PHYLO-ANALYSIS OF FOLK TRADITIONS: A METHODOLOGY FOR
THE HIERARCHICAL MUSICAL SIMILARITY ANALYSIS
Hilda Rome o-Velo1Gilbe o Be na des2Susana Lad a1
José R. Pa amá1Fe nando Sil a-Coi a1
1Uni e sidade da Co uña, CITIC, Da abase Labo a o y, Spain
2INESC TEC, Facul y o Enginee ing, Uni e si y o Po o, Po ugal
1{h. elo, susana.lad a, jose.pa ama, e nando.sil a}@udc.es, 2[email p o ec ed]
ABSTRACT
This s udy in oduces and e alua es a new me hodology
o c oss-cul u al e hnomusicological analysis o symbolic
music. We in es iga e music simila i y in popula adi-
ions oo ed in o al ansmission by iden i ying sha ed pa -
e ns a scale ac oss mul iple hie a chies. The no el y o
ou app oach lies in expanding musical simila i y phylo-
analysis- ypically adop ing alignmen me ics ha com-
pa e en i e sco es- o s uc u ally awa e ph ases and mac o-
s uc u e (i.e., o m) alignmen . Addi ionally, we explo e
pa e ns de i ed om mul iple ep esen a ions (ch oma ic
in e al, dia onic in e al, hy hmic a ios, and a combina-
ion o hem) o acili a e he explo a ion o s ylis ic a ini-
ies ac oss musical gen es and adi ions. Ou me hod is
es ed on a new da ase o 600 Galician and I ish pop-
ula music sco es, which includes expe anno a ions o
21 gen es ( ou sha ed be ween he wo adi ions) and
de ailed ph ase in o ma ion, all made a ailable as open-
access da a. We use he gen e sepa a ion a io o exam-
ine how alignmen s a egies cap u e s ylis ic s uc u e,
p o iding insigh s ha suppo musicological explo a ion
ac oss gen es and adi ions. The esul ing phylogene ic
ees and dis ance ma ices e eal ela ionships among a-
di ions, gen es, and sco es, acili a ing he explo a ion o
c oss-cul u al in luences and enabling he iden i ica ion o
sha ed pa e ns a mul iple hie a chies.
1. INTRODUCTION
The e hnomusicological s udies o popula music oo ed
in o al ansmission ac oss cul u es, he ea e e e ed o
as olk music, ha e inc easingly elied on la ge-scale da a
analysis o unco e deep s uc u al pa e ns and e olu ion-
a y p ocesses. As musical adi ions in e sec and e ol e
o e ime, esea che s ace he challenge o ansc ib-
ing, assembling and p ocessing as and complex music
© H. Rome o-Velo, G. Be na des, S. Lad a, J. R. Pa amá
and F. Sil a-Coi a. Licensed unde a C ea i e Commons A ibu ion 4.0
In e na ional License (CC BY 4.0). A ibu ion: H. Rome o-Velo, G.
Be na des, S. Lad a, J. R. Pa amá and F. Sil a-Coi a, “Phylo-Analysis
o Folk T adi ions: A Me hodology o he Hie a chical Musical Simi-
la i y Analysis”, in P oc. o he 26 h In . Socie y o Music In o ma ion
Re ie al Con ., Daejeon, Sou h Ko ea, 2025.
da ase s. This need has led o he adop ion o compu a-
ional me hods om disciplines like biology, whe e phy-
logene ic echniques—o iginally designed o acing ge-
ne ic e olu ion—a e now being used o explo e how musi-
cal ai s a e ansmi ed and ans o med ac oss cul u es.
These me hods ha e been applied o a wide ange o mu-
sical phenomena, om iden i ying melodic pa e ns [1, 2]
o s udying musical ansmission in o al adi ions [3]. Ad-
di ionally, phylogene ic ees ha e been used o ace he
e olu ion o polyphony om he Ba oque o Roman ic pe-
iods [4] and o analyze plainchan melodies [5]. While
some s udies ocus on hy hmic analysis [6], none ha e
ye examined musical o m hie a chically, ea ing s uc-
u e and sha ed pa e ns ac oss le els as e ol ing en i ies.
Iden i ying musical pa e ns and in e ing o m a e key
challenges in analysing music om di e en adi ions.
Volk e al. [7, 8] a gue ha expe s eliably ecognize
sha ed melodic pa e ns as indica o s o olk music am-
ilies. Building on his idea, we p opose a me hodology
o analysing la ge sco e collec ions based on sha ed pa -
e ns ac oss mul iple musical hie a chies. Ou app oach
ex ends p io wo k on musical simila i y [9–12] by ap-
plying alignmen me ics om bioin o ma ics no only o
pi ch and hy hm bu also o s uc u al elemen s such as
ph ases and o m. This s a egy educes spu ious ma ches
be ween songs ha do no ul ill he same s uc u al ole.
