.8.0 WhisPER is freely distributed using the electronic data repository
DepositOnce of the Technische Universit¨at Berlin. DepositOnce allows persistent identi-
fication of the current WhisPER version using the DOI (Digital Object Identifier) http:
//dx.doi.org/10.14279/depositonce-31. A more informative project website can be reached
from http://www.ak.tu-berlin.de/whisper.
1.3 Folders and Files
The current version of WhisPER maybe freely dowloaded from DepositOnce (see 1.2) as a
self-contained zip-file. After unpacking you will find four folders and some informative text
files on top level. In the following, the folder and file structure will be shortly explained.
1.3.1 The top-level folder ‘whisper X.Y.Z ’
The name of the top level folder contains the current WhisPER version number. Small
changes in WhisPER (such as, e.g., bugfixes) are reflected by a changes of the rightmost
version number (”Z”). Larger updates, such as, e.g., the implementation of new listening
test procedures are reflected in a incrementation of the midmost version number (”Y”). The
WhisPER v1.9.0 – User Documentation
81. Introduction
leftmost number (”X”) has not changed since WhisPER’s first release and is reserved for
indicating in-depth changes of the whole software.
The top level folder of the WhisPER distribution hosts the file release notes.txt which con-
tains information on the history of updates and the current software version. The file li-
cense notes.txt contains the license information, while Copying.txt contains a copy of the
GNU General Public License. The file authors.txt contains the names of all those, who
contributed to WhisPER so far.
1.3.2 The folder ‘1 literature’
This sub folder contains some publications giving background information on the implemented
procedures or on the development of WhisPERitself.
1.3.3 The folder ‘2 mfiles’
This sub folder contains the main WhisPER installation in the form of a number of MATLABr
scripts, figures, and data files. In order to execute WhisPER you will have to add this folder
to MATLABr’s path and run ’whisper’ from the MATLABrprompt (see 3).
1.3.4 The folder ‘3 doc’
This sub folder includes the user documentation, i.e. the document you are currently reading,
and a technical documentation, intended to be helpful when aiming at extending WhisPER’s
functionalities. If you plan to extend WhisPER– which is of course greatly appreciated – it
might be a good idea to get in contact with the Audio Communication Group before.
1.4 License Terms
For license information consult the file license notes.txt (see sec. 1.3.4) which is to be found
in the top level folder of the WhisPER installation.
1.5 Warranty and Safety Information
The use of the program – or any of its components (including this document) – for whatever
purpose, is under the complete and sole responsibility of the user. The authors of WhisPER
do not extend any kind of warranty.
Special attention is drawn to the fact that WhisPER does not comprise an automatic mech-
anism to prevent the delivery of too high sound pressure levels. It is the explicit responsibility
of the user to make sure that the stimuli in use, in combination with any further playback
equipment (sound card, amplifier, headphone or loudspeaker, etc.), will yield a sound pressure
level, that will not cause any damage to anyones health, especially the hearing system.
WhisPER v1.9.0 – User Documentation
9
2 A Short Characterization of the Program’s Features
2.1 Basic Concepts
In empirical auditory research experimenters often want to administer a number of different
listening tests to a number of ever identical subjects while the individual tests differ only
slightly in specification (factorial within-subjects test design). Hence, in order to simplify the
administration of such tests the user interface system of WhisPER is designed to allow a
central control of multiple test procedures from one point of the system. For this purpose
the program provides the administration of all data – including the empirical test results –
from within a virtual environment called test series, which is structurally represented by one
conjoint data folder (see sec. 10.1).
Moreover, WhisPER allows performing several successive – and even interleaved – sub-tests
within one test run while applying different test procedures. This feature tries to meet some
specific requirements of experimental research. For example, it is often desired to carry out a
familiarization experiment immediately before starting the actual listening test. Likewise, in
adaptive threshold estimation procedures, researcher often want to interleave adaptation runs
for different test conditions in order conceal the actual stimulus variation this way aiming at
a suppression of anticipation effects in test subjects.
Empirical data export to applications for statistical analysis such as, e.g., SPSS or MS-Excel
is feasible via automatically generated formatted results text files (*.csv format). Moreover,
data from repertory grid experiments (see sect. 2.2) may be imported into specialized software
packages such as Idiogrid (http://www.idiogrid.com/) and Gridsuite (http://www.gridsuite.
de/) through support of their respective native results data formats GridXML and GRD (see
sec. 10.1.4).
2.2 Implemented Test Procedures
In the present state, WhisPER provides about seven different listening test procedures. In
the following, these procedures and their basic features will be briefly reviewed together with
some essential literature. More detailed information on the first three procedures (AP, RGT,
and SD) can be found in the master theses of Simon Ciba (Ciba [2008]), and Stefanie Otto
(Otto [2008]), see also 1.3.2.
Adaptive Psychophysical Threshold Procedures: A group of test procedures that aim at
achieving efficient estimates of a single point on a psychometric function that is called
a psychophysical threshold. This is achieved by adapting the intensity of the stimu-
lus according to the history of preceding responses of the subject. There are different
adaption mechanisms and method for estimating the final threshold from the observed
response pattern. WhisPER currently supports simple and transformed staircase pro-
cedures with the adaption rules 1-down/1-up, 2-down/1-up and 3-down/1-up, halving
WhisPER v1.9.0 – User Documentation
10 2. A Short Characterization of the Program’s Features
of step sizes after a predefined number of reversals and a calculation of the threshold
estimate by averaging stimulus intensities over a defined number of last reversals.
Furthermore, more sophisticated adaptation methods can be chosen from, such as e.g.
parameter estimation by sequential testing (PEST), and a modified version of PEST re-
ferred to as More Virulent PEST. Moreover, maximum likelihood and Bayesian adaption
methods, such as, e.g. the Best PEST have been implemented. Finally, QUEST and
ZEST procedures have been implemented, too.
References:
•Overview: Treutwein [1995], [Gescheider, 1997, pp. 159], Leek [2001]
•Staircase: Cornsweet [1962], Wetherill [1963], Levitt [1971],
•PEST: Taylor and Creelman [1967]
•More Virulent PEST: Findlay [1978]
•ML/Bayesian procedure: Pentland [1980], Lieberman and Pentland [1982], Treutwein
[1995], Treutwein [1997]
•QUEST:Watson and Pelli [1983]
•ZEST: King-Smith et al. [1994]
At present, WhisPER allows employing these adaption procedures in detection and
discrimination tasks while using n-alternative forced choice (nAFC) response paradigms
for n=2, 3 or 4 alternatives. Furthermore, arbitrary adaptation tracks using the same
response paradigm may be randomly interleaved.
Repertory Grid Technique (RGT): RGT is a combination of auditory attribute elicitation and
scaling used for the sensory evaluation of complex stimuli. During the first part of the
procedure, stimuli (called elements) are combined to groups of three (called triads) and
presented to the subject which is supposed to report on similarities and contrasts in a
qualitative manner. The result of each triad comparison is a pair of verbal expressions
that are likely to represent the end-point descriptors of a bipolar attribute referred to
as a construct. In the second part of the procedure all elements are rated on bipolar
scales which were formed by using the elicited constructs.
The formation of triads can be carried out manually by the experimenter or automat-
ically by the program by creating a complete variation. Furthermore, the order of
presentation of triads and elements can be defined either by the user or be generated
randomly. Elicited constructs may be edited by the experimenter in between the two
parts of the procedure.
References: [Bech and Zacharov, 2006, 54], Berg and Rumsey [1999], [Bortz and D¨oring,
2006, 187], Fransella and Bannister [1977], Kelly [1955]
Semantic Differential (SD): The Semantic Differential is a procedure for rating (direct scal-
ing) the perceived amount of a certain auditory quality while being exposed to a specific
sound stimulus. Arbitrary sets of bipolar rating scales maybe defined for forming a Se-
mantic Differential. In WhisPER, the experimenter may define the presentation order
of objects or he might choose a randomly generated sequence.
References: Osgood et al. [1967],[Bortz and D¨oring, 2006, pp. 185],
WhisPER v1.9.0 – User Documentation
2.3 System and Software Requirements 11
ABX-Test: The ABX-test (also: duo-trio test) is a method for assessing the detectability
of subtle differences between two stimuli. In audio technology research ABX-testing
gained some publicity when being used for the controlled assessment of differences be-
tween (valve and transistor type) amplifiers or similar devices/technologies which are
typically expected to exhibit only very small perceptual differences.
During an ABX-test, two stimuli A and B are to be directly compared to a third stimu-
lus X which is deliberately chosen to be either A or B. The subject’s task is to identify
whether X equals stimulus A or B. The required number of trials (repetitions) reflecting
a certain type-1 error level, effect size and test power may be calculated based on the
binomial distribution. In WhisPERthe procedure is implemented as a double blind
test in order to prevent investigator effects.
References: Clark [1991], Leventhal [1986], Burstein [1988]
Spatial Audio Quality Inventory: The Spatial Audio Quality Inventory is a collection of qual-
itative descriptors which may be used for a differentiated, comparative auditory assess-
ment of real, imagined and simulated acoustic environments in order to reveal specific
shortcomings of a simulation under test and allow for a directed technical improvement.
The SAQI comprises 48 verbal descriptors of perceptual qualities assumed to be of prac-
tical relevance in the scope of virtual acoustic environments. The SAQI is implemented
as a widely customizable questionnaire-type listening test mostly operationalizating its
items with continuous or categorical rating scales or multiple choice check boxes. Ad-
ditionally, since WhisPER vs. 1.9.0 the SAQI is available as a multiple-stimulus test,
too. Hence, multiple stimuli may be assessed while referring to one SAQI item utiliz-
ing panels with multiple ratings scales. SAQI is further available in a German and an
English version, a French version is currently under development.
References: Lindau et al. [2014a], Lindau et al. [2014b], Lindau [2015]
ITU-R Rec. BS.1116-1 (ABC/HR) & ITU-R Rec. BS.1534-1 (MUSHRA): ABC/HR and MUSHRA
are methods for rating multiple conditions against one or more references for instance
in terms of difference (ABC/HR) or degradation (MUSHRA). ABC/HR includes for
each tested condition an individual detection task according to the ABX or duo-trio
paradigm (see above) and should be used when differences between test and reference
conditions are expected to be small. In contrast, MUSHRA is typically recommended
for the assessing intermediate sensory differences.
References: ITU [1997], ITU [2003], Bech and Zacharov [2006]
2.3 System and Software Requirements
Operating system WhisPER was developed to be executed on a 32-bit version of Windows
Vista. It has also been tested on a 32-bit version of Windows XP. Usage of the soft-
ware under different operating systems like Linux or OS X has generally been enabled
although not extensively been tested.
Running environment (MATLABrversion) Implementation of the software package was done
on MATLABrversion 7.4.0.287 (Release 2007a), additions have been made using
MATLABrversion 7.9.0.529 (Release 2009b). Version 1.9.0 of WhisPER was tested
to run on MATLABr2013a. It might be possible that earlier versions of MATLABr
WhisPER v1.9.0 – User Documentation
12 2. A Short Characterization of the Program’s Features
allow using WhisPER, too. However, it is known that it at least a MATLABrfrom
v7.0 upwards is required (not tested, though).
Important note: Since vs.1.8.0 the OSC (Open Sound Control) remote control capa-
bilities of WhisPER are realized by using Mark Marijnssen’s script oscsend.m which
is freely available from MATLABrCentral and which depends on MATLABr’s In-
strument Control Toolbox.
This implies that - for full OSC-capabilities - WhisPER will now require
the Instrument Control Toolbox (included in MATLABr’s student version)
to be pre-installed.
Display resolution The GUI system requires a display resolution of at least 1280x800 pixels.
Playback environment Presentation of the acoustical stimuli can be realized by playing .wav
files from a Direct Sound-compatible sound card 1. Another option is to send OSC
messages via network protocol to some external audio player application (for instance
puredata). This way, any type OSC-compatible playback periphery can be controlled.
Per stimulus event, one may send up to 6 independent OSC commands simultaneously,
allowing controlling a number of different playback devices (acoustical, optical, opto-
acoustical, etc.)at once 2.
1This applies to the Windows operating system. On other operating systems make sure MATLABr’s
audioplayer-object does work with your sound card.
2Technically speaking, MATLABrexecutes the six OSC-commands in sequence. However, normally this
proceeds very quickly unless the execution is not purposely delayed by difining a number of suitable delay
times.
WhisPER v1.9.0 – User Documentation
13
3 Running the Program
To start WhisPER, there have to be the following three steps taken in given order:
1. Start MATLABr.
2. Define the path of the program folder ‘2 mfiles’ as the current (working) directory
in MATLABr. This can either be done by using the Current Directory Browser or
using the function cd (change directory) by typing cd(’PathofProgramFolder’) into the
command window and afterwards pressing return.
3. Start WhisPER by typing whisper into the MATLABrcommand window and after-
wards pressing return.
After the conduction of these steps the GUI system’s main window will appear which is
labeled as WhisPER. It will be described in the next section.
WhisPER v1.9.0 – User Documentation
14 4. The Main Window
4 The Main Window
The starting point of any action is the GUI system’s main window which is labeled as Whis-
PER and appears right after starting the program as described in chapter 3. All functionality
is available directly or indirectly from this part of the user interface system. Some functions
can be addressed by using the menu bar, some by using control elements placed on the main
window’s desktop and some in both ways. The following two subsections supply an overview
and a short description of the functionality contained in the main window.
Figure 4.1: The main window.
4.1 Menu Bar
This subsection describes the items in the main window’s drop-down menu bar.
Main window (menu bar)
test series
new Opens a dialog window (DW)3for creating a new test series. You have to specify
name and location of a folder which will be created and later contain all the data
3This expression is used here in a generalized meaning, i.e. for any window of the GUI system that allows
for the information flowing in both directions – from the user to the program and vice versa.
WhisPER v1.9.0 – User Documentation
4.2 Desktop 15
of test runs referring to a common research issue (cp. sec. 2.1). The inherent
structure of this folder in terms of sub folders and files included will be described
in sec. 10.1.
open Opens a DW for opening an existing test series by browsing for the belonging
test series folder.
close Closes the current test series. Note: Every time one makes changes on settings,
those will be automatically saved after clicking on the OK button on the respective
window. That means that one does not have to save a test series’ data before
closing it.
export empirical data Manually creates the empirical data export file(s) (called export
sheets) in the sub folder export (see sec. 10.1.4).
exit program Closes the program.
edit
stimulus pool Opens a DW for the definition of stimuli (see sec. 5.2).
session setup Opens a DW for making global specifications necessary for the use of
multiple sessions. At the present state this is the number of sessions that is
required for the current test series. By default this value is set to 1.
test sections Opens a DW for selecting the desired test procedures, editing test sections
(see sec. 5.4) and setting up procedure parameters on a further DW (see sec. 5.5).
empirical data Opens a DW for deleting the empirical data of the latest completed test
run.
procedure’s defaults Opens a DW for configuring the default values of procedures as
they are initialised and added to a test series (see sec. 8).
program setup
audio <disabled, cp. sec. 9.1>
network Opens a DW for adjusting basic settings of the network environment (see
sec. 9.2).
plotting Opens a DW for adjusting settings referring to the data plots’ file format.
?
help Opens a pdf-document (if acrobat reader is available) containing the user docu-
mentation at hand.
about Displays information about the version of WhisPER currently used.
4.2 Desktop
In this subsection the control elements placed on the main window’s desktop are listed and
described.
Main window (desktop)
-no test series loaded-/name of current test series (dynamic superscription) Indication of
the test series currently loaded.
edit test series (panel)
stimulus pool (button) see sec. 4.1, menu item edit\stimulus pool
WhisPER v1.9.0 – User Documentation
16 4. The Main Window
session setup (button) see sec. 4.1, menu item edit\session setup
edit procedures (button) see sec. 4.1, menu item edit\test sections
next test run (panel)
run parameters (sub-panel)
subject id (input field) Space for entering the ID of the subject that is going to
take part in the next test run. The input is treated as a string variable.
session (input field) Space for entering the integer key that represents the session
key number which should be assigned to the next run.
run (output field) Displays the count number (i.e. the value of the respective
counter) of the next run.
execution of test sections (sub-panel)
all (radio button) If selected, all test sections (see sec. 5.4) will be performed during
the next test run.
selection (radio button) If selected, only a specified subset (see button edit) of
test sections will be performed during the next test run.
edit (button) Opens a DW for determining a subset of test sections to be performed
during the next test run. This button is active only if the second of the
preceding options has been selected.
start run (button) Starts the next test run and opens the GUI system for subject guid-
ance.
latest completed test run (panel)
run (output field) Displays the count number of the latest completed test run.
WhisPER v1.9.0 – User Documentation
17
5 Setting up a test series
This chapter describes how to set up a new test series from the very beginning. Additionally,
test series may be defined only in parts by carrying out only a few of these steps. Changes
will be saved instantaneously after confirmation by pressing the respective OK button. In
most cases dialog windows contain a cancel button allowing to discard changes made in the
current window.
Note that changes on the test series setup are only possible as long as the counter indicating
the coming test run shows the value 1. Otherwise, if you are still willing to make changes
on the setup, you have to reset the counters by deleting the total empirical data set which
is stored in the file TSD.mat contained in the current test series folder (see sec. 10.1). This
may either be done by deleting (or removing) the whole file at once or by using the item
edit\empirical data from the menu bar which allows for deleting data of test runs one after
the other. Note that in the first case you will have to restart the program so that the action
will take effect on the program’s current status. Additionally it is recommended to clear the
contents of the sub folders plot,export and logfiles manually. Otherwise this could lead to
confusion as old and new files will be mixed and old files might be gradually overwritten.
However, do not forget to make a copy of the folders’ contents if you want to preserve the
data.
5.1 First Step: Creation of a New Test Series
The creation of a new test series is done by selecting the corresponding item from the menu
bar (see sec. 4.1). Then a DW will appear for specification of name and location of the new
test series folder which will be automatically created including some sub folders and files.
