Speech Recogni ion in Complex Lis ening Si ua ions o
Bimodal and Single-Sided Dea Cochlea -Implan
Recipien s
Fabian Ebe ling1,2, Max Blüme 4, Ma k P ae o ius4,
Michael Schul e1,2, Jan Hee en1,2, Ki s en C. Wagene 1,2,
Thomas B and2,3
1Hö zen um Oldenbu g gGmbH, Oldenbu g
2Clus e o Excellence Hea ing4All, Oldenbu g
3Depa men o Medical Physics and Acous ics, Ca l on Ossie zky Uni e si y Oldenbu g,
Oldenbu g
4Depa men o O o hinola yngology, Uni e si y Medical Cen e Hambu g-Eppendo , Hambu g
Cu en A ilia ions:
Fabian Ebe ling:
Audiological Acous ics, Depa men o O ola yngology, Uni e si y Hospi al F ank u , Goe he
Uni e si y F ank u
Max Blüme :
Depa men o O o hinola yngology, Head and Neck Su ge y, Uni e si y Hospi al Essen,
Uni e si y Duisbu g-Essen
Abs ac
Cochlea implan (CI) e alua ions ypically use monosyllabic wo d ecogni ion (MWR) in quie ,
which does no e lec he complex lis ening en i onmen s CI use s encoun e , such as di use
backg ound noises, mul iple alke s, and di ided a en ion (DA) scena ios. This s udy in es iga ed
speech ecogni ion (SR) in CI ecipien s using he Oldenbu g Sen ence Tes (OLSA) and a mul i-
alke pa adigm, a ge ing ac o s beyond MWR. 8 CI use s wi h no mal hea ing in he
con ala e al ea (Single-Sided Dea ness, SSD) and 10 wi h con ala e al hea ing aid (bimodal,
BIM) pa icipa ed. MWR in quie was measu ed using he F eibu g Monosyllabic Tes
(con ala e al ea blocked). SR in noise was assessed using OLSA wi h di e en spa ial scena ios
and maske ypes. Spa ial elease om masking (SRM) was ex ac ed om speech ecep ion
h esholds (SRT). Selec i e a en ion (SA) bene i s we e e alua ed using he Concu en OLSA
(CCOLSA) bes aided in Single- and Dual-Task condi ions. Hea ing- ela ed cogni i e
pe o mance in di use noise was assessed using adap i e CCOLSA. MWR did no di e
signi ican ly be ween g oups. SSD pa ien s exhibi ed signi ican ly be e SRTs han BIM pa ien s.
Spa ial sepa a ion imp o ed SR o bo h g oups. SSD pa ien s pe o med be e in compe ing
speech han s a iona y noise; BIM pa ien s did no . Bo h g oups showed a be e -ea e ec . SSD
pa ien s demons a ed signi ican ly highe hea ing- ela ed cogni i e pe o mance and bene i ed
om SA. CCOLSA is easible o SSD and BIM pa ien s. Compa able MWR does no imply
simila SR in mo e ecologically alid si ua ions. The esul s sugges ha he hea ing loss in he
acous ic ea o BIM pa ien s limi s SA abili ies and inc eases cogni i e load, e en when bes
aided.
Keywo ds: Cochlea Implan ; Bimodal Hea ing; Single-Sided-Dea ness; Speech ecogni ion;
Selec i e A en ion
1
1 In oduc ion
Res o ing speech ecogni ion (SR) is he main objec i e when i comes o hea ing sys em
p o ision o people wi h impai ed hea ing. In Ge many, cochlea implan (CI) su ge y guidelines
as de ined in "Weißbuch Cochlea-Implan a (CI)-Ve so gung” (DGHNO-KHC, 2021) demand o
a SR es in quie o he indica ion and pos ope a i e de e mina ion o he p o ision success.
The e o e, monosyllabic wo d ecogni ion (MWR) is ypically e alua ed using he F eibu g
Monosyllabic Tes (F eibu ge Einsilbe es , FBE, Hahlb ock, 1953). The es is usually
conduc ed in quie wi h an occluded con ala e al ea . While i is an adequa e es o assess basic
SR abili ies, essen ial aspec s o SR in eal-li e si ua ions, such as di use backg ound noises,
con e sa ional u n- aking, a ying oices o di ided a en ion (DA) a e no aken in o accoun ,
bu o en highligh ed by pa ien s as demanding si ua ions (Wagene e al., 2008; Pang e al., 2019).
The e o e, when in es iga ing SR abili ies in e e yday li e, measu emen me hods ha a e mo e
ecologically alid han he FBE and ha ake in o accoun a b oade ange o he a o emen ioned
ac o s should be used.
Using a memo y es in which a ge wo ds had o be epea ed, McCoy e al. (2005) we e able o
show ha no mal-hea ing and hea ing-impai ed pa icipan s could achie e compa able esul s in
a es o speech in elligibili y. Howe e , people wi h hea ing loss had o pu in mo e e o as hey
epea ed signi ican ly ewe wo ds ha we e no pa o he a ge wo ds. O e all, hey we e able
o memo ize ewe wo ds. This aises he ques ion, whe he CI pa ien s could each compa able
esul s in SR asks bu show pe o mance di e ences in complex lis ening si ua ions. Hee en e
al. (2022) p esen ed he Concu en Oldenbu g Sen ence Tes (CCOLSA), which yields he
possibili y o measu ing a en ion abili ies and cogni i e abili ies in si ua ions wi h u n- aking
alke s. CCOLSA has al eady been success ully adminis e ed o pa ien s wi h HA p o ision, and
i s easibili y will now be assessed o pa ien s wi h CI p o ision. Fu he mo e, he di e ences in
SR abili ies be ween wo pa ien g oups ele an in Ge many will be explo ed.
CI candidacy c i e ia a y signi ican ly ac oss coun ies, pa icula ly wi h ega d o single-sided
dea ness (SSD). Al hough he equency o CI implan a ion o SSD pa ien s is inc easing, no all
coun ies ha e es ablished guidelines o hese pa ien s (Van De Heyning e al., 2022). Ge many,
howe e , has speci ic indica ions o SSD candida es (DGHNO-KHC /AWMF, 2020). No ably, a
subs an ial p opo ion o adul CI use s (74.5%) ha e unila e al CI p o ision, wi h 54% also using
a hea ing aid (HA) on he con ala e al side (Al akh i e al., 2024). The e o e, his s udy ocusses
on pa ien s wi h unila e al CI p o ision, encompassing hose wi h single-sided dea ness and hose
wi h bimodal hea ing (BIM). Bo h g oups combine elec ical and acous ical hea ing, bu wi h
di e en hea ing abili ies on he acous ically hea ing side. While SSD pa ien s a e no mally
hea ing on he non-implan ed side, BIM pa ien s ha e impai ed hea ing compensa ed by a HA. In
2008, Van de Heyning e al. conduc ed one o he pionee ing s udies on CI su ge y o pa ien s
2
wi h unila e al dea ness. In e es ingly, he p ima y aim was no necessa ily o enhance SR bu o
alle ia e inni us symp oms. None heless, alongside signi ican elie om inni us, a no iceable
imp o emen in SR was obse ed du ing he s udy (Ve mei e & Van De Heyning, 2009).