To suppo his hie a chical simila i y analysis, we com-
piled a new da ase o 600 Galician and I ish songs, anno-
a ed a he ph ase le el by expe s. These wo adi ions
sha e deep his o ical ies h ough Cel ic he i age [13, 14]
and 20 h-cen u y cul u al exchange [15], which in oduced
he use o he iddle in adi ional Galician music [16]. The
ise o Cel ic music es i als in he 1970s, such as O -
iguei a and Lo ien [17, 18], u he ein o ced hese con-
nec ions. S ill, o ou knowledge, no p io s udy di ec ly
compa es he sha ed gen es be ween I eland and Galicia.
Thus, wi h his da ase , we explo e whe he ou me hodol-
ogy can e eal sha ed s uc u al pa e ns, while also high-
ligh ing he dis inc i e ai s o each gen e and adi ion.
No ably, ou aim is o p o ide expe s wi h an analy i-
cal ool o explo ing la ge da ase s o musical sco es. To
acili a e his endea ou , supplemen a y igu es and dia-
g ams a e gene a ed, including in e ed gen e ee ep e-
sen a ions. I should be s essed, howe e , ha his is no a
me hodology o gen e iden i ica ion o classi ica ion.
190
Figu e 1. O e iew o he p oposed me hodology. In-
pu sco es a e encoded acco ding o a selec ed ea u e ep-
esen a ion and compa ed pai wise h ough a simila i y
me hod, p oducing he co esponding dis ance ma ix and
phylogene ic ee.
The pape is s uc u ed as ollows: Sec ion 2 in oduces
he anno a ed da ase ; Sec ion 3 de ails he me hodology;
Sec ion 4 e alua es he p oposed me hods; Sec ion 5 dis-
cusses a case s udy; and Sec ion 6 ou lines he conclusions.
2. DATASET AND ANNOTATIONS
We p esen a new anno a ed da ase o 600 music sco es
om Galician and I ish olk adi ions, wi h a balanced
dis ibu ion o 300 pieces pe adi ion. I ea u es 21 gen-
es, wi h 11 and 10 gen es pe Galician and I ish adi-
ion, espec i ely. Fou gen es a e sha ed be ween he
wo adi ions: ma ch (“ma cha” in Galician), mazu ka
(“mazu ca”), polka (“polca”) and wal z (“ alse”). The e-
maining Galician gen es a e: “alalás”, “ oliadas”, “jo as”,
“muiñei as”, “pasaco edoi as”, “pasodobles”, and “ um-
bas”. The I ish gen es also include: “ba ndance”, “ho n-
pipe”, “jig”, “ eel”, “slide”, and “s a hspey”. F om each
gen e, 30 songs we e selec ed, excep o he Galician
“mazu ca” and “polca”, wi h 15 songs each. The da ase
was sou ced om he Folko eca Galega 1and The Ses-
sion 2.
We pa sed each sco e o a oid epe i ion and Da Capo
ins uc ions, hus expanding he musical su ace o s ing
sequences o musical e en s. Music expe s anno a ed
each musical piece, indica ing ph ase endings. In o al, he
da ase includes 5,844 ph ases, o which 3,373 co espond
o he Galician adi ion and 2,471 o he I ish one. The
expanded sco es in **ke n 3and MusicXML o ma s and
he ph ase anno a ions a e openly published 4.
3. METHODOLOGY
This sec ion p oposes a me hodology o s udying he simi-
la i y be ween sco es, gen es, and adi ions. The app oach
is inspi ed by bioin o ma ics, which uses alignmen and
clus e ing echniques o analyse he simila i y o acid o
1h ps:// olko ecagalega.gal/pezas, Ma ch 2025
2h ps://gi hub.com/adac io/TheSession-da a,
Ma ch 2025
3h ps://www.humd um.o g/ ep/ke n/, June 2025
4h ps://gi hub.com/h ome o elo/ olk oo
Figu e 2. Sample o ea u e ex ac ion om a sco e.
p o ein chains. Fu he mo e, hei phylogene ic ees em-
body he idea o acing ela ionships be ween en i ies ha
sha e common oo s, analogous o he s udy o ou gen es
and cul u es ia he ‘music DNA’: pi ches and hy hms.
Figu e 1 shows he me hodology pipeline, highligh ing
i s main modules and da a low. I ou lines he ex ac ion
o musical ea u es om symbolic sco es, he simila i y
me hods used o compa e hem, and he cons uc ion o
phylogene ic ees based on pai wise dis ances.