The folder’s inherent structure will be described in sec. 10.1.
5.2 Second Step: Definition of Stimuli
Before stimuli can be chosen for application in certain test procedures, they have to be defined
on a global level. For this purpose use the button stimulus pool (see sec. 4.2) or the item
edit\stimulus pool from the menu bar (see sec. 4.1) to access a DW labelled as stimulus pool
(see fig. 5.1), which contains a list providing an overview on the stimuli which have been
defined already and their belonging playback properties. The supplied information for each
stimulus consists of an automatically generated ID, a user-defined label (name), if applicable
the name of a .wav file or a set of OSC commands or both.
New stimuli can be added by using the button new or multi new. While the former will only
generate an empty stimulus in the list which later will have to be edited in order to be useful,
the latter will open a file selection dialog window allowing multiple selection of .wav files. For
WhisPER v1.9.0 – User Documentation
18 5. Setting up a test series
Figure 5.1: DW stimulus pool
each .wav file selected a separate stimulus associated with that .wav file will be added to the
list 4. Names will be auto-generated from the file names, but of course can be edited later5.
Unwanted stimuli can be deleted by using the delete button 6.
Stimuli can be sorted using the move up and move down button, and the order the stimuli
are in here will be the order the stimuli will be presented to you in all the other DWs when
there are stimuli to be selected from the global stimulus pool.
If one wants to test if a selected stimulus’ properties (see below) have been set well, one can
use the button test stimulus for a presentation of it on a trial basis.
The editing of a stimulus’ properties is performed by selecting the concerning position of the
list and then using the edit button, which will open a suitable DW labeled as edit stimulus
whose control elements will be described below. On this window definitions can be made
regarding the stimulus’ label (name), duration and content. The latter can be chosen either
as a .wav file or a set of up to 6 OSC commands or even both at a time. That means that
both kinds of playback environment, auditory and visual, can be addressed. If you choose a
.wav file from any location of your data drive, it will be copied to the folder audiofiles, which
is a sub folder of the test series folder (see sec. 10.1). The idea behind this behavior is to be
able to easily transfer a test series’ configuration data set between computers with a minimum
amount of adjustments necessary. One restriction resulting from this is that the name of a
chosen .wav file must be unique and not allocated to several files. When uploading a .wav file
the stimulus’ duration will be automatically determined as the file’s length. The stimulus’
4For techniques of economically configuring multiple OSC stimuli instead please cf. sec. 8.
5Auto-generation of names from file names will be suspended however, if the default stimulus name has been
changed, cf. sec. 8.
6While it is possible to delete global stimuli which have been assigned inside procedures (see sec. 7), you
are highly encouraged to follow good practice and separate global stimulus setup from procedure stimulus
assignment and first perform the one, afterwards the other.
WhisPER v1.9.0 – User Documentation
5.2 Second Step: Definition of Stimuli 19
name will also be auto-generated 7. If playback of stimuli is triggered/involves additional
via OSC commands, the stimulus’ duration should be edited or must be manually entered
respectively (the latter is the case if no .wav file is chosen). This is necessary because the
software will receive no feedback information about the termination of playbacks controlled
by OSC. As a consequence the entered duration period has to be long enough to include all
presentations of the different contents, i.e. possibly a combination of a .wav file and several
OSC commands to be executed at a time.
Figure 5.2: DW edit stimulus
7This does not apply if the name has previously been altered from the default! Cf. also sec. 8
WhisPER v1.9.0 – User Documentation
20 5. Setting up a test series
DW “edit stimulus”
stimulus properties (panel)
name (input field) Space for entering the name or a short description of the stimulus
respectively. The input will be treated as a string. This should release the user from
distinguishing the selected stimulus from the remaining ones during the process of
setting up test procedures. If this field has not yet been altered from the initial
default and you select a .wav file by way of the appropriate dialog, this field
will be auto-filled with a name generated from the .wav file minus its extension.
Which name is displayed here upon stimulus creation as a default name can also
be manipulated, cf. sec. 8.
ID (output field) Displays the stimulus’ ID.
duration in sec (input/output field) In this field either the duration of the stimulus in
[sec] can be manually entered or will displayed when received by using the get from
file button.
get from file (button) Obtains the stimulus duration from a .wav file if there has been
one already specified in the file field.
file (input/output field) In this field the original path of a .wav file which shall be played
back directly from the sound card as part of the stimulus’ audio content, either
will be manually entered or displayed if received by using the file dialog which can
be accessed by using the browse button.
browse (button) Opens a file dialog for selecting the path of a .wav file which shall be
played back directly from the sound card as part of the stimulus’ audio content.
When leaving the current DW (stimulus properties) and confirming changes by
pressing the OK button, the selected file will be copied to the folder audiofiles
which is a subfolder of the test series folder (see sec. 10.1). After this the file field
will no longer contain the path but only the file’s name.
listen (button) Plays the .wav file if one has been specified. This function can be used
for identification and calibration purposes.
stop (button) Stops the .wav file’s playback.
rate [Hz] (output field) Displays the sampling rate in [Hz] if a .wav file has been speci-
fied.
bit depth (output field) Displays the resolution of the quantization process in [bit] if a
.wav file has been specified.
Server 1–6 (panel) Here a set of up to 6 OSC commands can be determined with each
addressing one of up to 6 servers which have to be specified globally (see sec. 9.2).
The commands will be executed in sequence with the order given by their ascending
label numbers.
pre-send pause (input field) Space for entering a time period in [sec] that lies
between the previous and the current command’s execution.
osc path (input field) Space for entering the path of the parameter one wants to
address.
data type (drop-down list) Here one can choose among five types of data formats
which are OSC-string8,int32,float32,True and False. For a description of
these the user is directed to e.g. http://opensoundcontrol.org/spec-1 0.
8A sequence of ASCII coded characters terminated by zero (C-String), filled up with zeros to a length of an
integer multiple of four.
WhisPER v1.9.0 – User Documentation
5.3 Third Step: Configuration of Session Setup 21
osc data (input field) Space for entering the data of the format which has been
previously specified. This field will be disabled if either True or False are
selected.
5.3 Third Step: Configuration of Session Setup
In order to be able to perform and control different test runs under several experimental
conditions within one test series (cp. sec. 2.1), a structural variable called session has been
defined. Each subset of runs under one equally determined set of experimental conditions
is thereby attached to a specific value of this variable. The variable’s range starts at 1
and extends by integer steps to its maximum which is called number of sessions. Before the
different experimental conditions can be specified for each test section, the number of sessions
has to be determined as a global parameter. This can be done by using the button session
setup placed on the main windows’ s desktop (see. sec. 4.2) or selecting the item edit\session
setup from the menu bar (see sec. 4.1). Thereupon a suitable DW will open. By default the
number of sessions is set to 1.
5.4 Fourth Step: Specification of Test Sections
As mentioned in sec. 2.1 WhisPER allows to sequentially perform several sub-tests within
one test run. For this reason there may be several test sections assigned to a single test run.
Moreover for each test section one must select an appropriate test procedure. Both tasks
can be carried out by either using the button edit test sections which is placed on the main
window’s desktop or selecting the item edit\test sections from the menu bar. In both cases a
DW appears which is labelled as edit test sections and whose control elements are described
in the following.
DW “edit test sections”
available test procedures (panel)
<no label>(list box) Shows all test procedures available under WhisPER.
add to list below (button) Adds a selected test procedure to the bottom of the list of
test sections below.
test sections (panel)
<no label>(multi-columned list box) Shows all test sections in an order defined by
the user. The entries of the three columns represent the count number, an internal
label which should support identification by the user (i.e. the experimenter) and
a synonym which will be used for subject guidance during a test run.
<no label>(output field 1st col) Shows the count number of the selected test section.
<no label>(input/output fields 2nd & 3rd col) Space for displaying and editing of
both the labels for the experimenter and the subject belonging to the selected test
section.
apply (button) Applies changes in the contents of the input/output fields above to the
selected test section.
WhisPER v1.9.0 – User Documentation
22 5. Setting up a test series
Figure 5.3: DW edit test sections.
delete (button) Deletes a selected test section.
move up/move down (button) Shifts the selected test section to a proximate list posi-
tion above/below.
specify test properties (button) Opens a DW for determining the properties of the test
procedure that is chosen for the selected test section (cp. sec. 5.5).
5.5 Fifth Step: Setting up Procedure Properties
The process of determining a specific test procedure’s properties is triggered by applying
the button specify test properties from the DW edit test sections which has been described
in sec. 5.4. Then a DW labeled as <abbreviatory denotation of the type of test procedure>
– main settings (<internal label of the current test section>)will open which has different
appearances in dependency on the type of test procedure that has been addressed to the
currently selected test section. This window and the further approach to setting up the
properties will be described in chapter 7, to be more precise in the respective subsection
belonging to the regarding test procedure.
WhisPER v1.9.0 – User Documentation
23
6 Performing a Test Run
6.1 Initializing and Starting a Test Run
Right before a test run’s performance only a few parameters have to be set by the experi-
menter. This is done on the panel next test run which is placed on the main window’s desktop
(see. sec. 4.2). The basic run parameter settings refer to the specification of the subject’s ID
and the input of the session key which is associated with certain experimental conditions (cp.
sec. 5.3). Both actions have to be performed in the sub-panel run parameters. Furthermore
one can choose only to perform a subset of test sections by making a selection using the sub-
panel execution of test sections. The test run is started by actuating the button Start run.
After this the GUI system for the subject’s guidance and the collection of the data appears
which will be described in the following section.
6.2 The Test Subject’s GUI System
Figure 6.1: GUI system for subject guidance – basic screen showing information on the test run’s
status of progression.
The basis of the subject’s GUI system is constituted by a window which is maximized to cover
the whole screen and which stays in the background during the whole test run’s performance.
This window is labeled as Versuchsdurchfuehrung (German for ”Test Execution”).
WhisPER v1.9.0 – User Documentation
24 6. Performing a Test Run
In WhisPER most test information on the subject side (besides self-definable
instruction texts) will be presented in German language. If this needs to be
changed, one has to modify the source code and is therefore directed to the tech-
nical documentation (see sec. 1.3.4).
In the following this window will be called the subject’s basic screen. It will alternatively dis-
play the test run’s current state of progress, or instructions for subjects. The test’s progress
is signaled at the beginning of a run and after each test section by displaying a list of the
individual test names as defined by the experimenter beforehand (see sec. 5.4). Thereby,
names representing completed test sections are shown in grey letters, the upcoming test is
displayed in green an all remaining ones in black (see fig. 6.1). After clicking the Weiter
(’Continue’) button, the initial test instructions will be displayed (see fig. 6.2). Instructions
may be freely defined by the experimenter beforehand when specifying a procedure’s prop-
erties (see chap. 7). After reading the instruction, the subject may click the OK button and
will then be presented with the current test procedures GUI. GUIs differ according to the
type of test conducted; more details on these windows can be found in chapter 7.
Figure 6.2: GUI system for subject guidance – basic screen showing the initial instructions. IMPOR-
TANT NOTE: The top right corner of the GUI shows a – normally hidden – input field
appearing only, if one actuates the window’s ’close’ icon. When entering the word
“break” one may stop (break) the current test and return to the program’s main win-
dow.
The subject’s basic screen will stay maximized in the background during a test run. To avoid
false aborts caused by subjects the ’close’ icons of all windows belonging to the subject’s
GUI system have been disabled. If it is required to break a test run in a controlled way,
the experimenter may click on the standard close icon at the right corner on the subject’s
basic window. This will open a text input field (see fig. 6.2) for entering the word “break”.
Clicking ok will then close the subject’s GUI system and return the user to the program’s
WhisPER v1.9.0 – User Documentation
6.2 The Test Subject’s GUI System 25
main window.In such a case the inclusion of the empirical data collected up to
this point into the test series’s data set might be unwanted. Therefore, one may
delete the data of the last test run by using the menu item edit\empirical data
(see sec. 4.1).
WhisPER v1.9.0 – User Documentation
26 7. Test Procedures – Configuration and Execution
7 Test Procedures – Configuration and Execution
7.1 Adaptive Psychophysical Threshold Procedures (AP)
7.1.1 The Test Run
Figure 7.1: The subject’s GUI when conducting an adaptive psychophysical threshold procedure.
The design of the GUI for collecting the subject’s responses depends on the selected response
paradigm. At the program’s current state one may apply nAFC-paradigms with n= 2,3or 4.
Figure 7.1 shows the interface for a presentation of three stimulus intervals. The interface
generally comprises nsquared icons arranged on a horizontal line representing the stimulus
intervals. The first stimulus’ playback starts automatically at the beginning of a trial. To
allow for a short delay the experimenter can determine a suitable time period (see sec. 7.1.2).
A further period of silence can be defined for in between the presentations of the stimulus
intervals. During the playback the respective icon which carries the current interval’s number
is highlighted to supply the subject visual information which should help assigning the current
perception to the correct interval. After all stimuli of one trial have been played, the subject
selects the intervals which it supposes to represent the deviant (target) stimulus by clicking
on the respective icon. At the same time the procedure will proceed with presenting the next
trial.
The current target stimulus’ characteristic is controlled by an adaptation mechanism which
one may chose from a larger collection being shortly described already in section 2.2. A
WhisPER v1.9.0 – User Documentation
7.1 Adaptive Psychophysical Threshold Procedures (AP) 27
singular test section may contain either a single adaptive track, or an arbitrary number of
interleaved tracks which use the same response paradigm.
7.1.2 Setting up Properties
The configuration of an adaptive procedure can be divided into several hierarchical “layers”
implying a stepwise setup action. This is due to some of the properties being set on lower
“layers” are affected by those defined on higher ones, whereas the reversed case does not
appear.
The top “layer’s” settings refer to the response paradigm, the time period between trials,
terms of instructions and the laying of the tracks one wants to run. These settings are
performed on the DW AP – main settings (...) (see fig. 7.2) which can be accessed as
explained in subsection 5.5 and will be described below. If several tracks have been created,
one may either randomly interleave all these tracks or define subsets of tracks which are then
assigned to different sessions. If a subset comprises more than one track, these tracks will be
randomly interleaved.
By selecting a single track from the respective list-box and actuating the edit button, one gets
to the DW AP – track settings (...) (see fig. 7.3) which represents the next lower “layer” which
is for determination of the track’s settings. These include specifying a label, setting the range
of stimulus intensities, assigning the stimuli from the stimulus pool to the single intensity
values and choosing the desired adaption and estimation mechanism out of five categories
– Staircase, PEST, ZEST, QUEST and (manually configured) ML/Bayesian procedure. By
clicking on the settings button belonging to the selected mechanism, one arrives at the third
“layer”, which refers to the mechanism’s parameters. Setting up those will be described in
the regarding paragraphs below. But before, there will be given a description of the DWs
referring to the first two “layers”.
Figure 7.2: DW AP – main settings (...)
WhisPER v1.9.0 – User Documentation
28 7. Test Procedures – Configuration and Execution
DW “AP – main settings (...)”
response paradigm (panel)
nAFC (radio button) If selected: selection of the nAFC paradigm.
yes/no (radio button)<this option is not available at the program’s current state>
settings (button) Opens a window for specification of the response paradigm’s settings.
For the nAFC paradigm these are the number nof stimulus intervals, which can
either be 2,3or 4 (default: 3) and the time period of silence between the intervals
in seconds (default: 1 sec). Moreover it can be specified, wether or not the subject
is allowed to listen to the stimuli as often as she/he wants.
break between trials (panel)
<no label>(input field) Space for entering the time period in seconds that lies between
the subject’s response and the initiation of the next trial’s first stimulus’ playback
(default: 2 sec). The input will be treated as a double value.
initial subject instruction (panel)
<no label>(multi-lined input field) Space for entering terms of instructions that will
be displayed on the subject’s basic screen (see sec. 6) before the current test section
is started.
short instruction (on screen each trial) (panel)
<no label>(input field) Space for entering a short term of instruction that will be
displayed on top of the subject’s user interface in each trial.
set/edit track settings (panel)
<no label>(multi-columned list box) This list shows all the tracks that have been
defined within the selected test section and across all sessions. The first column
views the track number and the second its label name. Thereby a track’s number
is only important for its identification. Just for reasons of clearness there exists
the possibility to shift a selected track to another list position by using the move
up/down buttons (see below).
edit (button) Opens a DW labeled as AP – track settings (...) and described below,
for the specification of properties like the track’s label name, the definition of the
stimulus intensity’s range and the choice of an appropriate adaption and estimation
mechanism.
move up/down (button) Shifts the selected item (i.e. track) to a proximate list position
above/below.
delete track (button) Deletes the selected item (i.e. track).
add new track (button) Adds a new item (i.e. track) to the list’s end.
interleaving and session usage (sub-panel)
interleave all tracks in each session (radio button) If selected, all defined tracks
will be performed across all (globally) defined (cp. sec. 5.3) sessions in an
randomly interleaved manner. Note: If the list consists of only one item
(which is supposed to be a rather common scenario), this single track will be
performed in the “normal” (i.e. non-interleaved) manner (across all defined
sessions).
multi session use (radio button) If selected, in each run a subset of tracks which
has been defined by the experimenter and assigned to a certain session key
number, will be performed in a randomly interleaved manner (exclusively
WhisPER v1.9.0 – User Documentation
7.1 Adaptive Psychophysical Threshold Procedures (AP) 29
within the respective session). Note: If a subset consists only of one track, this
single track will be performed in the “normal” (i.e. non-interleaved) manner
(within the belonging session).
organize sessions (button) If the respective option has been chosen, this button
opens a DW that allows for defining subsets of tracks with each of them
assigned to a certain session.
Figure 7.3: DW AP – track settings (...)