Since hen, a lo o s udies in es iga ed he e ec o unila e al CI p o ision on SR in pa ien s wi h
unila e al dea ness (e.g. Buechne e al., 2010; A nd e al., 2011; Ho h e al., 2016), because
un ea ed unila e al dea ness usually leads o di icul ies in SR, especially in noisy en i onmen s,
and nega i ely a ec s he localiza ion o sounds (e.g. Cab al Junio e al., 2016; Cañe e e al.,
2019; Snapp and Ausili, 2020). In hese cases, he p o ision o a CI can pa ially es o e binau al
hea ing, esul ing in enhanced speech ecogni ion, imp o ed sound localiza ion abili ies (e.g.
Zei le e al., 2015; A nd e al., 2017; Buss e al., 2018; Li o sky e al., 2019; Thompson e al.,
2022) as well as subjec i ely epo ed be e hea ing (e.g. Rösli e al., 2015; Dillon e al., 2017;
P ejban e al., 2018; Häußle e al., 2020). Despi e he obse ed imp o emen s in hea ing
ou comes ollowing cochlea implan a ion, he audi o y abili ies o SSD and BIM pa ien s emain
cons ained due o limi ed access o binau al cues, such as in e au al le el di e ences (ILDs) and
in e au al ime di e ences (ITDs). ITDs, which depend on p ecise empo al encoding, a e poo ly
ep esen ed in CI s imula ion s a egies, pa icula ly a he high pulse a es commonly employed
in speech coding s a egies (Laback e al., 2004, 2015). In con as , ILDs, based on ampli ude
di e ences, a e mo e eliably encoded and mo e easily pe cei ed by CI use s (G an ham e al.,
2007; Seebe & Fas l, 2008). Consequen ly, ILDs a e p ocessed mo e e ec i ely han ITDs due
o inhe en limi a ions in CI sys ems, which can only pa ially esol e he empo al ine s uc u e
o inpu signals while p edominan ly p ese ing en elope cues (Van Hoesel, 2004; Ausili e al.,
2020). Fo BIM pa ien s he access becomes e en mo e di icul due o he signal p ocessing o
he HA (Udesen e al., 2013), bu he access o ILDs and ITDs is ele an o SR in noise as well
as he localiza ion and sepa a ion o sound sou ces (A an e al., 2015). E en hough bo h g oups
bene i om binau al hea ing, hey also end o Be e -Ea -Lis ening (BE), which means ha he
subjec ’s lis ening pe o mance is p edominan ly de e mined by he be e hea ing ea a he han
by he imp ession o bo h ea s oge he , especially in noisy en i onmen s wi h sepa a ed a ge
and noise sou ces (Williges e al., 2019). Howe e , he measu emen condi ions used by Williges
e al. (2019), like s a iona y maske s p esen ed om ixed posi ions, a e no necessa ily
ep esen a i e o si ua ions aced in e e yday li e. In e e yday li e, we mus deal wi h complex
lis ening si ua ions, consis ing no only o s a iona y noise sou ces, bu also con aining compe ing
alke s, luc ua ing backg ound sounds and music (Wagene e al., 2008). Besides di e en signal-
o-noise a ios (SNR), ac o s like a en ion and cogni i e load play a c ucial ole in hese so-
called cock ail-pa y si ua ions (B onkho s , 2000). A e checking he easibili y o he CCOLSA
in CI pa ien s, his measu emen me hod will be used o in es iga e whe he and how di e ences
in SR be ween SSD and BIM pa ien s can be obse ed o complex lis ening si ua ions. The
in es iga ions also ocus on he c ucial ac o s leading o possible pe o mance di e ences
3
(masking h ough s a iona y noises o in e e ing speech as well as a en ion abili ies). S ickney
e al. (2004) ound, h ough compa ison o SR o no mally hea ing pa icipan s and bila e al CI
use s, ha one in e e ing alke had a la ge in luence on bila e al CI use s han on no mally
hea ing pa icipan s. Bo h subjec g oups showed wo se esul s o one in e e ing alke han o
speech-shaped noise. When changing he in e e ing alke ’s oice om he same male as he
a ge alke o ano he male o emale, he no mally hea ing pa icipan s showed an imp o emen
in speech in elligibili y while he di e en oices had no in luence on he pe o mance o he
bila e al CI pa icipan s. Koelewijn e al. (2014) ound compa able esul s o pa icipan s wi h
hea ing impai men . Using one emale a ge alke and one male in e e ing alke , he
pa icipan s showed wo se speech in elligibili y han o speech-shaped noise.
The aim o he p esen s udy was he compa ison o he SR pe o mance o SSD and BIM pa ien s,
who show compa able MWR on he CI side, in complex lis ening si ua ions measu ed using
CCOLSA. Fu he mo e, he alidi y o he ou come o con en ional measu emen p ocedu es
used daily in clinical con ex , like he FBE o he Oldenbu g Sen ence Tes (Oldenbu ge Sa z es ,
OLSA), will be analyzed.
4
2 Me hods
2.1 Pa icipan s
Eigh een pa icipan s (nine o each sex) wi h a mean age o 60.2 yea s we e assessed. The
younges pa icipan was 24 yea s old and he oldes 81 yea s. No limi a ions ega ding age,
gende , du a ion o CI o HA p o ision, implan , speech p ocesso o HA we e made. Each
pa icipan had o ha e a minimum o 60% MWR in quie while lis ening only ia CI (con ala e al
ea blocked), on he las annual check-up appoin men . Fo he sepa a ion in o SSD- and BIM-
pa icipan s, he pu e- one-a e age (PTA; equencies 500, 1k, 2k, 4k Hz) o he acous ically
hea ing side was added as c i e ion o he exis ing HA p o ision. Also, each pa icipan had o be
able o pe o m speech ecogni ion es s in noise. A PTA < 30 dB HL was se as limi o no mal
acous ical hea ing in he con ala e al ea . All pa icipan s wi h no mal acous ical hea ing we e
assigned o he SSD-g oup. Pa icipan s wi h a PTA 30 dB HL and exis ing HA p o ision in
he con ala e al ea we e assigned o he BIM-g oup. Con ala e al hea ing h esholds o all
pa icipan s a e depic ed in Figu e 1, including he a i hme ic mean o he wo pa icipan g oups.