3.1 Musical Fea u es
Conside ing he o al ansmission o olk music, melodic
con ou , hy hm, and ecu ing mo i s—iden i ied by Volk
and an K anenbu g [11] as key pe cep ual cues—p o ide
a basis o i s ca ego iza ion in o une amilies. Follow-
ing he amewo k p oposed by Zhu e al. [19], each piece
o music is de ined by h ee ea u es: he dia onic in e -
al (D), he ch oma ic in e al (C), and he hy hmic a io
(R). Using hese ea u es, a piece o music can be de ined
independen ly o i s key and ime signa u e o empo, as
a du a ion ela ion is calcula ed. This app oach is pa ic-
ula ly ad an ageous wi hin he olk music con ex , whe e
nume ous pieces a e ecompiled o di e en ins umen s
wi h a ying unes and hy hmic anno a ions.
The dia onic in e al (D) e e s o he di ec ion and
numbe o s eps o he nex no e in he dia onic scale—
o example, B o C yields a alue o 2 (see Figu e 2).
The ch oma ic in e al (C) measu es he numbe o semi-
ones be ween consecu i e no es (1 in ou p e ious exam-
ple). Unison is ep esen ed by 1 in he dia onic in e al and
by 0 in he ch oma ic in e al (no semi one di e ence).
Howe e , unisons a e emo ed du ing ea u e ex ac ion,
as hey do no e lec melodic p og ession. Nega i e al-
ues indica e descending mo ion in bo h ea u es. Toge he ,
hese ep esen a ions o e complemen a y, ansposi ion-
in a ian iews o pi ch, enabling a compa a i e explo-
a ion o melodic con en ac oss adi ions—dia onic in-
e als cap u e s epwise onal s uc u e, while ch oma ic
in e als p o ide ine pi ch esolu ion.
The hy hmic a io (R) is compu ed by di iding he
du a ion o wo consecu i e no es, cap u ing he ela i e
changes, i.e., whe he he du a ion inc eases o dec eases
a each s ep. Fo example, in Figu e 2, a qua e no e is ol-
lowed by wo ied qua e no es; in his s ep, he ollowing
no e’s du a ion is wice ha o he p e ious one. Res s a e
ea ed as no es, and hei hy hmic a io is also compu ed.
In addi ion, we conside wo o he ea u es de i ed om
combining D, C, and R: he dia onic- hy hmic (DR) and
he ch oma ic- hy hmic (CR) ea u es. This app oach en-
P oceedings o he 26 h ISMIR Con e ence, Daejeon, Ko ea, Sep embe 21-25, 2025
191
Figu e 3. GSR me ic o each ea u e using global, sha ed ph ases, o m, and combined simila i y me hods. A a ic ligh
scale highligh s bes (highe ) and wo s (lowe ) esul s. Highe alues indica e gen es a e be e in e ed.
Figu e 4. GSR d ops wi h noise, nea ing andom baseline.
ables sco e ep esen a ion ega dless o key o ime signa-
u e. To p ope ly combine pi ch and hy hm de ini ions, he
cha ac e ‘ ’ is in oduced o indica e melodic es s. This
ensu es ha odd a ay posi ions de ine pi ch, while e en
posi ions de ine hy hm. Figu e 2 p o ides an example o
a sco e agmen wi h he esul ing i e ea u es.
3.2 Simila i y Me hods
Fo each pai o sco es in he da ase , we compu e hei
simila i y pe musical ea u e. F om now on, le us con-
side ha a musical piece o ph ase is ep esen ed as a ea-
u e ec o . We explo ed ou me hods o compu e sim-
ila i y, aiming o assess hei obus ness in g ouping he
da ase ’s gen es, assuming ha pieces wi hin each gen e
end o sha e pi ch and hy hmic pa e ns.
The i s me hod consis s in pe o ming di ec align-
men be ween sco e pai s. This equen ly used me hod
[10, 20] se es as a baseline o check whe he hie a -
chical s uc u al knowledge imp o es in-gen e g oupings.
The nex wo me hods iden i y common ph ases be ween
sco es p io o compu ing simila i y. One ocuses on he
numbe o ph ases sha ed; he o he accoun s o musical
o m, i.e., he empo al alignmen o sha ed ph ases. The
inal me hod combines he p eceding wo.
3.2.1 Global Simila i y
This is he mos di ec compa ison me hod. I com-
pu es simila i y be ween he en i e ea u e ec o s o wo
pieces by compa ing hei global alignmen , as p oposed
by Needleman e al. [21] o sea ching simila i ies in he
amino acid sequences o wo p o eins. The algo i hm as-
signs a +1 penal y o misma ches and gaps, so lowe align-
men sco es indica e g ea e simila i y.