DW “AP – track settings (...)”
track name for Track <no. of selected track>(panel)
<no label>(input field) Space for determining a label name for the selected track.
range (panel)
<no label>(input field) Space for entering the maximum stimulus intensity9which
has to be an integer value greater than 1. The stimulus intensity’s range then
is defined from 1 to this value in steps of 1. Note: Usually stimulus intensities
are scaled in a specific unit and one wants to realize a different resolution than
integer steps. In this (normal) case one has to apply a transformation to the
physical domain outside the software with the discrete values kept in a look up
table. Basically stimulus intensities should be “chosen along some scale that yields
approximately equal sensory intervals” [Cornsweet, 1962]. At least for the appli-
cation of parametric methods (e.g. in ML/Bayesian procedures) a certain shape
of the psychometric function is assumed [Treutwein, 1995]. Further note: This pa-
rameter affects some properties of lower configuration layers, i.e. several of those
that depend on the range, e.g. the signals assigned to the stimuli and properties
of the selected adaption and estimation mechanism such as initial values, position
9For simplification this parameter by itself sometimes is called Range. Note that this denotation will appear
in some of the following statements and formulas.
WhisPER v1.9.0 – User Documentation
30 7. Test Procedures – Configuration and Execution
and shape parameters, etc.. So if this parameter’s value will be changed after the
mechanism’s properties have already been set, some of the latter could be changed
or overwritten by default values. It is therefore strongly recommended that in this
case one repeatedly checks these properties to be on the safe side.
stimulus assignment (button) Opens a DW for assigning the stimulus signals from the
stimulus pool to the discrete integer values of stimulus intensities. This includes
a reference stimulus for which also an intensity value can be specified, which has
rather to be seen as a label and won‘t be used by the program at the current
state for any calculation, but might be helpful for documentation and evaluation
purposes.
Adaption and estimation mechanism (panel)
Staircase (radio button) Selection of the Staircase procedure for placing stimuli and
the threshold’s estimation.
PEST (radio button) Selection of the PEST procedure for placing stimuli and the
threshold’s estimation.
max. likelihood/Bayes (radio button) Selection of ML/Bayesian procedures, based on
the Best-PEST but including extensions taken from several similar procedures;
selecting this enables the selection of one of the following three choices:
ZEST (radio button) Selection of the ZEST procedure for placing stimuli and the
threshold’s estimation; this preselects the appropriate options of the ML/Bayes
based procedures for ZEST and disables the others.
QUEST (radio button) Selection of the QUEST procedure for placing stimuli and
the threshold’s estimation. This preselects the appropriate options of the
ML/Bayes based procedures for QUEST and disables the others.
manually (radio button) Selection of ML/Bayes procedures for placing stimuli
and the threshold’s estimation; options from all extensions implemented in
WhisPER are available to the user (yet no obvious guidance how to sensibly
combine them).
settings (button (3x)) Opens a DW for specifications on the selected mechanism, la-
belled as AP – edit <selected type of adaptive procedure>settings (...) (see below).
7.1.2.1 Staircase
DW “AP – edit staircase settings (...)”
stimulus domain (panel)
halve step size (sub-panel)
no (radio button) If selected, there will be no decrease of step size during a run.
after the following reversals (radio button) If selected, the step size will be halved
after a set of reversals which has been determined by the experimenter.
<no label>(input field) Space for entering a set of numbers defining the track’s
reversals at which the step size should be halved. The single values should be
integers greater than 1 and at most equal to the maximum number of reversals
to be performed (see panel termination criterion). Each value must appear
only once. The items have to be separated by space. Let’s say one wants to
WhisPER v1.9.0 – User Documentation
7.1 Adaptive Psychophysical Threshold Procedures (AP) 31
Figure 7.4: DW AP – edit staircase settings (...)
achieve bisections after the 2., 4. and 6. reversal appearing during the run.
Then one has to enter “2 4 6”. Note: Be careful not to insert a space after
the last item!
calculate initial step size: (button) Calculates the size of the first step. This value
directly depends on the number of bisections defined before. For example if one
has chosen the reversals 2, 4, and 6 for halving the step size, the initial step size
will be 8, as the minimum step size is always 1. Note: This function has no direct
effect on the algorithm underlying the procedure. Its purpose only is to facilitate
the configuration process. It should prevent the experimenter from defining too
many bisections which might result in an absurd initial step size that is comparable
to or even exceeds the range.
<no label>(output field) Space for viewing the initial step size’s value which is cal-
culated after actuating the respective button (see above).
initial level (sub-panel)
<no label>(drop-down list) Here the initial value of the stimulus intensity can
be selected. By default this is set to the range’s maximum value.
adaption rule (panel)
1-down/1-up (radio button) If selected, a simple 1-down/1-up adaption rule will be
applied.
2-down/1-up (radio button) If selected, a transformed 2-down/1-up adaption rule will
be applied.
3-down/1-up (radio button) If selected, a transformed 3-down/1-up adaption rule will
be applied.
termination criterion (panel)
number of reversals (input field) Space for entering the maximum number of reversals
which will be performed and after that the procedure will be terminated.
jump out after ... trials (check box & input field) If the box is checked, one may specify
WhisPER v1.9.0 – User Documentation
32 7. Test Procedures – Configuration and Execution
a maximum number of trials that will be performed and after that the procedure
will be terminated if the maximum number of reversals has not been reached
yet. Note: This additional termination criterion was installed in order to avoid
unreasonably long trial sequences. If this criterion catches and the first does not,
the threshold’s estimate will not be calculated as the estimation method (see panel
below) then cannot be applied. In this case 0 is returned as the output. (Under
certain conditions the calculation of a threshold estimate might be caught up
manually outside the program by applying a slightly different scheme, e.g. using
previous reversals for averaging stimulus intensities.)
estimation method (panel)
averaging the last ... reversals (check box & input field) Space for entering the number
of last consecutive reversals whose associated stimulus intensities will be averaged
for calculating the threshold’s estimate. Note: This number must not exceed
the track’s maximum number of reversals specified before (see panel
termination criterion). Otherwise the algorithm will not work properly.
From a methodical point of view, i.e. in order to maximize accuracy and minimize
bias, averaging should only cover an even number of reversals including only those
that appear after the minimum step size has already been reached.
7.1.2.2 PEST
Figure 7.5: DW AP – edit PEST settings (...)
DW “AP – edit PEST settings (...)”
stimulus domain (panel)
initial level (sub-panel)
<no label>(drop-down list) Here the initial value of the stimulus intensity can
be selected. By default this is set to the range’s maximum value.
initial step size (sub-panel)
WhisPER v1.9.0 – User Documentation
7.1 Adaptive Psychophysical Threshold Procedures (AP) 33
<no label>(drop-down list) Here the size of the first step can be selected. By
default this is the second largest value that is possible in the case, i.e. the
second largest power of two that fits in the range.
maximum step size (sub-panel)
<no label>(drop-down list) Here the maximum step size can be selected. This
value will never be exceeded throughout the run no matter what value the
rules call for. By default it is set to the second largest possible value, i.e. the
second largest power of two that fits in the range. Note: As this parameter
already affects the first step, its value should be at least as high as the initial
step size.
PEST version (panel)
original PEST (Taylor & Creelman 1967) (radio button) If selected, the original PEST
procedure described by Taylor & Creelman [Taylor and Creelman, 1967] will be
performed.
deviation limit W (input field) Space for entering the value of the parameter called
deviation limit. The authors of the PEST procedure suggest a value of W= 1 if
one wants to yield a target probability of Pt= 0.75 and higher values like W= 1.5
or W= 2 for target probabilities closer to 0.5.
More Virulent PEST (radio button) If selected, a modified PEST procedure described
by Findlay (1978) [Findlay, 1978] will be performed.
parameter M (input field) Space for entering the value of the parameter M. Sadly, the
author gives no unambiguous recommendations on this parameter’s choice. By
default it is set to 8, as this is a moderate value used in the author’s simulations.
target probability (panel)
Pt(input field) Space for entering the value of the target probability Pt=ψ(ϑ), i.e. the
probability for a positive response at convergence. This value has to lie between
0 and 1 and should be chosen in dependency on the reponse paradigm. A reason-
able choice is sometimes made by using Abott’s formula (based on the classical
threshold concept [cp. Gescheider, 1997, chap. 4]) to calculate the ordinate’s value
ψ(x) of the psychometric function:
ψ(x) = γ+ [1 −γ−λ]·ψ∗(x) (7.1)
Thereby ψ∗(x) is the probability of sensory discrimination, λis the lapsing rate
and γthe guessing rate or false alarm rate. At convergence ψ∗(ϑ) is usually defined
as 50%. In case of the nAFC paradigm there is γ=1
n. Hence by default the target
probability is set to the respective value assuming λ= 0, i.e.
Pt=1
n+1−1
n·0.5 (7.2)
secondary termination criterion (panel)
jump out after ... trials (check box & input field) If the box is checked, here one may
enter the maximum number of trials that will be performed and after that the
procedure will be terminated if this has not been done by the PEST rules yet.
(The procedure regularly terminates when the minimum step size of 1 is under-
run.) Note: This additional termination criterion was installed in order to avoid
unreasonably long trial sequences. If this criterion catches and the first does not,
WhisPER v1.9.0 – User Documentation
34 7. Test Procedures – Configuration and Execution
the threshold’s final estimate cannot be properly calculated and is therefore set
to 0. In this case one might want to look for the stimulus intensity that would
have been taken for the next placement. This value can either be found in the log
file (see sec. 10.1.5) or in the track’s plot (see sec. 10.1.6). Note that this value
comprises a different accuracy, as the minimum step size has not been under-run
and possibly not even been reached yet.
7.1.2.3 ML/Bayesian procedure
Figure 7.6: DW AP – edit ZEST settings (...)
DW “AP – edit ZEST settings (...)”
DW “AP – edit QUEST settings (...)”
DW “AP – edit ML/Bayes settings (...)”10
psychometric function (panel) The shape of the underlying psychometric (model) function
ˆ
ψ(x) is described by one of the following two equations, which can be derived from
equation (7.1) on page 33 by assuming a certain function for ψ∗(x).
10This DW’s title depends on your previous choice of ML/Bayes based procedures, cf. sec. 7.1.2. The same
goes for the availability of certain options of this DW.
WhisPER v1.9.0 – User Documentation
7.1 Adaptive Psychophysical Threshold Procedures (AP) 35
logistic (radio button) The function used is the logistic function l(x;α, β) = 1
1+e
−(x−α)
β
.
This yields:
ˆ
ψ(x) = ˆ
ψ(x;θ, β) = γ+ (1 −γ−λ)·1
1 + e
−(x−θ)
β
(7.3)
Thereby λis the lapsing rate and γthe guessing rate or false alarm rate which
is assumed to be γ= 1/n for the nAFC paradigm. θis the unknown position
parameter (which in case of a logistic model function is equal to the threshold)
and βthe spread parameter (see further below).
Weibull (radio button) The function used is the Weibull function11, yielding:
ˆ
ψ(x) = ˆ
ψ(x;θ, β) = γ+ (1 −γ−λ)·exp[−10β(x−θ+)] (7.4)
Again λis the lapsing rate, γthe guessing rate (assumed to be γ= 1/n for
the nAFC paradigm), θthe threshold and βthe spread parameter. gives the
possibility of shifting what point of the psychometric function is used as threshold,
thereby influencing the threshold criterium (see below). Fields for input of λ,γ
and βare ”shared” with the fields for the corresponding input for the logistic
function (see below).
Please note that one of the mandatory requirements for psychometric functions is
that for different conditions it doesn’t change shape, but only its position along
the x-axis, see e. g. Watson and Pelli [1983]. The Weibull function alas specifically
meets this criterion only under the condition that the x-axis is scaled logarithmi-
cally, see e. g. [Treutwein, 1995]. Thus the Weibull function is best suited for use as
psychometric function in those cases, where the relationship of the tested stimuli’s
intensities is logarithmic, too.
spread parameter (beta) (input field) Space for entering the value of the spread pa-
rameter βwhich should be chosen based on prior knowledge. Sometimes it might
be helpful to consider its relationship to the psychometric function’s slope S:
S=(1 −γ−λ)
4β.(7.5)
In other cases one might note the relationship to the spread σlof the logistic
function l, which is approximately given by
σl≈1.81 ·β(7.6)
[Hastings and Peacock, 1975]. If one has prior knowledge about a psychometric
function being described by a cumulative normal ogive then one might use the
relationship
σN≈1.7·β(7.7)
with σNbeing the ogive’s standard deviation, as for this the logistic function can
be seen as a fairly good approximation to the cumulative normal [Treutwein, 1995,
textbox 2]. If the spread parameter’s true value is unknown the authors of Best
PEST recommend rather to use a larger than a smaller value [Lieberman and
Pentland, 1982]. By default it is set to β=Range/10 (cp. panel termination,
option confidence interval). (This parameter is used by both the logistic as well
as the Weibull function.)
11Used here in that of its many forms which is used by ZEST.
WhisPER v1.9.0 – User Documentation
36 7. Test Procedures – Configuration and Execution
guessing rate (gamma) (input field) Space for entering the guessing rate (nAFC para-
digm) or false alarm rate (yes/no paradigm). At the program’s current state one
only may use the nAFC paradigm and so this value is fixed to 1/n and therefore
the respective field is not editable.(This parameter is used by both the logistic as
well as the Weibull function.)
lapsing rate (lambda) (input field) Space for entering the lapsing rate λ. In the original
procedure this parameter was not included in the psychometric model function,
i.e. it was implicitly set to zero. As some scientists recommend to choose a value
of about 1% [eg. Hall, 1981; Harvey, 1986; Klein, 2001], this one is selected for
default setting. Although it should be stated that the true lapsing rate depends
on the individual subject and therefore is unknown [cp. Treutwein, 1995]. Note:
Due to the definition of the underlying psychometric model function (cp. eq. (7.3)
on p. 35 and eq. (7.4) on p. 35 ) this value naturally has to be selected from the
interval [0,1−γ]. Otherwise the algorithm cannot work properly.(This parameter
is used by both the logistic as well as the Weibull function.)
threshold modifier (epsilon) (input field) Space for entering the threshold modifier ,
only activated when the Weibull psychometric function is selected. This parameter
shifts the psychometric function along the x-axis and therefore influences which
point of the psychometric function is used for testing. This does not have an effect
on the accuracy of obtainable threshold measurements’ results, but it does have
on the procedure’s efficiency, see King-Smith et al. [1994]. King-Smith et al., 1994
suggest that for tests running between 8 and 16 trials long be set to 0, which is
also the default value used in WhisPER. For tests lasting longer than 16 trials
they assess the greater efficiency to be gained setting so that the testing occurs
at the ideal sweat factor (see below), which is what Watson and Pelli, 1983 used.
calculate (button) Automatically calculates so that testing occurs at the ideal
sweat factor of the psychometric function of the present values. The sweat
factor is a measure of efficiency describing the relationship between the amount
of ”work” to obtain a certain result and the accuracy of said result. For the
psychometric function it is calculated as the square of the ratio of the binomial
standard deviation and the slope of the psychometric function at the point in
question (Watson and Pelli [1983], Taylor [1971]):
K(X) = Ψ(X)[1 −Ψ(X)]
[dΨ(X)/dX]2with X=x−θ(7.8)
The ideal sweat factor is represented by the minimum of K(X). The calcu-
lation is done iteratively, the result is automatically written to the input
field. This button is only activated when the Weibull function is chosen as
psychometric function.
p(Threshold) (output field) Displays the answering probability of the psychometric
function at the testing point. This can be useful for comparing probabilities derived
from optimisation of (see above) with probabilities of more traditional threshold
criteria.
view psychometric function (button) Opens a window containing a plot of the psy-
chometric function’s shape which has been calculated for the previously specified
parameter settings. Note: This feature does not directly influence the algorithm
underlying the procedure. Its purpose only is to facilitate the configuration as it
might supply the experimenter a tool for a visual verification of the choices made
on the parameters’ values.
WhisPER v1.9.0 – User Documentation
7.1 Adaptive Psychophysical Threshold Procedures (AP) 37
initialisation (panel) For a discussion of the different possibilities that exist to do the initial-
isation see for example [Pokorny, 1998].
implicit trials (radio button) If selected, the procedure will be initialized by defining a
data set of stimulus intensities and corresponding (presumed) responses which is
sometimes referred to as implicit trials. This is done in the way it is described by
the authors of Best PEST, Lieberman and Pentland, 1982.
gaussian a priori p. d. f. (radio button) If selected, the procedure will be initialised with
a (normalised) gaussian a priori probability density function N(x;µ, σ), with µ
being its mean and σthe standard deviation.
mean (input field) Space for entering the mean µof the gaussian a priori prob-
ability density function. By default µis set to the middle of the range, i.e.
(1+Range)
2.
std. deviation (input field) Space for entering the standard deviation σof the
gaussian a priori probability density function. The experimenter should take
care that this function’s graph is relatively flat compared to the final p. d. f. so
that the latter never will be dominated by the former [Treutwein, 1995; Martz
and Waller, 1982, p. 184]. By default σis set to 0.7·Range which is a value
that has no theoretical justification but is intuitively chosen to yield such a
flat curve.
modified hyperbolic p. d. f. (radio button) If selected a modified hyperbolic secant is
used as a priori probability density function. Suggested by the authors of ZEST,
King-Smith et al., 1994, they used this function to fit a p. d. f. to histograms of
thresholds measured in earlier experiments. It takes the form of
q0(θ) = A
B·e−C(θ−t)+C·eB(θ−t)(7.9)
with θbeing (logarithmic) threshold, Abeing the overall height of the curve, Bits
lower and Cits upper slope and tbeing the most probable threshold, the function’s
maximum (i. e. tshifts the function along the x-axis). The advantage of using a
function like this is that in contrast to all other initialization options provided here
it is capable of describing asymmetric threshold distributions, which – seeing that
a modified hyperbolic p. d. f. has been reached fitting a curve to real data from
earlier experiments – obviously do exist.
A (height) (input field) Space for inputting the overall height of the curve. As the
height of the a priori p. d. f. is usually not relevant to the correct working of
e. g. the ZEST or the QUEST procedure (c. f. King-Smith et al., 1994), this is
simply set to 1 per default. As the function’s overall height is heavily affected
by both slope parameters, please see below for a remedy.
norm. (button) Pushing this button will calculate the height parameter for
the curve to have a maximum of 1. As the height is usually really not
important to the working of the procedures, this is simply for convenience,
e. g. when viewing the generated curve via the view a-priori p. d. f. button
this simply makes sure the whole graph is visible.