Figu e 1: The igu e shows he hea ing h esholds o he acous ical hea ing side o he SSD
(dashed) and BIM (do ed) g oups. Also, he a i hme ic mean o he SSD (g een) and
BIM (blue) g oup is shown.
One o he SSD-pa icipan s showed a p e iously unknown hea ing loss, which led o a PTA
o 33.3 dB HL. Since i is only a mild hea ing loss, he pa icipan pa icipa ed in he s udy as pa
o he SSD-g oup. Hence, he highes PTA in he SSD-g oup was 33.3 dB HL and he lowes PTA
o he BIM-g oup was 35.8 dB HL. In he end, 8 pa icipan s we e assigned o he SSD-g oup
and 10 pa icipan s o he BIM-g oup. The exac demog aphic da a as well as he p o ision
5
in o ma ion a e lis ed in Table 1. Each pa icipan was ec ui ed h ough he uni e si y medical
cen e Hambu g-Eppendo . Pa icipa ion was on a olun a y basis and was no emune a ed. The
expe imen was app o ed by he e hics commi ee o he Uni e si y o Hambu g.
Table 1: Demog aphic da a and e iology o hea ing loss o each pa icipan
ID Age E iology Sex
Dea ness
(yea s) CI ea Implan
CI
P ocesso CI Usage
(yea s) G oup
PTA
(dB HL)
VP 01 69 Sudden hea ing loss F ca. 20 yea s L Mi1250 Rondo 3 1,1 BiM 56,7
VP 02 81 Ch onic o i is media M min. 10 yea s R CI622 CP1150 1,6 BiM 64,2
VP 03 24 unknown F N/S R CI512 CP1000 12,8 BiM 66,7
VP 04 27 Ch onic o i is media M 3-4 yea s R CI622 CP1150 0,7 SSD 4,2
VP 05 56 Sudden hea ing loss F N/S L Mi1250 Sonne 2 0,9 BiM 54,2
VP 06 64 Sudden hea ing loss F N/S R Mi1250 Sonne 2 1,7 SSD 22,9
VP 07 32 Sudden hea ing loss F N/S L CI622 CP1150 0,5 SSD 17,9
VP 08 77 O oscle osis F N/S R Mi1250 Rondo 3 0,9 BiM 65,4
VP 09 43 unknown M 1,5 yea s L Mi1250 Sonne 2 0,6 BiM 80,4
VP 10 60 Mo bus Meniè e F N/S L Mi1250 Rondo 2 5,5 SSD 32,9
VP 11 70 Re ision s apes su ge y F N/S R CI622 CP1150 0,9 BiM 35,4
VP 12 80 Ves ibula schwannoma M N/S R Mi1250 Rondo 3 1 SSD 25,4
VP 13 62 Sudden hea ing loss M N/S L CI622 CP1000 3,9 SSD 0,0
VP 14 65 Mo bus Meniè e M N/S R Mi1250 Sonne 2 1,9 SSD 6,6
VP 15 80 P esbyacusis M N/S L Mi1250 Rondo 3 1,2 BiM 44,2
VP 16 81 unknown M N/S L CI622 CP1150 0,5 BiM 70,4
VP 17 59 Sudden hea ing loss F N/S L Mi1250 Rondo 2 8,8 SSD 1,3
VP 18 53 Sudden hea ing loss M N/S L CI622 CP1000 2,2 BiM 57,9
F = emale; M = male; N/S = no speci ied; CI = Cochlea Implan ; L = le ; R = igh ; SSD = Single-Sided-
Dea ness; BIM = bimodal; PTA = pu e one a e age
2.2 Measu emen Se up
All measu emen s we e conduc ed in he audiology labo a o y o he uni e si y medical cen e
Hambu g-Eppendo . The es ing oom (5 m x 4 m x 3,5 m) was op imized o acous ical
measu emen s, using sound abso be s on he ceiling, mol on cu ains on he walls and ca pe
loo ing. The pa icipan s we e sea ed on a chai in he middle o he oom. In 1 m dis ance, 16
Genelec 8030B (GENELEC, Finland) loudspeake s we e placed in equidis an dis ances a ound
he pa icipan s posi ion, s a ing in on o he pa icipan . The loudspeake ou pu was con olled
using TASCAR-scenes (G imm e al., 2015) un on a Dell (Dell Technologies Inc., USA) desk op
compu e wi h an Ubun u ope a ing sys em. Selec ion as well as ac i a ion/deac i a ion o he
TASCAR-scenes was con olled using OSC-commands in Ma lab 2023b (The Ma hWo ks Inc.,
USA) un on a Dell (Dell Technologies Inc., USA) desk op compu e wi h Windows 10 (Mic oso
Co po a ion, USA) as ope a ing sys em, which had a di ec cable connec ion o he Ubun u
compu e . Using an RME ADI-8 con e e (Audio AG, Ge many), he windows compu e was
connec ed o he loudspeake s. Fo he loudspeake calib a ion, a 2260 sound analyze (B üel &
Kjæ , Denma k) was used, calib a ed wi h a ype 4231 calib a o (B üel & Kjæ , Denma k) a 94
dB SPL. All measu emen s we e pe o med using a linea scale o he le els. The sound analyze
was placed a he pa icipan s posi ion a he heigh o he loudspeake s. A i s , he loudspeake
le els we e equalized o 75 dB SPL wi h maximum de ia ions o 0,5 dB, using whi e noise as
calib a ion signal. Subsequen ly he ou pu le el o all signals has been calib a ed.
Pu e one audiome y was conduc ed using a cus om MATLAB sc ip in MATLAB 2023b (The
Ma hWo ks Inc., USA), which p esen ed pu e ones h ough model DT770 (Beye dynamic,
6
Ge many) headphones. The measu ed equencies anged om 125 Hz o 8 kHz. The calib a ion
was done using a 2610 measu emen ampli ie (B üel & Kjæ , Denma k) and an a i icial ea ype
4153 wi h couple (B üel & Kjæ , Denma k). Sine ones we e p esen ed a 60 dB SPL as
calib a ion signal. Fo he de e mina ion o dB HL alues, i was assumed ha co ec h esholds
could be app oxima ed by aking in o accoun he di e ences in he equency esponse be ween
he DT770 and he HDA200. The e o e, he RETSPL alues o he HDA200 we e used, wi h an
addi ional co ec ion applied o compensa e o he equency esponse di e ences be ween he
HDA200 and he DT770 a each audiome ic equency measu ed.