3.2.2 Sha ed Ph ases Simila i y
As he sco es ha e been di ided in o ph ases, he numbe
o ph ases hey sha e can be compa ed. Be o e his, i is
necessa y o assess which ph ases a e conside ed equal a
he ea u e le el unde conside a ion. Gi en ha many di -
e only in a ew no es, we ea i as a clus e ing p oblem.
Fi s , global alignmen dis ances a e compu ed be ween
all ph ase pai s. We hen adap he Quali y Th eshold (QT)
Clus e ing algo i hm (Heye e al. [22]), seeding clus e s
wi h ph ases a ze o dis ance and cons aining clus e co-
hesion ia a maximum diame e a he han a minimum
size o p e en g ouping dissimila ph ases. Tha diame-
e h eshold is se o he 10 h pe cen ile o he ea u e’s
dis ance dis ibu ion—p o iding su icien sensi i i y o
g oup simila ph ases while a oiding o e - agmen a ion.
A ph ase joins he i s clus e whose dis ance o e e y ex-
is ing membe emains below his h eshold; i none qual-
i ies, i o ms a single on clus e .
Each clus e ecei es a unique ID. We encode each
sco e as he mul ise o i s ph ase-clus e IDs and com-
pu e he esul ing equency ec o . The sha ed-ph ases
P oceedings o he 26 h ISMIR Con e ence, Daejeon, Ko ea, Sep embe 21-25, 2025
192
Figu e 5. Gen e ee o hy hmic ea u e using combined
75% o m + 25% sha ed ph ases simila i y. I ish and Gali-
cian gen es highligh ed in blue and g een, espec i ely.
alignmen alue be ween wo sco es is hen de ined as he
Euclidean dis ance be ween hei equency ec o s — a
symme ic, me ic measu e ha na u ally a ises om he
geome y o ec o spaces, whe e simila i y co esponds
o spa ial p oximi y, and a pe spec i e consis en wi h es-
ablished ec o -space models in music analysis [23].
Example. Le Sco e A: P1, P2, P1, P3; and Sco e B:
P2, P4, P4, P1. Suppose he ph ases a e g ouped as ol-
lows: P1, P2∈C1,P3∈C2,P4∈C3. Then, Sco e
A co esponds o clus e sequence {C1, C1, C1, C2}, and
Sco e B o {C1, C3, C3, C1}. These a e encoded as e-
quency ec o s: [3,1,0] and [2,0,2], espec i ely. The Eu-
clidean dis ance is p(3−2)2+ (1−0)2+ (0−2)2=√6.
3.2.3 Fo m Simila i y
Gi en ha each sco e is de ined by he clus e ID o i s
ph ases, i is possible o compu e a global alignmen based
on his de ini ion. This app oach allows us o conside
ph ase o de explici ly, cap u ing s uc u al p ope ies ha
eme ge om he hie a chical o ganiza ion o he sco es.
Example. Clus e sequences: {C1, C1, C2, C2} s.
{C1, C2, C1, C2}. Bo h yield a equency ec o o [2,2]
(sha ed segmen s dis ance = 0), bu di e in s uc u e, e-
sul ing in non-ze o o m dis ance.
3.2.4 Combined Simila i y
The combina ion o sha ed ph ases simila i y and o m
simila i y p o ides a sui able model. The alignmen al-
ues ob ained o each compa ison me hod a e i s sub-
jec ed o min-max no malisa ion, a da a p ep ocessing
echnique ha ans o ms hem o a s anda d scale. The
no malised alues a e hen combined o each sco e pai
using a weigh ed sum. We examine h ee combina ions o
o m (F) and sha ed ph ases (S): 0.25 ×F+ 0.75 ×S;
0.5×F+ 0.5×S; and 0.75 ×F+ 0.25 ×S.
3.3 Phylogene ic T ees
A e calcula ing a dis ance ma ix be ween all he sco es
in he da ase o each combina ion o ea u es and com-
Figu e 6. Hea map o hy hmic dis ances be ween gen-
es using 75% o m + 25% sha ed ph ases. Ci idis scale
highligh s close gen es in yellow, dis an ones in blue.
pa ison me hods, i is now possible o cons uc a phyloge-
ne ic ee o all such models. The algo i hm used o his
pu pose is Neighbou Joining [24], a bo om-up me hod
ha i e a i ely joins he pai o elemen s ha minimises o-
al b anch leng h. This app oach yields un oo ed ees and
does no assume a cons an a e o change, making i well
sui ed o explo a o y con ex s whe e no common ances o
is p esumed. While he ee igu es shown in his pape a e
oo ed o be e isual depic ion, his does no imply em-
po al o genealogical de i a ion—only s uc u al simila i y
based on he selec ed compa ison ea u es.