B (low slope) (input field) Space for entering the slope of the curve in the range of
lower threshold values. For lack of different approaches this is simply initialized
to 1.22, the value King-Smith et al., 1994 used. Please note that although
they expressed their confidence in the possibility that their modeled a priori
p. d. f. might be of universal nature, they warned at the same time about it
WhisPER v1.9.0 – User Documentation
38 7. Test Procedures – Configuration and Execution
possibly not being usable in the same way for every possible psychometric
measurement. ”Wrong” values entered for the a priori p. d. f. should not harm
the accuracy of the test, however, but only its efficiency (cf. King-Smith et al.,
1994).
t (prob. thresh.) (input field) Space for entering the value for the value of the
function that is to be the most probable threshold. As mentioned above this
parameter slides the function along the x-axis. This, too, is simply initialized
to the value used by King-Smith et al., 1994, 2. If not fitted to actual data of
your own it should usually better be set to Range
2, for example.
C (high slope) (input field) Space for entering the slope of the curve in the range of
lower threshold values. For lack of different approaches this is simply initialized
to 5.07, the value King-Smith et al., 1994 used. Please note the warning about
this further above.
uniform a-priori p. d. f. (radio button) If selected, no prior information will be used.
The first stimulus’ intensity will be chosen from the middle of the range, i.e.
round 1+Range
2.
view a-priori p. d. f. (button) Opens a window containing a plot of the a priori p. d. f.
which has been calculated for the selected variant of initialization. Note: This
feature does not directly influence the algorithm underlying the procedure. Its
purpose only is to facilitate the configuration as it might supply the experimenter
a tool for a visual verification of the selected initialization approach and the choice
of the respective parameters’ values.
exclude prior information from calculation of the final estimate (check box) If checked,
the prior information will be excluded from calculation of the final estimate as this
has been recommended by some scientists [e.g. Watson and Pelli, 1983]. This is
internally performed in a way that is equivalent to dividing the final p. d. f. by the
a priori p. d. f. before selecting/calculating the average (i.e. the mode or the mean,
cp. below). Hence this option will affect the first three approaches to initialization
described above.
measure of central tendency (panel)
mode (ML) (radio button) If selected, the maximum of the likelihood function will be
taken for both placing stimuli and the threshold’s final estimation. (A procedure
involving this approach basically can be called a maximum likelihood (ML) proce-
dure as it is implied amongst others by the Best PEST [Lieberman and Pentland,
1982; Pentland, 1980].)
mean (radio button) If selected, the mean of the a posteriori probability density func-
tion (i.e. the likelihood, normalized to the area enclosed by the graph) will be
taken for both placing stimuli and the threshold’s final estimation.
termination (panel)
number of trials (radio button & input field) If selected, the procedure will be termi-
nated after a fixed number of trials has been carried out. The latter must be
entered in the corresponding input field. As ZEST has been proven to already
converge among the first 20 trials (cf. Otto [2008]), selecting ZEST will set this
value to 20. By default this parameter is set to 40, though this has been a rather
arbitrary choice.
confidence interval (radio button) This dynamic stopping criterion leads to a termi-
WhisPER v1.9.0 – User Documentation
7.1 Adaptive Psychophysical Threshold Procedures (AP) 39
nation if the current confidence interval12 (at a predefined confidence level) of the
threshold’s estimation falls below a preset limit (which is specified by the input
field maximum width). Note: A true confidence interval (as an absolute measure)
can only be calculated if the true spread of the psychometric function is known.
As this usually is not the case the probability intervals calculated in the given
context can only be regarded as comparative measures. Further notes: The num-
ber of trials which are carried out till this criterion catches strongly depends on
the setting of the spread parameter (see above). Higher values lead to a greater
number of trials and vice versa. As a rule of thumb it is suggested in Treutwein
[1995] to adjust one of the three parameters, the spread, the maximum width of
the confidence interval and the confidence level, until the procedure stops after the
following number of trials on average in dependency on the response paradigm:
yes/no, 20; 4-AFC, 30; 3-AFC, 37.5 and 2-AFC, 50. By default the maximum
confidence interval is set to round(0.15 ·Range) and the level to 0.95.13
maximum width (input field) Space for entering an upper limit for the total width
of the confidence interval belonging to the final estimate. Due to the range’s
resolution and the internal calculation of the current confidence interval, only
positive integer numbers constitute sensible choices (i.e. any value from the
half-open interval extended by an integer number and the next one above
which is not included, results in the same behavior of the procedure as the
integer by itself).
confidence level (input field) Space for entering the value of the confidence level
(also called confidence coefficient). The value must be from the interval [0,1].
secondary termination criterion (sub-panel)
jump out after ... trials (check box & input field) This option is only available
if a dynamic stopping criterion has been selected. If in this case the box
has been checked, one can enter a maximum number of trials that will
be performed and after which the procedure will be forcefully terminated
if the dynamic criterion has not caught yet. Note: This additional ter-
mination criterion was installed in order to avoid unreasonably long trial
sequences. If this criterion catches and the first does not, the threshold’s
final estimate cannot be properly calculated and is therefore set to 0. In
this case one might want to look for the stimulus intensity that would have
been taken for the next placement. This value can either be found in the
log file (see sec. 10.1.5) or in the track’s plot (see sec. 10.1.6).
12The calculation of the current confidence interval is performed as described in [Treutwein, 1995, eq. 23] and
as carried out in the YAAP implementation (published in Treutwein [1997]; seen version 3.0, 1993/12/1,
by courtesy of B. Treutwein).
13These values have been encountered in simulations which yielded the following numbers of trials on average
(stimulus range stated in brackets, resolved in steps of 1): 26,1 (30); 34,4 (50); 47,0 (100). The simulations
have been performed for the 3AFC paradigm and a spread parameter of β=Range/10. Note that criteria
of validity like accuracy and bias thereby have not been under consideration.
WhisPER v1.9.0 – User Documentation
40 7. Test Procedures – Configuration and Execution
7.2 Repertory Grid Technique (RGT)
7.2.1 The Test Run
The performance of this test procedure is split in two parts: At first, bipolar descriptors of
qualitative differences (typically called constructs in RGT) are elicited by triadic comparison
of some representative stimuli. Hereby, triadic comparison means that – at each trial – three
different stimuli (which have potentially been drawn from a larger pool) are being compared.
Secondly, the stimuli (typically called elements) a rated on bipolar scales whose endpoint
descriptors are formed by the respective elicited constructs.
In WhisPER there is a break between the two test parts that can be used by the experimenter
for editing the elicited constructs, which might be necessary for reviewing the elicited phrases.
Moreover, final decisions can be made with respect to the polarity of the scales.
Figure 7.7: The subject’s GUI for the performance of the RGT’s first part, i.e. the triadic comparison.
Part I: Elicitation of Constructs Figure 7.7 shows the elicitation GUI. To play one of the
triad’s stimuli the subject has to apply one of the three buttons denoted by X, Y and Z. The
subjects’ first task is to decide which two out of the three stimuli appear to be more similar
than the third one and then enter the labels of the corresponding two buttons (i.e. X, Y or
Z). The inputs must be confirmed by applying the OK button. This task is only required to
WhisPER v1.9.0 – User Documentation
7.2 Repertory Grid Technique (RGT) 41
correctly formulate the next two questions which will be presented only, if each stimulus has
been played for at least one time. First it is asked in how far the two stimuli which have been
previously identified as the most similar ones are alike. Second, it is asked for the difference
between these two stimuli and the third one. Please be aware that alternative methods for
posing these two questions – such as asking for a direct negation of the first descriptor – have
been discussed in literature, too.
After answering these questions, the subject may choose to either go on reporting further
similarites and differences or to continue with the next triad’s presentation. Note that the
test continues only, if each stimulus has been played for at least one time and all questions
have been answered (i.e., inputs have been made to the respective fields). The first part of the
procedure ends by displaying an instruction telling the subject to wait for the experimenter.
Break: Editing Constructs When the experimenter operates the button labeled ”nur vom
Versuchsleiter zu bedienen” a DW will appear which is labeled as RGT – edit constructs
(see Fig. 7.8). As with all other experimenter DWs the operating language in this DW is
English. Possible modifications of the elicited phrases range from revising construct phrasing,
determining assignments to either one of the rating scale’s poles, deleting entire constructs,
making decisions on the scale’s polarity, i.e. its orientation when being posed during the
rating process, and changing the order in that scales are posed one below the other. After
editing, the experimenter may press the ”resume” button in order to start the procedure’s
second part, i.e. the rating procedure.
DW “RGT – edit constructs (...)”
subject: ... (text) Displays the ID of the current subject.
session: ... (text) Displays the key number of the current session.
edit constructs for rating (panel)
<no label>(multi-columned list box) This list primarily shows the constructs arranged
in the order of their elicitation events which is represented by the numbers placed
in the first column. After modifications have been made with respect to the order
of the list’s entries (by using the move up/move down buttons, see below) these
numbers then represent the order in which the constructs’ scales will be posed one
below the other during the rating process. The further columns view the poles’
verbal descriptors and the scale’s polarity, i.e. its orientation when being posed
during the rating process. For the latter a “+” indicates that the low pole (i.e.
the pole belonging to a minimum scale value) will be posed on the left side of
the DW and the high pole (i.e. the pole belonging to a maximum scale value)
on the right side. The reversed case is indicated by a “-”. Note that at first
poles may be not properly defined as the verbal descriptors’ assignment either to
the scale’s minimum or maximum extension has been automatically done straight
forward (i.e. the inital pole is always assigned to the low pole and vice versa). A
correction then has to be performed manually by using the button switch labels
which is described below. When all editing work has been completed the list must
not contain more than 30 constructs. This restriction is due to the finite display
resolution which limits the number of scales to be displayed properly within one
window.
WhisPER v1.9.0 – User Documentation
42 7. Test Procedures – Configuration and Execution
Figure 7.8: The DW RGT – edit constructs (...) offers the possibility to the experimenter for editing the
constructs elicited in the first part.
<no label>(output field 1st col) Displays the number assigned to the selected con-
struct.
<no label>(input/output fields 2nd & 3rd col) Spaces for displaying and editing the
verbal descriptors of the scale’s poles belonging to the selected construct.
<no label>(output field 4th col) Indicates the polarity of the scale belonging to the
selected construct.
switch polarity (button) Switches polarity of the scale belonging to the selected con-
struct.
switch labels (button) Switches verbal descriptors of the scale’s poles belonging to the
selected construct.
apply (button) Applies changes in the contents of the input/output fields above to the
selected construct.
move up/move down (button) Shifts the selected construct to a proximate list position
above/below.
delete construct (button) Deletes a selected construct which then is shifted to the list
below.
deleted constructs (panel)
<no label>(multi-columned list box) This list views all constructs that have been
deleted from the list above. The idea behind this is to prevent the experimenter
from losing a construct by faulty operation, as the items contained in this list can
WhisPER v1.9.0 – User Documentation
7.2 Repertory Grid Technique (RGT) 43
be reassigned to the list above by using the button undelete which is described
below.
undelete (button) Reassigns the selected construct to the list above.
resume with rating (button) Resumption of the current test section by initiating the RGT’s
second part, i.e. the rating process. Note: make sure that all editing work has been
completed before applying this button, as there will be no possibility to return to this
DW afterwards.
Part II: Rating Procedure The rating part is introduced by displaying experimenter-defined
instructions. After operating the weiter-button the rating procedure begins.
The respective GUI is shown in figure 7.9. The playback of the respective stimulus starts
automatically at the beginning of each trial. To allow for a short delay the experimenter
can specify a suitable time period (see sec. 7.2.2). Rating is allowed from the beginning of a
trial (even during the playback) either on discrete scales which are realized by radio buttons.
Progression to the next stimulus is possible if the whole set of – at most 30 – constructs has
been rated. This restriction is due to the finite display resolution preventing displaying a
greater number of scales together with their poles’ labels within one window.
Figure 7.9: The subject’s GUI for the performance of the RGT’s second part, i.e. the rating of the
elements on the constructs which have been elicited and edited before.
7.2.2 Setting up Properties
Settings of the procedure’s properties have to be determined on the DW RGT – main settings
(...) which can be accessed as explained in section 5.5. The basis for the whole procedure
is formed by the definition of elements which is done in the respective DW accessible by
the button element pool. Thereby, a set of stimuli from the stimulus pool can be selected
WhisPER v1.9.0 – User Documentation
44 7. Test Procedures – Configuration and Execution
and assigned to element key numbers, gathered in the element pool. Further settings are
structured by the procedure’s two parts and hence adjustable in two distinct panels.
If elements have been defined, the generation of the triads can be performed on the panel
Part I: construct elicitation by applying the button edit triads which then opens a DW that
allows either to manually assign elements to the triads’ key numbers or to let the program
automatically generate a complete variation of (at most 15) elements over triads which then
may be edited manually.
After defining the triads the order of their presentation has to be determined. This can be
done in two different ways: On the one hand one may choose to play all triads in a random
order, whereas on the other hand one may select certain triads which can either be played
in a determined sequence or in random succession. When choosing the second approach one
has to apply the regarding edit button to reach the respective DW labeled as RGT – triad
sequence setup (...) which is described in more detail below. A further property that can be
set is the maximum possible number of constructs the experimenter wants to be drawn during
one triadic comparison (cp. sec 7.2.1). Moreover, two instructions may be entered. The first
one will be displayed in the beginning of the procedure’s first part whereas the second one
should consist only of a short phrase which will be placed on top of the triadic comparison
panel.
Setting the properties of the second part (rating) has to be performed on the panel Part II:
rating procedure. By applying the button edit scales one gets to the DW RGT – edit scale
format which is described below and allows adjusting the scales’ format settings such as the
number of categories, the categories’ numerical representatives, category labels if required.
Further parameters of the rating process refer to the order of the elements’ presentation and
will be adjusted in a manner that is completely analogous to the approach already described
above in the triads’ context. Moreover, the time period between the subject’s operation of the
weiter-button and the presentation of the next element can be fixed. Finally, two different
instructions may be entered. Again, two instructions may be entered. The first one will
be displayed in the beginning of the procedure’s second part whereas the second one should
consist only of a short phrase which will be placed on top of the rating panel.
WhisPER v1.9.0 – User Documentation
7.2 Repertory Grid Technique (RGT) 45
Figure 7.10: DW RGT – main settings (...)
DW “RGT – main settings (...)”
definition of elements (panel)
element pool (button) Opens a DW for definition of the elements by assigning stimuli
from the stimulus pool to the elements’ key numbers which are listed in the element
pool.
Part I: construct elicitation (panel)
edit triads (button) Opens a DW for generation of triads. The latter can either be
performed manually by assigning three elements respectively to each triad’s key
number or by automatically generating a complete variation. In the latter case it
is possible to permute 15 elements at most yielding a list of 455 triads.
sequence setup (sub-panel)
all triads in random order (radio button) If selected, the complete list of triads
generated before will be presented to each subject in a random order.
manual selection (radio button) If selected, only a selection of triads, i.e. a subset
of the triad list, will be presented to each subject with sequences either being
randomly generated or predefined (see button edit below).
edit (button) Opens a DW labeled as RGT – triad sequence setup (...) and de-
scribed below, for selection of subsets of triads and determination of their
orders of presentation.
WhisPER v1.9.0 – User Documentation
46 7. Test Procedures – Configuration and Execution
maximum number of constructs per triad (sub-panel)
<no label>(drop-down list) Here you can choose the maximum number of con-
structs that may result from one triad comparison. Subjects will then have
the opportunity to report on a respective quantity of attributes if they intend
to.
initial subject instruction (sub-panel)
<no label>(multi-lined input field) Space for entering terms of instructions that
will be displayed on the subject’s basic screen (see sec. 6) before the test
section (i.e. the first part of the RGT) is started. The input will be treated
as a string.
short instruction (on screen each trial) (sub-panel)
<no label>(multi-lined input field) Space for entering a short term of instruction
that will be displayed on top of the subject’s user interface in each trial. The
input will be treated as a string.
Part II: rating procedure (panel)
edit scales (button) Opens a DW labeled as RGT – edit scale settings (...) and de-
scribed below, for editing the format settings of the rating scales.
sequence setup (sub-panel)
all elements in random order (radio button) If selected, all defined elements will
be presented to each subject in a random order.
manual selection (radio button) If selected, only a selection of elements, i.e. a
subset of the element pool, will be presented to each subject with sequences
either being randomly generated or predefined (see button edit below).
edit (button) Opens a DW labeled as RGT – element sequence setup (...) and
described below, for selection of subsets of elements and determination of
their orders of presentation during the rating process.
break between trials (sub-panel)
<no label>(input field) Space for entering the time period in seconds that lies
between the subject’s activation of the weiter button after finishing rating on
the current stimulus (i.e. element) and the initiation of presentation of its
successor. The input will be treated as a double value.
initial subject instruction (sub-panel)
<no label>(multi-lined input field) Space for entering terms of instructions that
will be displayed on the subject’s basic screen (see sec. 6) before the second
part of the RGT (i.e. the rating of the elements) is started. The input will be
treated as a string.
short instruction (on screen each trial) (sub-panel)
<no label>(multi-lined input field) Space for entering a short term of instruction
that will be displayed on top of the subject’s user interface in each trial. The
input will be treated as a string.
The following description of a DW refers to the manual selection and determination of se-
quences of both triads and elements.
DW “RGT – triad/element sequence setup (...)”
contents of sessions (panel)
WhisPER v1.9.0 – User Documentation
7.2 Repertory Grid Technique (RGT) 47
Figure 7.11: DW RGT – triad/element sequence setup (...)