2.6 Gene al Measu emen P ocedu e
All measu emen s and espec i e condi ions a e lis ed in Table 2. Be o e s a ing any
measu emen , he ea canal o he acous ical hea ing ea was checked using o oscopy, o ensu e
op imal measu ing condi ions. A e wa ds, pu e one audiome y o he acous ical hea ing ea
Table 2: Measu emen Condi ions
Measu emen Condi ions No e
Pu e one audiome y Headphone measu emen Hea ing h eshold, PTA
FBE S0N0 Monosyllabic wo d ecogni ion
OLSA (Ca e e ia Noise) S0Ndi SRT in di use ca e e ia noise
OLSA (olnoise) S0N0/S0Nac/S0NCI SRT wi h s a iona y maske
OLSA (FIT) S0N0/S0Nac/S0NCI SRT wi h emale in e e ing alke
CCOLSA Adap i e O e lap ime a 50% co ec , CCO
CCOLSA DT Wo d sco ing wi h DA
CCOLSA ST Wo d sco ing wi h SA
PTA = Pu e Tone A e age; FBE = F eibu g Monosyllabic Tes ; SRT = Speech Recep ion Th eshold; FIT = Female
In e e ing Talke ; CCO = CCOLSAcos s; DA = Di ided A en ion; SA = Selec i e A en ion; DA = Dual Task;
ST = Single Task
and he FBE wi h CI only (acous ical hea ing ea blocked) we e conduc ed o de e mine he
pa ame e s used o g oup alloca ion. As a nex s ep, he indi idual SRTs o all h ee CCOLSA
alke s (see Sec ion 2.5) we e de e mined in di use ca e e ia noise. A e wa ds he adap i e
CCOLSA (see Sec ion 2.5.3) was measu ed o de e mine he indi idual o e lap ime. The speech
ecep ion h esholds (SRTs) as well as he o e lap ime a e needed o he emaining CCOLSA
condi ions. All emaining measu emen s (OLSA wi h olnoise/FIT, CCOLSA ST/DT) we e
conduc ed in andomized o de . Pa icipan s o he SSD g oup o e all had o pe o m wo mo e
measu emen s han he BIM g oup: The CCOLSA ST and DT condi ion we e conduc ed unaided
(SSD) and wi h HA p o ision (SSDaided) o he acous ical no mal hea ing ea . The HA p o ision
enables he in es iga ion o he in luence o a HAs signal p ocessing on he combina ion o elec ical
and acous ical hea ing. A andom HA was selec ed o he SSDaided condi ion and was i ed o he
indi idual hea ing h eshold using he p esc ip ion i ing o mula p o ided by he manu ac u e
and using closed domes as acous ic coupling. The choice we e he HAs Pu e Cha ge&Go 5 Nx
7
(Signia, Ge many), Opn S1 miniRITE (O icon, Denma k) and Audéo M70-R (Phonak,
Swi ze land). Fea u es like e.g. au oma ic scene de ec ion o noise educ ion we e deac i a ed i
possible o se o he lowes le el o he wise. Mic ophones we e se o omnidi ec ional. The gain
alues based on he eedback supp ession we e disca ded. Nei he he HAs o he BIM g oup no
he CIs o bo h g oups we e i ed o ma ch ce ain e e ences. Ins ead, he mos p e e ed se ings
o he pa icipan s we e used.
2.3 F eibu g Monosyllabic Tes
Monau al MWR a 65 dB SPL in quie , wi h CI only (con ala e al ea occluded), was assessed
using he FBE in ee ield. The measu emen se up is shown in Figu e 2(a). Speech was p esen ed
om he on and a combina ion o ea plugs and ea mu s was used o exclude he acous ical
hea ing ea in he con ala e al side. The pa icipan ’s ask was o epea he p esen ed wo ds. Fo
cases in which he pa icipan couldn’ unde s and he p esen ed wo ds co ec ly, guessing was
allowed. Fo each pa icipan one es lis , including 20 Ge man monosyllabic wo ds, was
measu ed. Following he ecommenda ion o Winkle and Holube (2014), only he lis s 6, 7 and
16 we e measu ed in pseudo andomized o de . Measu emen s we e con olled using an
applica ion un on a Dell (Dell Technologies Inc., USA) desk op compu e , designed using he
Ma lab 2023b (The Ma hWo ks Inc., USA) app designe .
(a) F eibu g Monosyllabic Tes (b) OLSA S0N0, S0N90ac, S0N90CI
(c) OLSA di use (d) CCOLSA
Figu e 2: The igu e schema ically illus a es he speake a angemen . The speake s ac i e du ing
he espec i e measu emen condi ion a e highligh ed in ed. a) F eibu g monosyllabic
wo d es . b) S0N0, S0N90ac and S0N90CI condi ions o OLSA. c) OLSA in di use
backg ound noise, d) CCOLSA measu emen : due o he di use backg ound noise, all
speake s a e ac i e. Addi ionally, he posi ions o he wo male alke s and he emale
alke a e indica ed.
14
Figu e 6: SRM o he SSD (g een) and BIM (blue) g oup, wi h maske s olnoise (le ) and FIT
( igh ), depending on he noise sou ce posi ion (N90ac, N90CI). The boxes ex end o he
in e qua ile ange, wi h he median ep esen ed by a ho izon al line. The whiske s
ex end o a maximum o 1.5 imes he leng h o he box and ou lie s a e ma ked wi h
’x’. S a is ical signi icance is indica ed by as e isks (*p<.05).
3.5 CCOLSA Wo d Sco ing
The wo d sco ing esul s om he CCOLSA ST and DT measu emen s a e shown in Figu e 7.
The di e ences in median wo d sco ing esul s be ween he ST and DT condi ions we e in he
SSD g oup 30.0%, in he SSDaided condi ion 37.5% and only 6.7% in he BIM g oup. The impac
o HA p o ision is e iden in he educed wo d sco ing ou comes o SSD compa ed o SSDaided.
This was obse ed in bo h he ST (median di e ence: 4.2%) and he DT (median di e ence:
11.7%). Fo each da ase , no mal dis ibu ion could be assumed (SW-Tes , p 0.051). Since he
esul s o SSD and SSDaided a e pai ed samples, bu no pai ed wi h he BIM esul s, wo sepa a e
ANOVAs we e conduc ed o a oid mixing pai ed and unpai ed samples wi hin a single ANOVA.
To in es iga e he di e ences be ween he SSD and BIM g oup, a wo-way mixed ANOVA (Type
III) wi h be ween-subjec ac o g oup (SSD, BIM) and wi hin-subjec ac o ask (ST, DT) was
pe o med. No signi ican main e ec s o ei he g oup o ask we e ound, bu a signi ican in e ac ion
be ween he ac o s g oup and ask was ound (F(1, 16) = 11.07, p = 0.004), indica ing a di e en
dependence o he wo d sco ing wi hin each g oup on he ask. To u he examine his in e ac ion
wo one-way ANOVAs we e conduc ed o in es iga e he in luence o bo h ac o s sepa a ely. The
e ec o ask wi hin each g oup appea ed o be signi ican o he SSD g oup (F(1, 7) = 64.3, p <
0.001) bu no o he BIM g oup (F(1, 9) = 1.05, p = 0.332). Fo he ac o g oup, a signi ican e ec
could only be obse ed in he ST (F(1, 16) = 8.4, p = 0.01) bu no in he DT condi ion (F(1, 16) =
0.767, p = 0.394).
15
To assess he in luence o he HA p o ision in he SSD g oup, a 2-way epea ed measu es ANOVA
(Type III) wi h wi hin-subjec ac o s g oup (SSD, SSDaided) and ask (ST, DT) was conduc ed.
While he in e ac ion (F(1, 7) = 0.074, p = 0.794) and he main e ec g oup (F(1, 7) = 1.529, p =
0.256) emained non-signi ican , he main e ec ask (F(1, 7) = 58.443, p < 0.001) appea ed o be
signi ican .
Figu e 7: Amoun o co ec ly epea ed a ge wo ds (in %) o he CCOLSA ST and DT condi ion o
he SSD, BIM and SSDaided g oups. The boxes ex end o he in e qua ile ange, wi h he
median ep esen ed by a ho izon al line. The whiske s ex end o a maximum o 1.5 imes
he leng h o he box. S a is ical signi icance is indica ed by as e isks (*p<.05,
**p<.01, ***p<.001).
3.4 CCOLSAcos s
The dis ibu ions o he CCO a e shown in Fehle ! Ve weisquelle konn e nich ge unden
we den.8. The median CCO o he SSD g oup was 4.6 dB and 5.9 dB in he BIM g oup. Fo he
SSD da a, a de ia ion om he no mal dis ibu ion was e ealed (SW-Tes , p = 0.045), while
no mal dis ibu ion could be assumed o he BIM g oup (SW-Tes , p = 0.2). The pai ed
compa ison e ealed a s a is ically signi ican di e ence in CCO be ween he SSD and BIM
g oups (U = 11, Z = 2.579, p = 0.01).
16
Figu e 8: CCO o he SSD (g een) and BIM (blue) g oup. The boxes ex end o he in e qua ile
ange, wi h he median ep esen ed by a ho izon al line. The whiske s ex end o a
maximum o 1.5 imes he leng h o he box and ou lie s a e ma ked wi h ’x’. S a is ical
signi icance is indica ed by as e isks (*p<.05).
17
4 Discussion
This s udy in es iga ed he speech ecogni ion abili ies o SSD and BIM pa ien s in di e en
si ua ions. The e o e, SRTs we e measu ed using di e en maske s and di e en spa ial scena ios
om which SRM alues we e ex ac ed. Addi ionally, he speech ecogni ion o complex
si ua ions was assessed using a mul i- alke es which has ne e been applied o CI use s be o e
and which also cap u es he in luence o di e en a en ion si ua ions. Fu he mo e, he hea ing-
ela ed cogni i e pe o mance in e ms o CCO was assessed.
4.1 Speech Recogni ion
Fo he OLSA e e ence condi ion, S0N0 wi h olnoise, he SRTs we e −8.5 dB SNR o he SSD
g oup. This co esponds o SRTs o no mal hea ing lis ene s (Wagene e al., 1999b)
(−7.1 dB SNR) and was also expec ed acco ding o Williges e al. (2019) (app ox. −6.0 dB SNR)
o he SSD g oup. SRTs o −4.5 dB SNR we e measu ed o he BIM g oup, which is 4 dB highe
(poo e speech ecogni ion) han he SSD g oup’s SRTs and is also in line wi h he indings o
Williges e al. (2019) (app ox. −4.0 dB SNR). This indica es ha o SR wi hou spa ially
sepa a ed sound sou ces, be e ea (BE) pe o mance is decisi e. The BE can be de e mined
h ough compa ison o he S0N90 condi ions wi hin a g oup. Fo he SSD g oup, he acous ical
hea ing ea , and o he BIM g oup he elec ical hea ing ea appea ed o be he BE.
In scena ios wi h spa ially sepa a ed sound sou ces SRTs o −10.4 dB SNR (S0N90ac) and
−12.5 dB SNR (S0N90CI) we e measu ed o he SSD g oup in he olnoise condi ion. This co esponds
o he expec a ion acco ding o Williges e al. (2019), which s a ed ha in he SSD g oup, be e SR
was o be expec ed wi h he maske on he CI side. The measu ed esul s a e compa able o hose
de e mined by Williges e al. (2019), who measu ed SRTs o app ox. −10.0 dB SNR in he S0N90CI
condi ion. Howe e , he SSD g oup in his s udy showed 6.0 dB lowe SRTs (be e speech ecogni ion)
o he S0N90ac condi ion han he SSD g oup o Williges e al. (2019) (app ox. −4.5 dB SNR). When
compa ing SRTs o condi ions wi h spa ially sepa a ed sound sou ces, subs an ial di e ences can a ise
due o a ia ions in loudspeake se ups and oom acous ics, as illus a ed by he e e ence cu es epo ed
by Winkle e al. (2025). No mal hea ing lis ene s can each SRTs o app ox. −17.0 dB SNR in hese
scena ios and he e o e dis inc ly be e SR (Beu elmann & B and, 2006).
Wi h SRTs o −10.7 dB SNR (S0N90ac), he BIM g oup eached compa able esul s o he SSD
g oup, bu no imp o emen was obse ed when p esen ing he noise on he CI side. P esen ing
he noise om he CI side leads o a deg ada ion o speech ecogni ion and SRTs o −7.0 dB SNR
(S0N90CI). This is also in line wi h he indings o Williges e al. (2019), which s a ed ha he
elec ical hea ing ea is BE o he BIM g oup.
Using he FIT as maske in a S0N0 scena io, SRTs o −16.1 dB SNR we e obse ed in he SSD
g oup, which was 10.2 dB lowe (be e speech ecogni ion) han he BIM g oups SRTs o
−5.9 dB SNR. G ossmann e al. (2016) measu ed SRTs o −1.6 dB SNR o SSD pa ien s using
18
he OLSA and a a ying male wo- alke babble noise in a S0N0 scena io. Hence, he SSD g oup
in his s udy exceeded he expec a ions demons a ing lowe SRTs (be e speech ecogni ion) han
he SSD g oup in G ossmann e al. (2016), which could be explained due o he di e ence be ween
he used maske s. The BIM g oups SRTs o he same condi ion exceeded he expec a ions
acco ding o Culling on and Zeng (2011), who measu ed SRTs o app ox. 1 dB SNR, using
sen ences o he HINT da abase wi h one emale in e e ing alke om he Ins i u e o Elec ical
and Elec onics Enginee s (IEEE) sen ence ma e ial.