<no label>(multi-columned list box) Shows all sessions that have been defined earlier
(see sec. 5.3) and their belonging subsets of selected triads/elements.
separate items by space (input field) Space for entering the key numbers of the tri-
ads/elements one wants to be presented during the regarding session. Items have
to be separated by space.
assign triads/elements (button) Assigns the content of the previously described input
field to the session currently selected. Preexisting content will be overwritten.
view triads/elements (button) Shows the list of triads/elements defined earlier (see
button edit triads/element pool above).
sequence (panel)
as defined (radio button) If selected, the sequence of presentation will be the same as
the order of the items entered to the field previously described.
random order (radio button) If selected, the sequence of presentation will be randomly
generated before each run.
DW “RGT – edit scale format (...)”
range settings (panel)
number of categories (drop-down list) Here you can choose a number of categories
that is inbetween 2 and 11. By default it’s set to 5.
minimum value (input field) Space for entering the minimum value of the scale’s range.
By default it’s set to -2.
category width (input field) Space for entering the category width. By default it’s set
to 1.
apply (button) Calculates the categories’ numerical representatives.
your category values are (output field) Displays the categories’ numerical representa-
tives if calculated before by using the apply button.
category labels (if required) (input field) Space for entering the category labels if re-
quired. The single inputs will be treated as strings that have to be separated
WhisPER v1.9.0 – User Documentation
48 7. Test Procedures – Configuration and Execution
Figure 7.12: DW RGT – edit scale format (...)
by space and are assigned to the categories in ascending order of their numerical
representatives. If you want to skip a category enter a $-sign.
value selection mode (panel)
discrete choice (radio button) If selected, the subject will only be able to choose a
point on the scale that belongs to one of the categories’ marks. Selection is made
by using radio buttons.
continuous slider (radio button)<not available at the program’s present state>
resolution (drop-down list)<not available at the program’s present state>
7.3 Semantic Differential (SD)
7.3.1 The Test Run
The procedure-specific user interface is depicted in figure 7.13. The playback of the respective
stimulus starts automatically at the beginning of each trial. To allow for a short delay the
experimenter can specify a suitable time period (see sec. 7.3.2). Rating is allowed from the
beginning of a trial (even during the playback) either on discrete scales which are realized by
radio buttons. Progression to the next stimulus is possible if the whole set of – at most 30 –
constructs has been rated. This restriction is due to the finite display resolution preventing
displaying a greater number of scales together with their poles’ labels within one window.
7.3.2 Setting up Properties
The procedure’s settings may be edited in the DW SD – main settings (...) which is depicted
in figure 7.14 and can be accessed as explained in subsection 5.5. Its control elements will
WhisPER v1.9.0 – User Documentation
7.3 Semantic Differential (SD) 49
Figure 7.13: The subject’s GUI for performing the SD test.
be described below. In the beginning, objects (i.e. stimuli) to be rated have to be defined
using the DW SD – object pool (...) (see fig. 7.15 on p. 51) accessible by the button object
pool. Thereby, a set of stimuli from the stimulus pool can be selected and assigned to object
key numbers, gathered in the object pool. By applying the button edit scales the DW SD –
edit scales (see fig. 7.16 on p. 52) is opened. The DW allows adjusting the scales’ format
settings such as the number of categories, the categories’ numerical representatives, and
category labels if required. Moreover, verbal descriptors of the scales’ endpoints can be
defined. Further settings refer to the order of the objects’ presentation and can be adjusted
in the panel sequence setup. Here, one can choose the generation of a random sequence
or define a sequence manually by applying the button edit whic opens a DW SD – object
sequence setup (...) (see fig. 7.17 on p. 53). Moreover, the time period between the subject’s
actuation of the weiter-button and the presentation of the next object can be fixed. Finally,
two instruction texts can be defined. The first one will be displayed at the beginning of the
procedure, the second one, which should consist of a short phrase only, which will be shown
on the top of the subject’s user interface in each trial.
DW “SD – main settings (...)”
object pool (button) Opens a DW for definition of the objects by assigning stimuli from the
stimulus pool to the objects’ key numbers which are listed in the object pool.
edit scales (button) Opens a DW labeled as SD – edit scales (...) and described below,
for editing the formal settings of the rating scales and defining the endpoints’ verbal
descriptors.
sequence setup (panel)
all objects in random order (radio button) If selected, all objects will be presented to
each subject in a random order.
WhisPER v1.9.0 – User Documentation
50 7. Test Procedures – Configuration and Execution
Figure 7.14: DW SD – main settings (...)
manual selection (radio button) If selected, only a selection of objects, i.e. a subset
of the object pool, will be presented to each subject with sequences either being
randomly generated or predefined.
edit (button) Opens a DW labeled as SD – object sequence setup (...) and described
below, for selection of subsets of objects and determination of their orders of
presentation.
break between trials (panel)
<no label>(input field) Space for entering the time period in [sec] that lies between
the subject’s actuation of the weiter button after finishing rating on the current
stimulus (i.e. object) and the initiation of its successor’s presentation. The input
will be treated as a double value.
initial subject instruction (panel)
<no label>(multi-lined input field) Space for entering terms of instructions that will
be displayed on the subject’s basic screen (see sec. 6) before the test section is
started.
short instruction (on screen each trial) (panel)
<no label>(multi-lined input field) Space for entering a short term of instruction that
will be displayed on top of the subject’s user interface in each trial.
WhisPER v1.9.0 – User Documentation
7.3 Semantic Differential (SD) 51
Figure 7.15: DW SD – object pool (...) (see text on p. 49)
DW “SD – edit scales (...)”
range settings (panel)
number of categories (drop-down list) Here you can choose a number of categories
that is in between 2 and 11. By default it’s set to 5.
minimum value (input field) Space for entering the minimum value of the scale’s range.
By default it’s is set to -2.
category width (input field) Space for entering the category width. By default it’s set
to 1.
apply (button) Calculates the categories’ numerical representatives.
the categories’ values are (output field) Displays the categories’ numerical represen-
tatives if calculated before by using the apply button.
category labels (if required) (input field) Space for entering the category labels if re-
quired. The single inputs will be treated as strings that have to be separated
by space and are assigned to the categories in ascending order of their numerical
representatives. If you want to skip a category enter a $-sign.
value selection mode (panel)
discrete choice (radio button) If selected, the subject will only be able to choose a
point on the scale that belongs to one of the categories’ marks. Selection is made
by using radio buttons.
continuous slider (radio button)<not available at the program’s present state>
resolution (drop-down list)<not available at the program’s present state>
item definition (panel)
<no label>(multi-columned list box) This list shows the scales’ endpoint descriptors
representing the items under test, i.e. the attributes underlying the affective or
WhisPER v1.9.0 – User Documentation
52 7. Test Procedures – Configuration and Execution
Figure 7.16: DW SD – edit scales (...)
perceptual measurement. During the test run the items and their belonging scales
will be placed on the screen in the same order as they are created in the list. The
further columns view the poles’ verbal descriptors and the scale’s polarity, i.e. its
orientation when being posed during the test run. For the latter a “+” indicates
that the low pole (i.e. the pole belonging to a minimum category value) will be
posed on the left side of the DW and the high pole (i.e. the pole belonging to a
maximum category value) on the right side. The reversed case is indicated by a
“-”. The list may not contain more than 30 items. This restriction is due to the
finite display resolution as a greater number would lead to problems with viewing
all scales and their labels within one common window.
<no label>(output field 1st col) Displays the selected item’s number which also rep-
resents its position within the order of presentation during the test run.
<no label>(input/output fields 2nd & 3rd col) Spaces for displaying and editing the
verbal descriptors of the scale’s poles belonging to the selected item.
<no label>(output field 4th col) Indicates the polarity of the scale belonging to the
selected item.
switch polarity (button) Switches polarity of the scale belonging to the selected item.
switch labels (button) Switches verbal descriptors of the scale’s poles belonging to the
selected item.
apply (button) Applies changes on the contents of the input/output fields described
above to the selected item.
move up/move down (button) Shifts the selected item to a proximate list position
above/below.
delete item (button) Deletes a selected item.
add new item (button) Adds a new item to the list after its last position.
WhisPER v1.9.0 – User Documentation
7.4 ABX Detection Test (ABX) 53
Figure 7.17: DW SD – object sequence setup (...)
DW “SD – object sequence setup (...)”
contents of sessions (panel)
<no label>(multi-columned list box) Shows all sessions that have been defined earlier
(cp. sec. 5.3) and their belonging subsets of selected objects.
enter the objects’ key numbers (...) (input field) Space for entering the key numbers
of the objects one wants to be presented during the regarding session. Items have
to be separated by space.
assign objects (button) Assigns the content of the previously described input field to
the session currently selected. Preexisting content will be overwritten.
view objects (button) Shows the list of objects defined earlier.
sequence (panel)
as defined (radio button) If selected, the sequence of presentation will be the same as
the order of the items entered to the field previously described.
random order (radio button) If selected, the sequence of presentation will be randomly
generated before each run.
7.4 ABX Detection Test (ABX)
The ABX test (see also sec. 2.2) is a 2AFC double blind detection test allowing to assess
the detectability of very subtle differences between two stimuli while controlling type-1 error
level and test power for a pre-defined effect size (the remaining detection rate). For fur-
ther information about the ABX procedure and its parameters see Burstein, 1988, Clark,
1991,Leventhal, 1986.
WhisPER v1.9.0 – User Documentation
54 7. Test Procedures – Configuration and Execution
7.4.1 The Test Run
The subject’s GUI while performing an ABX test is divided into two parts (see fig. 7.18): In
the upper one, the subject can start/stop the audition of three stimuli whereas in the lower
section the subject may answer the question, which of the stimuli A or B equals the reference
X.
When hitting the buttons A or B or X, the respective stimulus will be started from the
beginning. In the lower section of the test subjects’s GUI (below an optional short ques-
tion/instruction definable in the DW ABX – edit main settings) you will find two answer
buttons A and B. After hitting one of them, the weiter-button will be activated, allowing the
subject to go on to the next trial. WhisPER registers only the last decision of a subject on
a trial, so it is possible to change the given answer multiple times before going to the next
trial.
Figure 7.18: The subject’s GUI during an A/B/X trial.
7.4.2 Setting up Properties
To perform the ABX test, the investigator has to:
WhisPER v1.9.0 – User Documentation
7.4 ABX Detection Test (ABX) 55
1. Define the number of trials n (at best using a calculation based on pre-defined type I
and II error levels and effect size which can be done directly in the DW “ABX – edit
main settings (...)”),
2. Assign the stimuli to A and B, and
3. Define the instruction texts for the subjects.
Figure 7.19: ABX Trial Main Settings
DW “ABX – edit main settings (...)”
initial subject instruction (input field)
The text shown to the subject before the ABX test starts
short instruction (input field)
The text shown in every trial on the subject’s GUI
number of trials n (input field)
Number of repetitions, Value that defines the length of the ABX test
number of correct answers c (input field)
Setting this value and fill out the ”number of trials”-field and press the ”Calculate”-
button will result in the corresponding ”Significance Level”
WhisPER v1.9.0 – User Documentation
56 7. Test Procedures – Configuration and Execution
Significance level alpha (input field)
setting this value as well as the ”Type II error risk” and the ”Effect size” while leaving
empty the ”Number of trials”-field and the ”Number of correct answers”-field will
– on operating the ”Calculate”-button – result in the corresponding entries to be
filled in automatically.
type 2 error risk beta (input field)
see ”Significance level”
Effect size (input field)
see ”Significance level”
Set Stimuli (input field)
by pressing ”Edit Assignment” the DW ABX – stimulus assignment(...) (see Fig. 7.20
will be opened. Here, one may define, which of the stimuli (as defined in the main
menu) shall be compared in the ABX test. By pressing the ”Assign to ...”-buttons
one may assign certain stimuli to be either stimulus A and B in the test. By hitting
the ”refresh”-button in the main settings menu, the assigned stimuli will show up
in the ”Set Stimuli” display, too, for control. Note again here, that while A and
B are assigned to be fixed throughout the whole test, the reference X is of course
randomly assigned to be either A, or B resp. in each trial.
WhisPER v1.9.0 – User Documentation
7.4 ABX Detection Test (ABX) 57
Figure 7.20: Assigning the stimuli to be judged in the current trial.
WhisPER v1.9.0 – User Documentation
58 7. Test Procedures – Configuration and Execution
7.5 Spatial Audio Quality Inventory (SAQI)
The Spatial Audio Quality Inventory (SAQI) is intended for a qualitatively differentiated,
comparative auditory assessment of real, imagined and simulated acoustic scenes in order to
to reveal specific shortcomings of a simulation under test and allow for a directed technical
improvement. The SAQI comprises 48 verbal descriptors of perceptual qualities assumed to
be of practical relevance when comparing virtual environments (VAEs) to real or imagined
references or amongst each other. It was generated by a Focus Group of 20 German experts for
virtual acoustics. Five additional experts helped verifying the unambiguity of all descriptors
and the related explanations. Moreover, an English translation was generated and verified
by seven bilingual experts. Rationale and methodology pursued in constructing the English
(SAQI-EN) and the German (SAQI-GER) vocabulary are described in more detail in Lindau
et al. [2014a] and Lindau et al. [2014b].
The SAQI vocabulary in its entirety (including perceptual descriptors, circumscriptions, scale
end label, and - if given - illustrative sound examples) is intended to enable experts in the
field to train any laymen to use it for assessments of VAEs. An extensive Test Manual is
freely available for download (Lindau [2015]). It is strongly recommended that you read it, in
order to better understand the following descriptions of the SAQI test in WhisPER. Further,
there is a SAQI project website at http://www.ak.tu-berlin.de/saqi.
7.5.1 The Test Run
The SAQI is a rather extensive test instrument, which can be widely customized to the user’s
specific needs. If one chooses to conduct a complete SAQI test, rating a singular perceptual
difference quality will involve presenting five concurrent GUIs (Graphical User Interfaces) to
the listener:
1. quality name and description
2. rating scale
3. assignment of modifications (2 GUIs)
4. choice of assessment entities
Whether this comprehensive way of testing will be actually required will of course depend
on the researcher’s aims. Whereas only the second GUI will be mandatory (quality name
and rating scale) the other four may be optionally used in a test. As an example, in case
of more exploratory research questions one might decide to collect as much information as
possible, whereas for a more confirmatory study, the SAQI test may be reduced to cover
only specific aspects. Hence, the user may conveniently chose from the test settings which
assessments shall be conducted (i.e. which GUIs will be shown to the user) in a test run. In
the following section it will be shown how a (complete) SAQI evolves from the view of the
test subject. In section 7.5.2 it will be explained, how a SAQI test may be parameterized by
the experimenter.
By the time, SAQI tests may be conducted in either English and German language (see
below). A French translation is currently in the review phase and should be available by the
end of 2015.
WhisPER v1.9.0 – User Documentation
7.5 Spatial Audio Quality Inventory (SAQI) 59
Description GUI (Fig. 7.21)
This GUI is intended for instructing test subjects to focus on a specific auditory quality to be
assessed. For each quality it contains the quality’s name and a closer circumscription. Note:
As described in the SAQI Test Manual (Lindau [2015]) all subjects should be suitably trained
to the full understanding of the SAQI beforehand. Therefore, presenting the qualities
circumscriptions here is only thought to work as a short reminder.
Figure 7.21: Example of a description GUI presenting a name and circumscription of an auditory qual-
ity
Rating GUI (Fig. 7.22-7.27)
This GUI is intended for rating the perceived amount of the (described) auditory difference
when comparing a test stimulus to a given or imagined reference. For each quality it presents
its name, the rating scale and play buttons for playing back the reference and the compari-
son/test stimuli. The ’OK’-button saves the rating and guides the subject to the next GUI.
The design of this GUI will change slightly according to the type of rating scale that belongs
to a quality. Most of the qualities in SAQI were found to demand bipolar rating scales repre-
sented by sliders, however, two qualities, the ’overall difference’ and ’sequence of events’, are
understood as unipolar. To support a better comparability of ratings, unipolar sliders are
displayed with half the length of the bipolar ones.
The two qualities ’horizontal direction’ and ’vertical direction’ are rated using an edit field
allowing subjects to directly type in the perceived difference in degree instead of a slider.
In addition, there are two radio buttons for indicating the perceived difference in direction
(clockwise, or counterclockwise).
Finally, one SAQI quality was defined to be dichotomous (difference in ’front-back position’).
This descriptor will be rated by using radio buttons: ’not confused’ and ’confused’.
If required, up to nine sliders, text fields, or pairs of radio buttons may be presented on one
rating screen, this way allowing more efficient testing of multiple stimuli at a time.
Note: For rating the first quality (’overall difference’) the user has to listen at least once to
the stimuli (i.e. use the play button), otherwise the ’OK’-button won’t be enabled. In case
WhisPER v1.9.0 – User Documentation
60 7. Test Procedures – Configuration and Execution
the test conditions/stimuli, or the assignment of the test conditions/stimuli to the play A,
and Bbuttons is randomized, the user has to listen at least once to the stimuli on each rating
screen.
Note: By clicking on the text field showing the name of the auditory quality, the description
GUI will pop up again.
Figure 7.22: Example of a GUI for rating a quality on a bipolar, open-ended rating scale
Figure 7.23: Example of a GUI for rating a quality using a unipolar, closed/open-ended rating scale
WhisPER v1.9.0 – User Documentation
7.5 Spatial Audio Quality Inventory (SAQI) 61
Figure 7.24: Example of a GUI for rating a quality using direct bipolar, closed-ended rating
Figure 7.25: Example of a GUI for rating a quality using a dichotomous scale
WhisPER v1.9.0 – User Documentation
62 7. Test Procedures – Configuration and Execution
Figure 7.26: Final GUI asking for formulating and rating any so-far unmentioned differences (SAQI
qualifier: other)
Figure 7.27: Example for a rating GUI with eight sliders which allows rating multiple stimuli at a time
WhisPER v1.9.0 – User Documentation
7.5 Spatial Audio Quality Inventory (SAQI) 63
Optional GUIs (Fig. 7.28-7.30)
Since both, a closer description of the perceived qualities by indicating certain modifications,
and the assignment of the perceived differences to certain assessment entities is optional,
these GUIs will only appear if they were selected before in the test set up.