Fo he S0N90CI condi ion SRTs o −18.2 dB SNR o he SSD g oup and −6.0 dB SNR o he BIM
g oup we e obse ed. The SSD g oup again demons a ed clea ly lowe SRTs (be e speech
ecogni ion) han he BIM g oup, wi h a di e ence o 12.2 dB and again exceeded he expec a ion
acco ding o G ossmann e al. (2016), who obse ed SRTs o −8.8 dB SNR o he S0N90CI
condi ion. The BIM pa icipan s o Gi o d e al. (2014) achie ed SNR50 sco es o 2.0 dB SNR o
he same condi ion, e ealing a di e ence o app ox. 4 dB o he S0N0 condi ion, which is di e en
om he BIM g oup in his s udy who eached compa able SRTs o he S0N0 condi ion.
A simila end was obse ed o he S0N90ac condi ion, whe e SRTs o −17.1 dB SNR we e
obse ed o he SSD g oup, which is 8.1 dB lowe (be e speech ecogni ion) han he
−9.0 dB SNR o he BIM g oup. G ossmann e al. (2016) obse ed SRTs o −3.1 dB SNR (wo se
speech ecogni ion) o hei SSD g oup, hus he SSD g oup in his s udy again exceeded he
expec a ions. In con as o his he BIM g oup o Gi o d e al. (2014) demons a ed simila
esul s like he BIM g oup o his s udy, showing SNR50 sco es o app ox. 8.0 dB SNR.
In he di use ca e e ia condi ion (S0Ndi ), he SSD g oup demons a ed lowe SRTs o 0.2 dB SNR
(be e speech ecogni ion) compa ed o he BIM g oup wi h 3.6 dB SNR. No only is he e a
signi ican di e ence be ween he wo g oups, bu bo h g oups also exhibi highe SRTs (poo e speech
ecogni ion) han in he s a iona y noise condi ions.
To analyze he bene i o spa ial sepa a ion o sound sou ces, he SRM was ex ac ed om he
indi idual SRTs o each pa icipan . The highes SRM obse ed in his s udy was 3.9 dB wi h
olnoise o he SSD g oup and 6.0 dB o he BIM g oup. The SRM o he SSD g oup aligns wi h
he esul s o Williges e al. (2019), whe e he SSD g oup showed an SRM o 3.4 dB. Howe e ,
in hei s udy, he BIM g oup demons a ed a lowe SRM o only 2.3 dB, which con as s wi h he
highe SRM obse ed o he BIM g oup in his s udy. The SRM epo ed by Williges e al. (2019)
was signi ican ly di e en be ween SSD and BIM lis ene s. Since no signi ican di e ences in
SRM we e ound be ween g oups in his s udy, hese esul s will no be discussed u he .
4.2 Maske influence
S a is ical analysis e ealed a signi ican di e ence be ween all maske s o he SSD g oup, bu
no o he BIM g oup. The e was no signi ican di e ence be ween he maske s olnoise and FIT
in he BIM g oup. The olnoise p ima ily causes ene ge ic masking (EM), which in e e es wi h
speech ecogni ion by o e lapping he a ge speech signal. In con as , he FIT induces
19
in o ma ional masking (IM), whe e he compe ing alke ’s speech con en dis ac s he lis ene
and makes i ha de o ocus on he a ge speech. In o de o sepa a e wo compe ing alke s, i is
c ucial o de ec and di e en ia e be ween he alke s’ undamen al equencies, like
di e en ia ing be ween he male OLSA a ge alke and he emale in e e ing alke (FIT
maske ). Indi iduals wi h hea ing loss o en ha e a poo e equency selec i i y compa ed o
no mal hea ing lis ene s, while CIs s uggle wi h he p esen a ion o ha monic signal componen s
(Oxenham, 2008). Bo h ac o s - impai ed equency selec i i y and p oblema ic ha monics
p esen a ion - a e ound in he BIM g oup and could explain why none o bo h maske s used in
his s udy showed a signi ican in luence on he SR o he BIM g oup. Pyschny e al. (2011)
in es iga ed he in luence o IM on he SR o BIM pa ien s bu used a di e en p ocedu e. They
used OLSA sen ences as a ge speech, wi h a ixed le el and SNR, o de e mine he co ec SR
in pe cen age. As maske s hey used OLSA sen ences, bu wi h inc eased undamen al equency.
The measu emen s we e conduc ed in a S0N0 scena io o he condi ions HA only, CI only and
bimodal (HA + CI). No imp o emen o he sepa a ion by di e en undamen al equencies o
a ge and in e e ing alke was ound h ough he bimodal lis ening mode compa ed o CI o HA
only. Mülle and Lang-Ro h (2021) used a simila p ocedu e and s imuli like Pyschny e al. (2011)
o in es iga e he in luence o IM on SSD pa ien s and addi ionally es ed he in luence o di e en
spa ial scena ios. Fo he S0N0 condi ion, no signi ican imp o emen was obse ed compa ing
he bila e al lis ening mode o he no mal hea ing ea alone, indica ing ha sepa a ion o di e en
alke s p ima ily depends on he pe o mance o he no mal hea ing ea . Vongphoe and Zeng
(2005) compa ed he alke ecogni ion based on empo al signal in o ma ion be ween acous ical
and elec ical hea ing and ound wo se pe o mance o CI use s. This also suppo s he
assump ion ha , since he CI ea is he BE o he BIM g oup, his ype o maske has only a
mino in luence. The esul s a e in line wi h he indings o his s udy, demons a ing a signi ican
di e ence be ween olnoise and FIT only o he SSD g oup and no o he BIM g oup.
4.3 A en ion eec s and cogni i e pe o mance
An impo an pa o his s udy was o in es iga e whe he CCOLSA could be measu ed wi h CI
pa ien s. I was shown ha he measu emen s we e easible o bo h SSD and BIM pa ien s. The
pa icipan s’ age anged om 24 o 81 yea s; hus, i can be assumed ha i can also be conduc ed
wi h o he CI pa ien s in he same age ange as long as he measu emen o o he speech in noise
es s is easible o hem. The SR in complex lis ening si ua ions was signi ican ly di e en
be ween bo h g oups. Fu he mo e, only he SSD g oup had an ad an age in he SA ask, which
could no be obse ed in he BIM g oup. A HA p o ision on he con ala e al ea had no in luence
on he SR pe o mance o he SA bene i o he SSD g oup. Signi ican ly highe hea ing- ela ed
cogni i e pe o mance in e ms o CCO was measu ed o he SSD g oup compa ed o he BIM
g oup.
20
All g oups eached a wo d sco ing close o 50% co ec (SSD: 54.1%, BIM: 52.5%, SSDaided:
42.5%) o he DT condi ion, which was expec ed due o he usage o he indi idual o e lap imes
de e mined in he adap i e condi ion. Hence, each pa icipan g oup demons a ed plausible
esul s o he CCOLSA. The wo d sco ing esul s o he ST condi ion (SSD: 84.1%, BIM: 59.1%,
SSDaided: 80.0%) a e in line wi h he e e ence condi ion (S0N0) o he OLSA wi h olnoise (SSD:
-8.5 dB, BIM: -4.5 dB), showing be e SR o he SSD han o he BIM g oup. Th ough
compa ison o he ST and DT esul s, a selec i e a en ion bene i was obse ed o he SSD and
SSDaided g oup, which could no be obse ed o he BIM g oup. This di e ence could be
explained by he di e ence o he acous ical hea ing side. Acco ding o Shinn-Cunningham and
Bes (2008), he e is a decisi e di e ence be ween he bene i h ough SA in no mal hea ing
lis ene s and he unc ionali y o HA ea u es. Taking in o accoun ha he HA p o ision o he
acous ical hea ing ea in he SSDaided g oup showed no signi ican in luence on he pe o mance
o he SSD pa icipan s, he di e ence be ween he g oups is p obably no based on he addi ional
signal p ocessing o he HAs. Ra he , i can be assumed ha i is de e mined by he hea ing loss.
The CCO alues we e calcula ed based on he indi idual o e lap ime de e mined in he adap i e
CCOLSA and p o ide in o ma ion abou he hea ing ela ed cogni i e pe o mance o he
pa icipan s. Wi h CCO alues o 4.6 dB he SSD g oup eached signi ican ly lowe (be e
hea ing ela ed cogni i e pe o mance) CCO han he BIM g oup wi h 5.9 dB. This s udy was he
i s ime CCO we e measu ed in any CI pa ien g oup, so he e is a lack o e e ence da a using
he same measu e. When in oducing he adap i e CCOLSA, Hee en e al. (2023) p o ided CCO
sco es o no mal hea ing lis ene s, hea ing impai ed lis ene s wi hou HA p o ision (unaided
non-HA use s) and unaided HA use s. Pos hoc, only he di e ence be ween no mally hea ing
lis ene s and HA use s was ound o be signi ican by Hee en e al. (2023). The no mally hea ing
lis ene s achie ed CCO o app ox. 5.0 dB, a e which he SSD g oup in his s udy appea ed o
possess a sligh ly be e pe o mance han he no mal hea ing lis ene s es ed by Hee en e al.
(2023). Wi h CCO o app ox. 6.1 dB o he unaided HA use s and app ox. 5.8 dB o he unaided
non-HA use s g oup, bo h demons a ed compa able CCO o he BIM g oup in his s udy. Since
he pe o mance on he CI side was ma ched be ween bo h g oups, he di e ence in hea ing
ela ed cogni i e pe o mance appea s o be de e mined by he acous ical hea ing side. Hea ing-
impai ed lis ene s can pe o m simila ly o no mal hea ing lis ene s in asks including SR, bu
hey ha e o dedica e mo e cogni i e esou ces on he ask (McCoy e al., 2005). The esul s o
his s udy show ha e en wi h he combina ion o an elec ically hea ing and an acous ically
no mally hea ing side, he pe o mance o SSD pa ien s desc ibed by he CCO is simila o ha
o no mal hea ing lis ene s. This con as s wi h he inding ha he pe o mance o BIM subjec s
is simila o ha o unaided non-HA use s and unaided HA use s.
21
Since SSD pa ien s gain an ad an age h ough SA and can each CCO compa able o no mally
hea ing lis ene s esul s, SRM appea s o be he only di e ence be ween SSD pa ien s and no mal
hea ing lis ene s. SRM is measu ed mos commonly in scena ios using s a iona y noise sou ces.
Hence, he ques ion a ises whe he no mal hea ing lis ene s show an ad an age in complex
lis ening si ua ions compa ed o SSD pa ien s. Puglisi e al. (2021) in es iga ed he in luence o
e e be a ion ime and di e en noises on he SR o no mal hea ing lis ene s in eal, complex
acous ical scena ios. One o he key indings was ha e en o hese si ua ions, SRM o up o 3
dB could be obse ed. I is no iceable ha signi ican SRM o EM only appea ed o low
e e be a ion imes and o IM only o la ge e e be a ion imes. The e o e, i can be concluded
ha SRM also plays a ole in complex lis ening si ua ions. I can be assumed ha no mal hea ing
lis ene s ha e an ad an age o e SSD subjec s in hese si ua ions.
4.4 S udy limi a ions
One issue when looking a BIM pa ien s is he high a iabili y o measu emen esul s inside he
g oup, due o he high amoun o possible hea ing losses. I is di icul o p ecisely es ima e he
in luence o hea ing loss. Howe e , he esul s o his s udy s a e ha he CI ea is he BE o BIM
pa ien s, as also ound by Williges e al. (2019) and hus he con ibu ion o he acous ical ea o
SR is a he small, especially o complex lis ening si ua ions. Based on he ecommenda ions o
he G-BA (2021), an inc ease o 20% o MWR is enough o a success ul HA p o ision. Hence,
e en wi h pe ec ly i ed HAs, big di e ences be ween he subjec s a e possible. No es ic ions
we e made ega ding he SR o he acous ical hea ing ea . Fo be e in e p e a ion and analysis
o u u e esul s, i could be use ul o de ine subg oups inside he BIM g oup, based on he hea ing
loss o he SR wi h HA only. When in e p e ing he in luence o HAs on SR by compa ing SSD
and SSDaided esul s, i should be no ed ha no acclima iza ion pe iod was p o ided and ha
measu ed pu e one h esholds do no necessa ily indica e he need o HA p o ision in mos SSD
pa ien s. Ano he aspec , ha should be aken in o accoun when in e p e ing he esul s, is he
small numbe o pa icipan s, leading o a a he poo s a is ical powe . Howe e , o a s udy
including CI pa ien s and ca ied ou a a single ins i u ion, he numbe o pa icipan s is a he
la ge. As a compa ison Williges e al. (2019) ec ui ed eigh SSD and eigh BIM pa ien s om
wo di e en ins i u ions. While ou SSD g oup’s esul s a e compa able o he indings o o he
s udies, he BIM g oup showed p edominan ly be e esul s compa ed o li e a u e. This may be
a ibu ed o he selec ion o he pa icipan s. Based on bo h he measu emen esul s and he
subjec i e assessmen o he esponsible audiologis , all pa icipan s in his s udy we e a ed as
high pe o me s. The e o e, he gene aliza ion o he indings o he en i e y o SSD and BIM
pa ien s should be app oached wi h cau ion.