Note: Optional GUIs are not available if more then one test condition/stimuli are rated at
once (c.f. Fig. 7.27).
The GUIs contain questions referring to the modifications and assessment entities for each
single descriptor and will be displayed right after the rating of an perceptual difference in the
main GUI. More information about the meaning of modifications and assessment entities can
be found in the SAQI Test Manual (Lindau [2015]); related information on the SAQI test
setup can be found in sec. 7.5.2.
Please note, that even if the modifications and/or entities are enabled for the test run, they
are not going to be displayed if a quality was not rated or was rated with 0 (no difference).
Figure 7.28: GUI for the rating the temporal variability of perceived differences
Figure 7.29: GUI for the rating the cause of perceived differences
WhisPER v1.9.0 – User Documentation
64 7. Test Procedures – Configuration and Execution
Figure 7.30: GUI for assigning assessment entities to perceived differences
7.5.2 Setting up Properties (Fig. 7.31)
Figure 7.31: SAQI Main Settings
DW “SAQI – edit General Settings (panel)”
Language (drop-down list)
As mentioned in the beginning, it is possible two choose between conducting SAQI tests in
either German or English language. The default language is German. In programming the
SAQI, possibilities have been foreseen for adding new languages. In case you plan to add a
new national version of SAQI, please check section 7.5.3.
Show circumscription (check box)
Check to show the written circumscriptions of the perceptual qualities in a separate window
before assessing each quality. The default setting is ’on’.
WhisPER v1.9.0 – User Documentation
7.5 Spatial Audio Quality Inventory (SAQI) 65
Edit perceptual qualities (push button; Fig. 7.32)
Pushing the ’Edit’ button opens another window for selecting the perceptual qualities. It
contains all 48 original descriptors included in the SAQI. By default they are entirely selected.
Furthermore, all qualities are subdivided in categories.
In order to customize a SAQI test, it is possible to either select singular qualities separately
(left window pane) or to select entire categories (right window pane).
Please note, that the first (’overall difference’) and the last quality (’other’) must be selected,
otherwise a warning window will pop up, telling you to do so. If qualities are chosen by select-
ing complete ’categories’ (right window pane), these two qualities are selected automatically.
The ’De/Select all’ button enables you to select or de-select all (also including both, the first
and last) qualities.
Figure 7.32: GUI for the selection of perceptual qualities. Left: singular selection, right: category-
based selection.
Randomize qualities (check box)
Randomizes the selected perceptual qualities during test presentation according to the fol-
lowing scheme: Firstly, the order of categories is randomized, and secondly, the order of
qualities within the categories. Exceptions are again the first and the last quality, which will
be invariably presented at first and last position, respectively.
Ask for modification (check boxes)
(Only available for single condition/stimulus SAQI, see sec. 7.5.2)
By default, all three types of modifications a pre-selected (checked) for assessment. Please
note, that modification 2 depends on modification 1. Hence, if modification 1 is disabled
for assessment, modification 2 will automatically be disabled, too. Modification 2 cannot be
chosen for exclusive assessment.
1. modification 1 (temporal variation): constant/varying periodically or otherwise rule-
based/varying non-regularly
2. modification 2 (temporal variation): continuous/discontinuous
3. modification 3 (causality): depending on scene events/depending on user interaction/independent
WhisPER v1.9.0 – User Documentation
66 7. Test Procedures – Configuration and Execution
Ask for assessment entities (check box / push button; Fig. 7.33)
(Only available for single condition/stimulus SAQI, see sec. 7.5.2)
By pushing the ’Edit’ push button it is possible to define five entities a perceived difference
may be further ascribed to. If needed, the default ones can be overwritten. Furthermore, the
entities can be selected exclusively by the check boxes to the left. The last two entities ’I
don’t know’ and ’other’ cannot be renamed. By default, all entities are enabled.
Figure 7.33: Default assessment entities
DW “SAQI – Scale settings (panel)”
Set scale steps (radio button)
The user can choose between the scale steps ’7’, ’5’ and ’3’. The default setting is ’5’. Scale
steps only refer to the bipolar and unipolar qualities.
For example: A bipolar scale with ’7’ steps will range from +/-3, where 0 marks the middle
(’no difference’). At the same time, an unipolar scale will range from 0 to 3, enabling to
encode the amount of a perceived difference, only, and not its direction.
Important note: Please note that scale steps are intended only as visual guidance for test
subjects. Internally, all ratings will always be coded in a range of -1 to +1 in floating point
resolution independent of the chosen number of scale steps to be displayed.
Randomize scale poles (check box)
Check to randomize scale orientation and corresponding label while test presentation. The
default setting is ’off’. During default setting scales are arranged in a way that perceived
increases are ’logically’ encoded by moving a slider upwards. The label, that is listed first
in the definition file saqi scale label language.m is coded with −1 (bipolar/dichotom/degree
quality) or 0 (unipolar quality), regardless of randomization. The label that is listed in
second place always is coded with +1. Hereby, throughout the whole SAQI the logic was
followed to orient scales by default in a way that positive values may intuitively related to a
perceived increase of the respective quality. Hence, as an example, a positive rating value of
the distance-quality indicates an increase in perceived distance, or, as another example, for
the sound color-items positive values will naturally indicate perceived amplifications/boosts.
WhisPER v1.9.0 – User Documentation
7.5 Spatial Audio Quality Inventory (SAQI) 67
DW “SAQI – Subject instruction (edit text field)”
Space for entering terms of instructions that will be displayed on the subject’s basic screen
(see sec. 6) before the test section is started.
DW “SAQI – Set stimuli (panel)”
Number of groups, test and anchor conditions (text edit fields)
These three parameters can be used to define the number and kind of test conditions. Their
behaviour is best explained by three examples:
1. Single stimulus SAQI
One test/comparison stimulus is compared to a given or imagined reference, see
Tab. 7.5.2, (1). For example a binaural simulation (comparison stimulus) could
be compared to the sound field of a loudspeaker (reference). In this case on slider
will be displayed per rating screen and modifications and assessment entities can
be included in the test (see above).
2. Multiple stimuli, single group SAQI
Multiple test/comparison and anchor stimuli are compared to a given or imagined
reference, see Tab. 7.5.2, (2). For example, auralizations obtained from different
room acoustic modeling algorithms (test conditions) could be compared to binaural
recordings (reference), while also providing an auralization using a monaural room
impulse response (anchor). In case the number of test conditions, plus the number
of anchor conditions is larger than nine, ratings will be split among two or more
rating screens. The anchor condition(s) will be provided on every rating screen,
however only the last rating of each anchor conditin will be saved.
3. Multiple stimuli, multiple groups SAQI
Groups of multiple test/comparison and anchor stimuli are compared to given
or imagined references, see Tab. 7.5.2, (3). To understand hza is meant simply
expand the example given above (2) for similar conditions but using, say, two
different audio contents (groups). If one group contains more than four conditions
(test plus anchor), only one group will be displayed per rating screen.
Parameter (1) (2) (3)
Number of groups 1 1 2
Number of test condi-
tions per group
1 6 3
Number of anchor con-
ditions per group
0 1 1
Table 7.1: Exemplary parameters for obtaining different SAQI modes: (1) Single stimulus; (2) Multiple
stimuli, single group; (3) Multiple stimuli, multiple groups
Randomization of groups and conditions (drop-down list)
1. none — Groups and conditions are presented in the order specified by the investigator.
WhisPER v1.9.0 – User Documentation
68 7. Test Procedures – Configuration and Execution
2. per subject — Groups and conditions are randomized per subject, but their order
remains identical across auditory qualities.
3. per quality — Groups and conditions are randomized per subject and quality. Note
that in this case the subjects have to listen to each stimulus for each auditory quality
before being able to rate (sliders are disable beforehands).
Compare to inner reference (check box)
Check this option to allow the user to rate the test stimulus against his/hers inner reference.
In this case, the second play button and stimulus will be omitted in the respective GUIs and
GUI instructions will be reformulated accordingly. Refer to the SAQI Test Manual Lindau
[2015] for hints towards a proper test subject instruction.
Randomize Assignment of stimuli to ’A’ and ’B’ button (check box)
Check to randomize assignment of the the stimuli ’A’ and ’B’ to play buttons when assessing
the next perceptual quality. In this case, the subject must listen to the stimuli before being
able to rate the current quality. This will noticeably increase the test duration. If stimuli are
randomized, the rating is decoded by switching the sign, and thus, the effect of randomization
is compensated in the rating.
Stimulus assignment across conditions (drop-down lists)
Stimuli from the global stimulus pool can be assigned to conditions by first selecting a con-
dition from the top-left drop-down list and, second selcting a stimulus from the top-right
drop-down list. The drop-down list at the bottom gives an overview of the current assign-
ments (Fig. 7.34). Stimuli are specified by their ID and name as specified by the investigator.
Figure 7.34: Stimulus assignment across conditions
GUI settings (push botton, text edit fields)
Controls the maximum and minimum number of conditions and groups to be displayed on
each rating screen, and the positions of the scale labels (Fig. 7.35). The latter was introduced
to provide correct positions for various operating systems and Matlab versions. Only change
these parameters if the default bahviour does not match your needs.
7.5.3 How to: Implementing SAQI for a New Language
Adding a new language is possible, and the required steps are described in the following.
However, this reconfiguration will need some understanding of Matlab and is not thought
to be a normal usage procedure of WhisPER. Hence, if you do not feel save with these
requirements, please contact us for help. We can also - as a service - add national versions
WhisPER v1.9.0 – User Documentation
7.5 Spatial Audio Quality Inventory (SAQI) 69
Figure 7.35: Settings for distribution of conditions across rating screens.
of SAQI to WhisPER. In any case, please send us the new language files, so we can update
the WhisPER release on the project homepage.
Step 1: Create new language files
As a first step, it will be necessary to re-create some of WhisPER’s Matlab scripts in your
desired language, containing the names for:
•quality descriptors (original example file: saqi items german.m),
•categories (original example file: saqi categories german.m),
•circumscriptions (original example file: saqi definitions german.m),
•GUI phrases (original example file: saqi phrases german.m),
•scale label ends (original example file: saqi scale label german.m),
•assessment entities (original example file: saqi entities german.m)
The best way will be to copy the original (German or English) versions of these files at a
save place and change them accordingly, one by one. The structure of these files is pretty
straight-forward and may be checked out by retrieving them from the WhisPER source code
folder. Please note, that your new files need to be known in the Matlab’s path environment
(at best they are saved in the WhisPER source code folder).
In case letters with accents occur in the implemented language please decode them as done in
saqi scale label german.m to ensure compatibility across different system and system locales.
Step 2: Adding a new language case to the setup GUI
Now, you should update the language selection switch of the setup GUI by adding another
case for your language. You can find the program code in the saqi edit main file. The switch
is arranged in a function called ’pop language Callback’ (line 365 et seq.).
The easiest way of adding, is to copy-and-paste one of the existing cases and to rename the
applied m-files (e.g. saqi items french) in the new one (e.g. case 3).
Step 3: Additions on the GUI
WhisPER v1.9.0 – User Documentation
70 7. Test Procedures – Configuration and Execution
For the last step you will need to edit the figure ’saqi edit main.fig’, for example using the
GUIDE-Tool of Matlab (right-click on the fig-file and choose ’open with GUIDE’).
Next, double click on the language drop-down list in the panel ’General Settings’ and a Matlab
’Inspector’ will open. In this ’Inspector’ it is possible to change settings. Open the edit-box
by double clicking on ’String’ an add your language under the last enlisted.
7.6 ITU-R Rec. BS.1116-1 (ABC/HR) & ITU-R Rec. BS.1534-1
(MUSHRA)
ABC/HR and MUSHRA are methods for rating multiple conditions against one or more
references. ABC/HR includes a detection task according to the ABX paradigm and should
be used when differences between test and reference conditions are expected to be small,
whereas MUSHRA is recommended for usage with intermediate differences.
7.6.1 The Test Run
The rating GUIs (Fig. [7.36, 7.37]) are subdivided in three parts: At the top, the question
stating the task for the subject is displayed. The middle part shows the sliders and text fields
for rating the test conditions, as well as the play and stop buttons, and at the bottom, the
ratings can be saved or reset. Sliders and edit fields are initially disabled and will be enabled
after the subject listened to the corresponding stimuli. The OK -button will be enabled after
the subject listened to all stimuli.
Figure 7.36: ABC/HR rating GUI
WhisPER v1.9.0 – User Documentation
7.6 ITU-R Rec. BS.1116-1 (ABC/HR) & ITU-R Rec. BS.1534-1 (MUSHRA) 71
Figure 7.37: MUSHRA rating GUI
7.6.2 Setting up Properties (Fig. 7.38)
DW “ABC/HR & MUSHRA – edit General Settings (panel)”
Type (drop-down list)
Choose ITU-R Rec. BS.1116-1 (ABC/HR) or ITU-R Rec. BS.1534-1 (MUSHRA) depending
on your needs. Thill will affect the layout of the rating screen (see Fig. [7.36, 7.37]).
Set scale steps (radio button)
The user can choose between the scale steps ’7’, ’5’ and ’3’. The default setting is ’5’. A
scale with ’7’ steps will have seven eqidistant markers wich can be individually labeled (see
below).
Important note: Please note that scale steps are intended only as visual guidance for test
subjects. Internally, all ratings will always be coded in a range of -1 to 0 (MUSHRA), and
-1 to +1 (ABC/HR) in floating point resolution independent of the chosen number of scale
steps to be displayed. In both cases, ’0’ denotes that the subject perceived no difference, i.e.
the sliders were not moved, and ’-1’ denotes a large difference, i.e. the slider was moved all
the way down. In case of ABC/HR, positive numbers indicate the the subject failed to detect
the reference, i.e. the subject moved the wrong slider.
Scale label assignment (drop-down lists & text edit field)
Individual labels can be assigned to the scale markers, by selecting a marker using the top
left drop-down list and writing the label into the edit field on the top right of the panel. The
drop-down list at the bottom can be used to show the currently assigned label.
Note that the default label (’identical’, and ’different’) differ from the suggestions made
by the ITU (ABC/HR: ’Imperceptible’, ’Perceptible, but not annoying’, ’Slightly annoying’,
’Annoying’, ’Very annoying’; MUSHRA: ’Excellent’, ’Good’, ’Fair’, ’Poor’, ’Bad’). This was
done because the ITU suggestions are believed to be multi-dimensional (i.e., mixing evaluative
WhisPER v1.9.0 – User Documentation
72 7. Test Procedures – Configuration and Execution
Figure 7.38: ABC/HR and MUSHRA setting
and magnitude related connotations) and non-equidistant and therefore possibly bias the test
results.
DW “ABC/HR & MUSHRA – instructions (panel)”
Subject Instruction (text edit field)
Use this to specify the instructions that subjects will read before rating.
Rating Instruction (text edit field)
This is the question that will be displayed at the top of the rating GUIs (Fig. [7.36, 7.37]).
DW “ABC/HR & MUSHRA – set stimuli (panel)”
For Number of groups, test and anchor conditions,Stimulus assignment across conditions,
and GUI settings see Sec. DW “SAQI – Set stimuli (panel)” on page 67.
Group randomization (check box)
If enabled, the presentation order of groups will be randomized. Elsewise, the presentation
order will be as specified.
Condition randomization (check box)
If enabled, the presentation order of conditions (text, and anchor) will be randomized. Else-
wise, the presentation order will be as specified.
WhisPER v1.9.0 – User Documentation
73
8 Test procedures - Default Values’ Presets
This section is about a feature of WhisPER that allows the user to manipulate all the proce-
dures’ default values (as they are mentioned in chap. 7), which the procedures are initialized
with upon adding them to a test series via the test section setup (cf. sec. 5.4).
The DW that allows you to do this is accessible via the main window’s menu edit\procedure’s
defaults.
Figure 8.1: DW Configure defaults’ presets
This DW consists of a middle part comprising four tabs containing all the configurable default
values for all the procedures (plus for the stimuli), a lower part consisting of four buttons for
WhisPER v1.9.0 – User Documentation
74 8. Test procedures - Default Values’ Presets
closing, writing, resetting and canceling your configurations and an upper part consisting of
four preset buttons, each selecting one of four presets. Following that order the description
in detail:
Adaptive psychophysical threshold procedures (tab) This tab contains spaces for all the
configurable defaults of the parameters of the AP procedures as explained in sec. 7.1.2.
Please note: As stimulus definitions take place separately from procedure setup (cf.
sec. 5.2 and 5.4 of chap. 5) all AP configuration options involving (already defined)
stimuli in any way (e. g. stimulus assignments to intensities or reference level) sensibly
simply cannot be available here. They have to be set later.
Please also note that as this DW is intended for the power user all the input fields are
presented in a condensed and often abbreviated way, thus lacking the guidance of the
DWs of the in procedure setup as in sec. 7.1.2, but instead allowing all the options to
be set in one window only in the flattest hierarchy and the fastest way possible. Please
note again that changing values here will of course not disable you from making changes
later, after adding a procedure to your test series – changing values here will only affect
the values your procedure already has just after you have added it. Thus it is perfectly
possible to just once set those values here that are most likely to not change from
listening experiment to listening experiment (like for example the psychometric function
or the a priori p. d. f. or the number of trials etc.) and thus are always initialized to your
defaults, and later fill in those values that are specific to a single listening experiments
(like e. g. the participants’ instructions).
Of course you can also use this DW to just make all your wanted changes for one specific
listening experiment in case you simply prefer the availability of all the configuration
options in one place without having to click through multiple configuration DWs.
Repertory grid technique / Semantic differential (tab) This tab contains spaces for all the
configurable defaults of the parameters of the RGT procedure as well as the SD proce-
dure as explained in sec. 7.2.2 and in sec. 7.3.2 respectively. For notes and hints please
see above.
A / B / X trials (tab) This tab contains spaces for all the configurable defaults of the parameters
of the ABX procedure as explained in sec. 7.4.2. For notes and hints please see above.