22
5 Conclusions
This s udy sys ema ically assessed speech ecogni ion pe o mance o si ua ions wi h di e en
maske s as well as o dynamic complex lis ening si ua ions, o SSD and BIM pa ien s.
Fu he mo e, he hea ing- ela ed cogni i e pe o mance o he pa icipan s was assessed. The
main conclusions a e:
1. CCOLSA is easible o bo h SSD and BIM pa ien s.
2. Compa able monosyllabic wo d ecogni ion wi h he CI does no indica e compa able
speech ecogni ion o complex lis ening si ua ions.
3. SSD pa ien s eached lowe SRTs (be e speech ecogni ion) han BIM pa ien s o he
maske s olnoise, FIT and di use ca e e ia noise.
4. SSD pa ien s gain an ad an age h ough selec i e a en ion in complex lis ening
si ua ions; BIM pa ien s do no .
5. SSD pa ien s showed be e hea ing- ela ed cogni i e pe o mance han BIM pa ien s.
Speech ecogni ion in e e yday li e is in luenced by a a ie y o ac o s. The success ul
measu emen o he CCOLSA e ealed new di e ences be ween SSD and BIM pa ien s. As he
esul s o his s udy show, cu en s anda d p ocedu es, such as he measu emen o he FBE and
he use o s a iona y signals, do no su icien ly cap u e hese e ec s. The e o e, an ex ension o
include es s ha inco po a e speech-in-speech condi ions and assess cogni i e abili ies is
ecommended.
Howe e , u he esea ch is needed o shed mo e ligh on he di e ences be ween SSD and BIM
pa ien s. In pa icula , he in luence o hea ing loss in BIM pa ien s seems o be a p omising
app oach and dese es u he in es iga ion.
23
Re e ences
Ah lich, M. (2013). Op imie ung Und E alua ion Des Oldenbu ge Sa z es s Mi Weibliche
Sp eche in Und Un e suchung Des E ek s Des Sp eche s Au Die
Sp ach e s ändlichkei . Op imiza ion and e alua ion o he emale OLSA and
in es iga ion o he speake `s e ec s on speech in elligibili y. Bachelo Thesis.
Al akh i, M., Campbell, N., Line on, B., Rowan, D., & Boyle, P. (2024). In e na ional su ey o
bimodal hea ing and bila e al cochlea implan se ice p o ision o adul s. Cochlea
Implan s In e na ional, 25(4), 260–274. h ps://doi.o g/10.1080/14670100.2024.2413267
A nd , S., Laszig, R., Aschendo , A., Beck, R., Schild, C., Hassepass, F., Iho s , G., K oege , S.,
Ki chem, P., & Wesa g, T. (2011). Einsei ige Taubhei Und Cochlea -implan -Ve so gung:
Audiologische Diagnos ik Und E gebnisse. HNO, 59(5), 437–446.
h ps://doi.o g/10.1007/s00106-011-2318-8
A nd , S., Laszig, R., Aschendo , A., Hassepass, F., Beck, R., & Wesa g, T. (2017). Cochlea-
Implan a -Ve so gung on Pa ien en Mi Einsei ige Taubhei Ode Asymme ischem
Hö e lus . HNO, 65(Suppl 2), 98–108. h ps://doi.o g/10.1007/s00106-016-0297-5
Ausili, S. A., Ag e be g, M. J. H., Engel, A., Voel e , C., Thomas, J. P., B ill, S., Snik, A. F. M.,
Daze , S., Van Ops al, A. J., & Mylanus, E. A. M. (2020). Spa ial Hea ing by Bila e al
Cochlea Implan Use s Wi h Tempo al Fine-S uc u e P ocessing. F on ie s in
Neu ology, 11, 915. h ps://doi.o g/10.3389/ neu .2020.00915
A an, P., Gi aude , F., & Büki, B. (2015). Impo ance o Binau al Hea ing. Audiology & neu o-
o ology, 20 Suppl 1, 3–6. h ps://doi.o g/10.1159/000380741
Beu elmann, R., & B and, T. (2006). P edic ion o Speech In elligibili y in Spa ial Noise and
Re e be a ion o No mal-Hea ing and Hea ing-Impai ed Lis ene s. The Jou nal o he
Acous ical Socie y o Ame ica, 120(1), 331–342. h ps://doi.o g/10.1121/1.2202888
B and, T., & Kollmeie , B. (2002). E icien Adap i e P ocedu es o Th eshold and Concu en
Slope Es ima es o Psychophysics and Speech In elligibili y Tes s. The Jou nal o he
Acous ical Socie y o Ame ica, 111(6), 2801–2810. h ps://doi.o g/10.1121/1.1479152
30
Wagene , K. C., Hansen, M., & Lud igsen, C. (2008). Reco ding and Classi ica ion o he
Acous ic En i onmen o Hea ing Aid Use s. Jou nal o he Ame ican Academy o
Audiology, 19(4), 348–370. h ps://doi.o g/10.3766/jaaa.19.4.7
Williges, B., Wesa g, T., Jung, L., Ge en, L. I., Radelo , A., & Jü gens, T. (2019). Spa ial Speech-
in-Noise Pe o mance in Bimodal and Single-Sided Dea Cochlea Implan Use s. T ends
in hea ing, 23, 2331216519858311. h ps://doi.o g/10.1177/2331216519858311
Winkle , A., & Holube, I. (2014). Was Wissen Wi Übe Den F eibu ge Sp ach es ? Zei sch i
ü Audiologie (Audiological Acous ics), 53, 146–154.
Winkle , A., Wa ken in, L., Denk, F., Huss ed , H., Sankowksy-Ro he, T., Blau, M., & Holube, I.
(2025). Re e ence Speech- ecogni ion cu es o a Ge man monosyllabic es in noise:
E ec s o loudspeake con igu a ion and oom acous ics. In e na ional Jou nal o
Audiology, 64(7), 695–704. h ps://doi.o g/10.1080/14992027.2024.2401519
Zei le , D. M., Do man, M. F., Na ale, S. J., Loiselle, L., Yos , W. A., & Gi o d, R. H. (2015).
Sound Sou ce Localiza ion and Speech Unde s anding in Complex Lis ening
En i onmen s by Single-sided Dea Lis ene s A e Cochlea Implan a ion. O ology &
neu o ology: o icial publica ion o he Ame ican O ological Socie y, Ame ican
Neu o ology Socie y [and] Eu opean Academy o O ology and Neu o ology, 36(9), 1467–
1471. h ps://doi.o g/10.1097/MAO.0000000000000841