Stimulus (tab) This tab contains spaces for all the configurable defaults of stimuli (cf. sec. 5.2),
see fig. 8.2. Notes and hints above also apply here, further explanation see below.
New Stimulus default name (input field) Space for entering what is to be the default
name of a stimulus each time one is created either by pushing the new button
or the multi new button in the stimulus pool DW (as explained in sec. 5.2). As
changing this from ” - no name - ” to whatever you desire will effectively inhibit
automatic name generation for stimuli based on a .wav file name, this will only
be useful under two conditions: First, when you just do not want your stimuli to
be named corresponding to the .wav files loaded, and second, when there are no
.wav files to be loaded to begin with, i. e. your stimuli are OSC only. Please be
advised, that in both cases the name entered here will only be a ”base name”, as
it will result in the same default name for every stimulus generated. With that
”base name” already filled in the stimulus’ name field, however, individualizing
each name by simply manually appending a number, for example, is facilitated a
lot. Cf. below for OSC usage scenarios.
New Stimulus default OSC data (panel) Contains spaces for entering what is to be the
default OSC parameters of newly generated stimuli. As explained above for the
New Stimulus default name the values entered here will result in exactly the same
WhisPER v1.9.0 – User Documentation
75
default OSC data values for every stimulus newly generated. Thus their use is to
serve as ”base parameters”, with later only individual addition of e. g. numbers in
the OSC data field for each stimulus facilitating the creation and configuration of
a bigger set of OSC only stimuli.
Figure 8.2: DW Configure defaults’ presets – tab ”Stimulus”
Ok (button) This button will save your changes and close the DW. Please note that ”saving
your changes” means making them immediately effective (for the addition of test pro-
cedures for example) as well as storing them so they stay the way you configured them
even over the course of exiting the program (WhisPER as well as MATLABr) and
starting it again.
Write (button) This button will save your changes and leave the DW open. Again ”saving your
changes” means making them immediately effective (for the addition of test procedures
for example) as well as storing them so they stay the way you configured them even over
the course of exiting the program (WhisPER as well as MATLABr) and starting it
again.
WhisPER v1.9.0 – User Documentation
76 8. Test procedures - Default Values’ Presets
The purpose of not closing the DW but making your changes effective is the following:
You might have noticed that when you open the Configure defaults’ presets DW it does
not force the main program window to disappear or close. As the addition of a test
procedure to a test series (or a test section respectively) or stimuli to the pool are not
done in the main window but ”some DWs away”, having to go back and forth after
changing the defaults and wanting to add a test procedure would result in a lot of
unnecessary clicking. So instead the Configure defaults’ presets DW stays open after
pushing Write, all changes are effective, you navigate to e. g. the test section setup and
add a procedure and you can still change the defaults in this still open DW, push write
again and your new defaults are immediately in effect again.
Cancel (button) This button will close the DW discarding your changes – neither will they
come to effect, nor will they be stored for later. Please note that the state of this button
will also signal to you if changes have been made at all – it will only be enabled if they
have.
Reset (button) This button will reset the values and selections in all the above explained
four tabs – which constitute the current preset (see below) – to the values originally
intended by the programmers of WhisPER and made available to you when you first
got the software (as introduced above in sec. 7.1.2, sec. 7.2.2, sec. 7.3.2 and sec. 7.4.2).
The reset values will also be immediately effective and will be stored for later, too – this
means a ”Reset” action cannot be canceled via the Cancel button! Please note that the
state of this button will also signal to you if the values in the current four tabs differ at
least in one respect from the original provided ones – it will only be enabled if they do.
Preset 1-4 (buttons) These buttons select one of the corresponding four presets. A preset
comprises all values and selections of all the four tabs, and not just the momentarily
visible one. So preset 1 has a tab for AP, a tab for RGT/SD, a tab for ABX and a
tab for stimulus, and these four make preset 1, preset 2 has a tab for AP, a tab for
[. . . ] and so on. Changing tabs and entering values will only affect the current preset,
symbolized by the preset button at the top that is pushed. Pushing Ok,Write and
Reset will also only act on the current selected preset 14. Changing presets by pushing
a different preset button at the top will do two things: a) It will always store the preset
you are leaving for later (this means you cannot have unsaved changes over the course
of a preset change), and b) it will make the newly selected preset immediately effective.
This preset system can be used in an even more refined way than what was described
above along with the Write button: Leaving the Configure defaults’ presets DW open
in the background and having configured two presets differing e. g. only in the types of
the adaptive mechanism and their corresponding settings, it is easily possible to first
select one of those presets, add a new track to your AP test section, then select the other
preset and just again add a new track – thus with just four clicks you really quickly gen-
erated two tracks containing the same procedures, albeit with totally different settings
– enabling e. g. the comparison of different adaptive mechanisms.
Previously unaltered presets default to the initial defaults (cf. above, (Reset button),
presets stay persistent over the course of exiting the program and opening it again, and
the choice of which preset is currently active is persistent to ending the program and
starting it up again, too.
Presets are stored each in a file in the program folder, named PC.1.mat to PC.4.mat,
these files are only generated when that preset has been selected at least once.15 Whis-
14And Cancel will actually ”act” on nothing
15The original defaults are present in the file PC.mat, this should better not be touched.
WhisPER v1.9.0 – User Documentation
77
PER will not crash if one or even all of the PC.n.mat is missing – even if that preset
had been the selected one – but will simply generate a new one as a clone of the original
defaults values. 16
16Should you wish to use more than the provided four presets, you could thus simply move (or rename) your
configured PC.n.mat files and later exchange them or put them back – WhisPER will read them as if they
had never been elsewhere.
WhisPER v1.9.0 – User Documentation
78 9. Program Setup
9 Program Setup
In this chapter the approach of adjusting the program settings will be described. These refer
to technical properties which are mostly independent of methodical issues and stored in the
file PS.mat which will be described in section 10.1.7.
9.1 Basic Audio Setup
As WhisPER currently uses the internal MATLABrfunction audioplayer, there are no
general audio properties to be set.
9.2 Basic Network Setup
Basic network settings can be specified by selecting the item program setup\network from
the menu bar. Then a DW will appear for entering host names or IP addresses respectively
and port numbers of up to 6 OSC-servers. Each server is related to a single parameter which
has to be determined when defining the stimuli’s contents (see sec. 5.2). As in some cases
one might want to address several parameters on one server, the servers need not necessarily
be different. Note that the order in which the servers are defined has to be the same as the
order in which the respective OSC-commands should be executed.
WhisPER v1.9.0 – User Documentation
79
10 Data Handling
In this chapter basic information about the program’s data structure will be provided as far
as it is required to make use of the functionalities implemented. The aim is not to give a
complete description which one needs to make modifications or to work on upgrades or kind
of these things. For the latter purposes one is directed to the technical documentation (see
sec. 1.3.4). There are basically two kinds of data sets which may be accessed by the user.
One contains all data belonging to a certain test series and is represented by the test series
folder which may be placed at an arbitrary location (cp. sec. 5.1). The second data set is
contained in the file PS.mat which is located in the program folder ‘2 mfiles’ and consists of
global program settings (cp. chap. 9). Moreover there is a sub folder containing this user
documentation.
10.1 The Test Series Folder
This folder is quite essential to know for the user as it contains both the test series’ empirical
data set and its setup including audio data. In the following subsections the several files and
subfolders, which will be automatically generated if a new test series is created, are described.
10.1.1 The File ‘testseries.info’
This file contains information about the configuration of the test series and may be regarded
as a documentation of the test series setup that should enable the user to understand the
information provided by the log files or the export sheets. This information can be summed
up by the following aspects:
•Global settings: What stimuli have been defined? How many sessions have been set?
•Decoding of test section key numbers: What are the label names? What type of test
procedure is used in each test section?
•Properties of the test procedures: How are stimuli grouped and assigned to single
test units (elements, objects, triads, tracks)? What scales (format/labels) have been
defined? (RGT/SD) Which response paradigm and which adaption mechanism have
been selected? (AP)
•Decoding of session key numbers (experimental conditions): What subsets of test units
have been selected and what sequences have been defined (manually defined vs. random
sequences)?
Note: The file also serves as an identifier whose existence is a premise for the test series
folder to be accepted as such by the program.
WhisPER v1.9.0 – User Documentation
80 10. Data Handling
10.1.2 The Files ‘TSD.mat’ and ‘TSP.mat’
These two files constitute some of the program’s most important internal data storages which
in general should be handled with great care and therefore not be manipulated by the common
user. Though there are a few exceptions which can be applied to simplify the experimenter’s
life and which, besides a short general description of the files’ contents, will be noted in the
following two paragraphs.
TSD.mat (“test series data”) In this file all so far collected empirical data (i.e. more precisely
the contents of the dependent variables which are to be exported) are stored. The only
occasion for the common user to manually access this file is when one wants to clear all these
data or remove them from the test series. This might be necessary either if one wants a kind
of “virginal” test series folder (e.g. for replication of the test series) or if one wants to reset
the counter of the current test run (i.e. more precisely of both the counters of the latest
completed and the next test run) in order to be able to make some changes on the test series’
setup after the latter has already been completed (cp. chap. 5). The task at hand will be
performed by deleting the file TSD.mat or at least moving it to a location outside the test
series folder if one wants to preserve the data. Additionally it is recommended to clear the
contents of the sub folders plot,export and logfiles. Otherwise this could lead to confusion as
old and new files will be mixed and old files might be gradually overwritten (not in general
but in most cases if old and new files carrying the same labels). Do not forget also to make
a copy of the folders’ contents if you want to preserve the data.
Important Note on SAQI! Please note, that as referring to the previous paragraph SAQI-
tests resemble an exception. In case of SAQI tests the TSD.mat should NOT be deleted,
because it contains all empirical data collected from all subjects so far! Instead, it should be
saved after every test run in another folder to avoid loosing collected data. As already stated
in section 5 you can find the TSD.mat in the current test series folder.
TSP.mat (“test series properties”) Here all data concerning the test series’ setup are stored.
In the standard case it is not necessary for the user to manually access this file, as the setup
normally is performed by using the GUI system (see chap. 5). Anyhow in some cases it might
be necessary to export these configuration data to another test series folder if it is desired to
use a quite similar set up there. This can be done by copying the TSP.mat to the destination
test series folder and replacing the local file with it. Note that in such a case one also has to
copy the folder audio if the files in it should be played.
10.1.3 The Subfolder ‘audio’
This subfolder contains the audio files which are copied to there by the program in the context
of the stimuli’s definition (cp. sec. 5.2). Note that once an audio file has been uploaded, it will
not be deleted at a later time even if it is no longer used at a certain point. Of course, in the
latter case this may be manually done in order to save storage space. Note furthermore that
files all must have different names as otherwise the program will not be able to distinguish
between them and errors will occur. This can be ensured by using the GUI system to define
the stimuli. If then a .wav file of the same name already exists it will be overwritten.
WhisPER v1.9.0 – User Documentation
10.1 The Test Series Folder 81
10.1.4 The Subfolder ‘export’
This subfolder contains the export sheets, i.e. text files that include all so far collected
empirical data (i.e. more precisely the contents of the dependent variables which are to be
exported). The export sheets are saved as .csv files and therefore may directly be opened
by the spreadsheet software Excel. They can also be imported to the statistical evaluation
software SPSS17. Each file includes the data of a certain test section across all runs. Therefore
there are as much files as there have been test sections defined in a test series and the
information provided by a file’s name only consists of the section number. After each run the
export sheets are updated by being created anew. This can also be done manually at any
time by using the menu item test series\export empirical data (see sec. 4.1).
The internal structure of an export sheet is constituted in the following way. Each line
represents the data belonging to a different run. Lines are sorted in ascending run numbers.
Each column comprises a different variable’s values. The first line includes the column labels,
i.e. the variables’ names in an abbreviated format. Columns are separated by semicolons.
The first variable always is the run number, the second is the subject id and the third the
session key number. Further variables and columns respectively depend on the specific type
of test procedure applied. They will be stated in the following three paragraphs (see also
fig. 10.1).
Important Note on SAQI! Please note, that in case of an SAQI test run the previous expla-
nation is not valid. For extracting the empirical data please look up the paragraph about
SAQI in this section.
Adaptive Psychophysical Threshold Procedures For an AP the dependent variable is the
threshold value estimated in the end of each track. As there may be several tracks defined,
i.e. if tracks are interleaved and/or performed under different experimental conditions (i.e.
assigned to different sessions), each column comprises the threshold estimates of all the test
runs but for a different track. So if the number of tracks is given by T, there will be T
additional columns reserved. Thereby columns are arranged by ascending track numbers.
Note that there may be empty entries because not every track necessarily has to be performed
in every run. In many applications there is only one track defined and performed by each
subject, so that there is only one dependent variable and one additional column respectively
in the export sheet.
Repertory Grid Technique For the repertory grid technique there are two groups of elicited
data: the first is of a qualitative format and constituted by the elicited constructs, the second
is numeric data and consists of the ratings. Considering the first, the number of elicited
constructs depends on the test run. For each construct there are two verbal descriptors –
one for the high pole and one for the low pole18. So if Cis the maximum number of constructs
which has been reached in a run at all, there will be 2 ·Ccolumns reserved for the storage
17Tested on version 15.0. Note: Use the menu item “Read text data ...” and select the option “All Files (*.*)”
from the drop-down list in the file dialog to make .csv files visible.
18Only those constructs which have been used for rating, i.e. those that passed the editing in between the
two parts of the procedure, will be exported. Because the information about whether a descriptor evolved
from either a similarity or a contrast between elements is lost, we do not speak of initial and contrast pole
anymore.
WhisPER v1.9.0 – User Documentation
82 10. Data Handling
of the verbal descriptors. Moreover if E is the total number of different elements involved in
the rating process at all, there will be E·Cfurther columns reserved for storing the rating
values.
Semantic Differential For the semantic differential the elicited data consists of numerical
values evolved from the ratings. If Ois the total number of objects involved and Sis the
number of items (i.e. scales), then O·Scolumns will be reserved for the storage of the
dependent variables’ values.
ABX Test As results from the ABX test only the number of correct answers will be exported.
Please note: The absolute number of trials per condition is not exported! This means, that,
for later statistical analysis you will have to note this number by yourself! The number
of correct answers will be saved in 4th column in the export sheet, i.e. in the row of the
corresponding subject.
SAQI The internal structure of an SAQI-TSD-file is constituted in two levels. On the first
level one can find information about the run number, the subject ID, the session number and
the subject data. The second level is located in the fourth column named ’Subject Data’.
There, you will find the elicited ratings in a 49x30xN cell. Each line in this cell represents
the rating of one perceptual quality. Even if a quality/modification/entity was not included
or rated one will find an entry about it. N equals the number of test conditions (see table
10.1).
Important note on encoding of SAQI results!
When saving SAQI results, these are always encoded as if being assigned to the stimulus un-
der test when compared to the reference stimulus. This direction of encoding is also retained
in saved results, if test and references stimuli are chosen to vary randomly for each assessed
perceptual quality. Further, ratings are encoded to reflect perceived differences according to
a logical increase of the quality under test. Thus, raw SAQI ratings may be interpreted intu-
itively: Positive difference ratings in terms of perceived high frequency coloration, sharpness,
distance or clarity etc. refer to a perception of increase distance, emphasized high frequencies
and a sharper, more distant and more clear sound of the test stimulus as compared to the
reference.
In case of an open-ended bipolar or a unipolar quality the value of the slider will be suitably
transferred. In order to maintain an identical representation of values in the TSD.mat, in the
case of the two bipolar qualities with closed ends, the subject’s input will be divided by 180
before being saved. Furthermore, the entry will be set negative, if ’shifted clockwise’ (’hori-
zontal direction’) or ’shifted down’ (’vertical direction’) was chosen. For further information
about the TSD-entries please look up the schematic representation (see table 10.1).
For exporting and plotting the collected empirical data there are two Matlab functions avail-
able from here: //(http://dx.doi.org/10.14279/depositonce-1.
For further information on these files, please check section 9 in the SAQI Test Manual (Lindau
[2015]).
WhisPER v1.9.0 – User Documentation
10.1 The Test Series Folder 83
Other File Formats: GridXML and GRD Since there exist some software packages on the
market which are specialized in evaluating repertory grid data, export of their respective
(proprietary) and most widely adopted formats was implemented into WhisPER for users’
convenience. Files will be placed into the export subfolder anytime you hit the menu item
test series\export empirical data manually. (see sec. 4.1) Note: Only rating data of RGT and
SD sections will be written into these formats (for obvious reasons).
Idiogrid (http://www.idiogrid.com/) is a free package by James W. Grice with a lot of pow-
erful features like Multi Grid Analysis (MGA) and Adobe Illustrator (AI) export for figures
(file extension: GRD).
GridXML is an open grid format suggested by Martin Fromm, author of the commercial pack-
age Gridsuite (http://www.gridsuite.de).19 GridXML Files can be imported into Gridsuite
and converted to several other formats of commercial packages like Flexigrid, RepGrid/RepIV,
EnquireWithin, Gridstat and Gridcor in a second step.
19Walter et al. [2004]
WhisPER v1.9.0 – User Documentation
84 10. Data Handling
run subject session track 1 … track T
1 sub_id(1) ses_key(1) θ(1, 1) … θ(T, 1)
…
…
…
…
…
…
R sub_id(R) ses_key(R) θ(1, R) … θ(T, R)
T
x R
threshold estimates
(a) Adaptive psychophysical procedures
construct 1 construct C element 1 … element E run subject session low pole high pole …
low pole high pole construct1 …
constructC … construct1
…
constructC
1 sub_id(1) ses_key(1) lo(1, 1) hi(1, 1) …
lo(C, 1) hi(C, 1) r(1, 1, 1) …
r(1, C, 1) … r(E, 1, 1) …
r(E, C, 1)
…
…
…
…
…
…
…
…
…
…
…
…
…
…
…
R sub_id(R) ses_key(R) lo(1, R) hi(1, R) …
lo(C, R) hi(C, R) r(1, 1, R) … r(1, C, R) … r(E, 1, R) … r(E, C, R))
C x 2
x R
verbal descriptors
E x C
x R
ratings
(b) Repertory grid technique
object 1 object O run subject session item 1 item S … item 1 … item S
1 sub_id(1) ses_key(1) r(1, 1, 1) … r(1, S, 1) … r(O, 1, 1) … r(O, S, 1))
…
…
…
…
…
…
…
…
…
…
R sub_id(R) ses_key(R) r(1, 1, R) … r(1, S, R) … r(O, 1, R) … r(O, S, R)
O x S
x
R
ratings
(c) Semantic differential
Figure 10.1: Schematic representations of the data export sheets (cp. text on p. 81). Note that conceptually the schemes may include missing values, as not
every test unit (object, element, triad, track) necessarily has to be assigned to each test run.
WhisPER v1.9.0 – User Documentation
10.1 The Test Series Folder 85
run subject session section 1
1 sub_id(1) ses_key(1) correct answers(1, 1) ...
...correct answers(1, R)
T x R correct answers
correct answers(T, 1)
correct answers(T, R)
…
…
…
…
…
…
R sub_id(R) ses_key(R)
… section T
(a) ABX
run subject session session 1
1 sub_id(1) ses_key(1) subject data(1, 1) (T, 1)
…
…
…
…
…
…
R sub_id(R) ses_key(R) (1, R) (T, R)
T
x R
subject data sets
…
…
… session T
version(1, 1)
…
(1, R)
session 1
scale size(1, 1)
…
(1, R)
session 1
(b) Spatial Audio Quality Inventory - Level 1
1 ‘difference’
‘other’
…
…
48
30 properties per quality
‘other’ selection
selection ‘other’
48 qualities
…
Entity 7 NameQualitiy ID Qualitiy
... see table for details ...
…
entity 7
(c) Spatial Audio Quality Inventory - Level 2 ’Subject Data’
Figure 10.2:
WhisPER v1.9.0 – User Documentation
86 10. Data Handling
SAQI Empirical Data
Column number Column name Encoding
1 Quality ID 1: difference, 2, 3, 4 etc.... 48: other
2 Quality Quality Name
3 Category ID 0: ’difference’ and ’other’
other items have category 1-8
4 Category Category Name
5 Included 1: quality was included in test run
0: quality was not included in test run
6 Rated 1: quality was rated (subject heard a difference)
0: quality was not rated (subject did not hear a difference)
NaN: was not presented in the test
7 Modification 1 1: modification 1 was used
0: modification 1 was not used
8 Answer 1 1: constant
2: varying periodically or otherwise rule-based
3: varying non regularly
NaN: was not presented in the test
9 Modification 2 1: modification 2 was selected by the subject
0: modification 2 was not used
10 Answer 2 1: steady
0: not steady
NaN: was not presented in the test
11 Modification 3 1: modification 3 was used
0: modification 3 was not used
12 Answer 3.1 1: ’depending on scene events’ was used
0: ’depending on scene events’ was not used
NaN: was not presented in the test
13 Answer 3.2 1: ’depending on user interaction’ was used
0: ’depending on user interaction’ was not used
NaN: was not presented in the test
14 Answer 3.3 1: ’independent’ was used
0: ’independent was not used
NaN: was not presented in the test
15 Rating Value of the slider or radio button
NaN: was not presented in the test
16 Entity1str 1: assessment entities were included
0: assessment entities were not included
NaN: was not presented in the test
17 Entity 1 User specific name of entity 1
NaN: was not presented in the test
18 Entity1str 1: Entity 1 was selected
0: Entity 1 was not selected
NaN: was not presented in the test
19 Entity2str User specific name of entity 2
NaN: was not presented in the test
WhisPER v1.9.0 – User Documentation
10.1 The Test Series Folder 87
Column number Column name Encoding
20 Entity2str 1: Entity 2 was selected
0: Entity 2 was not selected
NaN: was not presented in the test
21 Entity3str User specific name of entity 3
NaN: was not presented in the test
22 Entity3str 1: Entity 3 was selected
0: Entity 3 was not selected
NaN: was not presented in the test
23 Entity4str User specific name of entity 4
NaN: was not presented in the test
24 Entity4str 1: Entity 4 was selected
0: Entity 4 was not selected
NaN: was not presented in the test
25 Entity5str User specific name of entity 5
NaN: was not presented in the test
26 Entity5str 1: Entity 5 was selected
0: Entity 5 was not selected
NaN: was not presented in the test
27 Entity6str User specific name of entity 6
NaN: was not presented in the test
28 Entity6str 1: Entity 6 was selected
0: Entity 6 was not selected
NaN: was not presented in the test
29 Entity7str User specific name of entity 7
NaN: was not presented in the test
30 Entity7str 1: Entity 7 was selected
0: Entity 7 was not selected
NaN: was not presented in the test
31 Test Condition Specifies: Test condition (Stimulus ID & name) and
Reference condition (Stimulus ID & name), e.g.
B1 Test condition (S7 G) vs. A1 Reference condition (S8 H)
32 Scale size 3, 5 or 7
33 SAQI version String, e.g. v.1.3
Table 10.1: Content of the TSD cell ’Subject Data’
WhisPER v1.9.0 – User Documentation
88 10. Data Handling
10.1.5 The Subfolder ‘log files’
This sub folder contains the log files of the runs that have been performed up to the current
point. A new log file will always be written right after a test run has been completed. A log
file’s name consists of the following information in given order: the run number, the subject
ID and the session key number. Note that once a log file has been created, it will not be
deleted automatically even if the run counter has been reset (cp. sec. 5). So if one wants the
log files to be deleted, e.g. to achieve a “virginal” test series folder, one has to empty this
sub folder by hand. Note that if a log file is to be created and one of the same name already
exists, the latter will be overwritten.
10.1.6 The Subfolder ‘plots’
This sub folder contains all data plots that have been created during the runs so far. At
the program’s current state plots are only produced for psychophysical adaptive procedures.
Each time after a procedure of this type has been performed, a MATLABrfigure depicting
the respective track will be automatically created. The respective files’ names consists of the
following information in given order: the run number, the subject ID, the session key number,
the section number and the track’s number. Additionally or alternatively to the MATLABr
figure format one may choose a different file format for the plot being saved in. This can be
done by using the menu item program setup\plottings from the main window’s menu bar (see.
sec. 4.1). Note that once a plot has been created, it will not be deleted automatically even if
the run counter has been reset (cp. sec. 5). So if one wants the plots to be deleted, e.g. to
achieve a “virginal” test series folder, one has to empty this subfolder by hand. Note that if a
plot is to be created and one of the same name already exists, the latter will be overwritten.
10.1.7 The File ‘PS.mat’ (“program settings”)
This file is placed in the program folder and contains all settings concerning the program’s
basic setup. These relate to more technical properties which are (mostly) independent of
methodical issues belonging to a certain test series. The standard procedure for adjusting
these settings is to use the GUI system (see. chap. 9). Though there may be some cases
when a different approach yields quicker results. This is for example when one wants to take
along these program settings to another computer or to switch between several set ups very
quickly and there is no time for making all the necessary inputs by hand. In such cases one
just might exchange the versions of the file PS.mat which is located in the program folder (‘2
mfiles’). Further adjustments then again can be made by using the GUI system.
WhisPER v1.9.0 – User Documentation
89
Bibliography
S. Bech and N. Zacharov. Perceptual Audio Evaluation. John Wiley & Sons, Ltd., Chichester,
2006.
J. Berg and F. Rumsey. Spatial attribute identification and scaling by repertory grid technique
and other methods. In AES 16th International Conference Munich, pages 51–56. Audio
Engineering Society, 1999.
J. Bortz and N. D¨oring. Forschungsmethoden und Evaluation. Springer, Berlin, 4 edition,
2006.
H. Burstein. Approximation formulas for error risk and sample size in abx testing. Journal
of the Audio Engineering Society, 36(11):879–883, 1988.
S. Ciba. Erstellung einer Softwarebibliothek f¨ur H¨orversuche - Programmkonzept und zu
implementierende Testverfahren. Magisterarbeit, FG Audiokommunikation, Institut f¨ur
Sprache und Kommunikation, Fakult¨at I, TU Berlin, 2008.
D. L. Clark. Ten years of a/b/x testing. In AES 91st Convention New York. Audio Engi-
neering Society, 1991.
T. N. Cornsweet. The staircase-method in psychophysics. American Journal of Psychology,
75:485–491, 1962.
J.M. Findlay. Estimates on probability functions: A more virulent pest. Perception &
Psychophysics, 23(2):181–185, 1978.
F. Fransella and D. Bannister. A manual for repertory grid technique. academic press,
London, 1977.
G.A. Gescheider. Psychophysics: The Fundamentals. Lawrence Erlbaum Associates, Mahwah,
NJ., 1997.
J. L. Hall. Hybrid adaptive procedure for estimation of psychometric functions. Journal of
the Acoustical Society of America, 69(6):1763–1769, 1981.
L. O. Harvey, jr. Efficient estimation of sensory thresholds. Behavior Research Methods,
Instruments & Computers, 18(6):623–632, 1986.
N. A. Hastings and J. B. Peacock. Statistical Distributions. Butterworth & Co (Publishers)
Ltd, London, 1975.
http://opensoundcontrol.org/spec-1 0. http://opensoundcontrol.org/spec-1 0 (18/10/2008).
ITU. ITU-R Rec. BS.1116-1: Methods for the subjective Assessment of small Impairments
in Audio Systems including Multichannel Sound Systems. ITU, Geneva, 1997.
WhisPER v1.9.0 – User Documentation
90 Bibliography
ITU. ITU-R Rec. BS.1534-1: Method for the subjective assessment of intermediate quality
level of coding systems. ITU, Geneva, 2003.
G.A. Kelly. The Psychology of Personal Constructs. Norton, New York, 1955.
P. E. King-Smith, S. S. Grigsby, A. J. Vingrys, S. C. Benes, and A. Supowit. Efficient and
unbiased modifications of the quest threshold method: Theory, simulations, experimental
evaluation and practical implementation. Vision Research, 34(7):885–912, 1994.
S. A. Klein. Measuring, estimating and understanding the psychometric function: A com-
mentary. Perception & Psychophysics, 63(8):1421–1455, 2001.
M.R. Leek. Adaptive procedures in psychophysical research. Perception & Psychophysics, 63
(8):1279–1292, 2001.
L. Leventhal. Type i and type ii errors in the statistical analysis of listening tests. Journal
of the Audio Engineering Society, 34(6):437–453, 1986.
H. Levitt. Transformed up-down methods in psychoacoustics. Journal of the Acoustical
Society of America, 49(2):467–477, 1971.
H. Lieberman and A. Pentland. Microcomputer-based estimation of psychophysical thresh-
olds: The best pest. Behavioral Research Methods & Instrumentation, 14(1):21–25, 1982.
A. Lindau. Spatial Audio Quality Inventory. Test Manual. v1.2. http:// dx.doi.org/ 10.14279/
depositonce-1. TU Berlin, Berlin, 2015. http://dx.doi.org/10.14279/depositonce-1.
A. Lindau, V. Erbes, H.-J. Maempel, S. Lepa, F. Brinkmann, and S. Weinzierl. A spatial
audio quality inventory for virtual acoustic environments (saqi). In EAA Joint Symposium
on Auralization and Ambisonics. TU Berlin, 2014a.
A. Lindau, V. Erbes, H.-J. Maempel, S. Lepa, F. Brinkmann, and S. Weinzierl. Eine fokus-
gruppe f¨ur die entwicklung eines vokabulars zur sensorischen beurteilung virtueller akustis-
cher umgebungen. In Fortschritte der Akustik: Proc. of the 40th German Annual Confer-
ence on Acoustics, DAGA. DEGA, 2014b.
H. F. Martz and R. A. Waller. Bayesian reliability analysis. John Wiley & Sons, Ltd., New
York, 1982.
C.E. Osgood, G.J. Suci, and P.H. Tannenbaum. The measurement of meaning. University of
Illinois Press, Urbana, 1967. Neuauflage der Originalausgabe von 1957.
S. Otto. Vergleichende Simulation adaptiver, psychometrischer Verfahren zur Sch¨atzung von
Wahrnehmungsschwellen. Magisterarbeit, FG Audiokommunikation, Institut f¨ur Sprache
und Kommunikation, Fakult¨at I, TU Berlin, 2008.
A. Pentland. Maximum likelihood estimation: The best pest. Perception & Psychophysics,
28(4):377–379, 1980.
M. A. Pokorny. Using maximum-likelihood adaptive methods to estimate difference thresholds
and points of subjective equality. Behavior Research Methods, Instruments, & Computers,
30(4):624–636, 1998.
M. M. Taylor. On the efficiency of psychophysical measurement. Journal of the Acoustical
Society of America, 49(2):505–508, 1971.
WhisPER v1.9.0 – User Documentation
Bibliography 91
M.M. Taylor and C.D. Creelman. Pest: Efficient estimates on probability functions. Journal
of the Acoustical Society of America, 41(4):782–787, 1967.
B. Treutwein. Adaptive psychophysical procedures. Vision Research, 35(17):2503–2522, 1995.
B. Treutwein. Yaap: yet another adaptive procedure. Spatial Vision, 11(1):129–134, 1997.
O. B. Walter, A. Bacher, and Martin Fromm. A proposal for a common data exchange format
for repertory grid data. Journal of Constructivist Psychology, 17(3):247–251, 2004.
A. B. Watson and D.G. Pelli. Quest: A bayesian adaptive psychometric method. Perception
& Psychophysics, 33(2):113–120, 1983.
G. Wetherill. Sequential estimation of quantal response curves. Journal of the Royal Statistical
Society, B25:1–48, 1963.
A. Wlodarski. Erstellung einer Softwarebibliothek f¨ur H¨orversuche - Implementierung und
Strukturierung des Quellcodes. Magisterarbeit, FG Audiokommunikation, Institut f¨ur
Sprache und Kommunikation, Fakult¨at I, TU Berlin, 2008.
WhisPER v1.9.0 – User Documentation
92 List of Figures
List of Figures
4.1 The main window. ................................. 14
5.1 DW stimulus pool .................................. 18
5.2 DW edit stimulus .................................. 19
5.3 DW edit test sections. ............................... 22
6.1 GUI system for subject guidance – basic screen showing information on the
test run’s status of progression. . . . . . . . . . . . . . . . . . . . . . . . . . . 23
6.2 GUI system for subject guidance – basic screen showing the initial instructions. 24
7.1 The subject’s GUI when conducting an adaptive psychophysical threshold pro-
cedure......................................... 26
7.2 DW AP – main settings (...) ........................... 27
7.3 DW AP – track settings (...) ........................... 29
7.4 DW AP – edit staircase settings (...) ....................... 31
7.5 DW AP – edit PEST settings (...) ........................ 32
7.6 DW AP – edit ZEST settings (...) ........................ 34
7.7 The subject’s GUI for the performance of the RGT’s first part. . . . . . . . . 40
7.8 DW RGT – edit constructs (...) .......................... 42
7.9 The subject’s GUI for the performance of the RGT’s second part. . . . . . . . 43
7.10 DW RGT – main settings (...) .......................... 45
7.11 DW RGT – triad/element sequence setup (...) .................. 47
7.12 DW RGT – edit scale format (...) ........................ 48
7.13 The subject’s GUI for performing the SD test. . . . . . . . . . . . . . . . . . 49
7.14 DW SD – main settings (...) ........................... 50
7.15 DW SD – object pool (...) (see text on p. 49) . . . . . . . . . . . . . . . . . . 51
7.16 DW SD – edit scales (...) ............................. 52
7.17 DW SD – object sequence setup (...) ....................... 53
7.18 The subject’s GUI during an A/B/X trial. . . . . . . . . . . . . . . . . . . . . 54
7.19 ABX Trial Main Settings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
7.20 Assigning the stimuli to be judged in the current trial. . . . . . . . . . . . . . 57
7.21 Example of a description GUI presenting a name and circumscription of an
auditoryquality................................... 59
7.22 Example of a GUI for rating a quality on a bipolar, open-ended rating scale . 60
7.23 Example of a GUI for rating a quality using a unipolar, closed/open-ended
ratingscale ..................................... 60
7.24 Example of a GUI for rating a quality using direct bipolar, closed-ended rating 61
7.25 Example of a GUI for rating a quality using a dichotomous scale . . . . . . . 61
7.26 Final GUI asking for formulating and rating any so-far unmentioned differences
(SAQI qualifier: other)............................... 62
7.27 Example for a rating GUI with eight sliders which allows rating multiple stimuli
atatime ...................................... 62
WhisPER v1.9.0 – User Documentation
List of Figures 93
7.28 GUI for the rating the temporal variability of perceived differences . . . . . . 63
7.29 GUI for the rating the cause of perceived differences . . . . . . . . . . . . . . 63
7.30 GUI for assigning assessment entities to perceived differences . . . . . . . . . 64
7.31SAQIMainSettings ................................ 64
7.32 GUI for the selection of perceptual qualities. Left: singular selection, right:
category-based selection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
7.33 Default assessment entities . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
7.34 Stimulus assignment across conditions . . . . . . . . . . . . . . . . . . . . . . 68
7.35 Settings for distribution of conditions across rating screens. . . . . . . . . . . 69
7.36ABC/HRratingGUI................................ 70
7.37MUSHRAratingGUI ............................... 71
7.38 ABC/HR and MUSHRA setting . . . . . . . . . . . . . . . . . . . . . . . . . 72
8.1 DW Configure defaults’ presets .......................... 73
8.2 DW Configure defaults’ presets – tab ”Stimulus” . . . . . . . . . . . . . . . . 75
10.1 Schematic representations of the TSD content. . . . . . . . . . . . . . . . . . 84
10.2 Schematic representations of the TSD content. . . . . . . . . . . . . . . . . . 85
WhisPER v1.9.0 – User Documentation