A e cons uc ing he phylogene ic ees o all combi-
na ions o ea u es and simila i y me hods, we explo e how
each model o ganises he sco es in ela ion o he gen e
labels p o ided in he da ase . This is done by compu -
ing a dis ance be ween each pai o gen es, de ined as he
a e age ee dis ance be ween all sco es assigned he e-
spec i e gen es. The esul ing gen e-le el dis ance ma ix
is used o gene a e a phylogene ic ee ia he Neighbou
Joining algo i hm. The ma ix is s o ed in XLSX o ma ,
and a hea map is gene a ed om i s min-max no malised
e sion o isual analysis. This ep esen a ion is no in-
ended as a classi ica ion o alida ion o gen e ca ego ies,
bu a he as an analy ical ool o con as models and sup-
po musicological in e p e a ion.
An e hnomusicologis analysing a gi en ea u e and
compa ison me hod may begin by examining he gen e ee
and hea map o obse e p oximi y be ween gen es o in e -
nal dispe sion wi hin one. Based on his o e iew, speci ic
gen es can be selec ed o u he analysis by compu ing a
ee o sco es es ic ed o hem. This enables ocused ex-
plo a ion o c oss-gen e and c oss- adi ion ela ionships,
po en ially e ealing o he wise unno iced connec ions.
P oceedings o he 26 h ISMIR Con e ence, Daejeon, Ko ea, Sep embe 21-25, 2025
193
Figu e 7. Example o wo ph ases om di e en adi-
ions. Rhy hmic a io shown in ed below each ph ase.
4. FEATURE-BASED ANALYSIS OF
PHYLOGENETIC TREES
In o de o asce ain which phylogene ic ee om he com-
pa ison me hods o ganises he gen es mo e cohe en ly, a
me ic is equi ed o e alua e whe he sco es om he same
gen e end o clus e oge he , and whe he hose om di -
e en gen es emain u he apa . Gi en he absence o a
e e ence ee o I ish and Galician musical gen es, he
In e -In a Class Dis ance Ra io me ic is adap ed om
i s equen use in machine lea ning o check how well a
model sepa a es classes. The concep o measu ing in a-
class cohesion and in e -class sepa a ion is undamen al o
clus e ing e alua ion me ics [25] and has also been used
o assess musical gen e sepa a ion models [26]. We will
e e o his me ic as he gen e sepa a ion a io (GSR).
The GSR is calcula ed o each gen e as he a io o
he a e age be ween-gen e dis ance o he a e age wi hin-
gen e dis ance. Values g ea e han one indica e ha gen es
a e well sepa a ed and ha sco es om di e en gen es a e
mo e dis an om each o he han om he same gen e.
Con e sely, alues lowe han one sugges ha he dis ance
wi hin a gen e exceeds he dis ance be ween gen es. This
las case indica es ha he gen es a e qui e simila and ha
we need sub-gen es o dis inguish hem be e .
GSR =a e age in a-gen e dis ance
a e age in e -gen e dis ance
Figu e 3 p esen s he GSR esul s by gen e ac oss phy-
logene ic ees de i ed om each ea u e, acco ding o
he di e en simila i y compa ison me hods. On a e age,
o m simila i y—cap u ing he ph ase-le el s uc u e o
music—p oduces highe GSR alues, wi h he 75% o m
+ 25% sha ed ph ases combina ion yielding he s onges
esul s. The analysis e eals gen es ha a e mo e clea ly
di e en ia ed (e.g. “Alalás”, “Polka”, “Slides”, “S a h-
spey”) and o he s wi h g ea e o e lap o lowe in a-gen e
cohesion (e.g. “Wal z”, “Valses”, “Jig”). O e all, he GSR
me ic p o ides a nuanced cha ac e iza ion o s ylis ic di -
e en ia ion in he I ish and Galician adi ions.
To e alua e he in e p e abili y o he GSR scale, we
conduc ed a sensi i i y analysis (see Figu e 4). An ide-
alized gen e ee p oduced a GSR o 51.0. Noise was g ad-
ually in oduced in o he gen e assignmen s (up o 50%),
leading o a sha p decline in he a e age GSR, which ap-
p oached a baseline. This baseline was es ablished using
10,000 ees based on andom dis ance ma ices, esul ing
Figu e 8. Gen e ee o ch oma ic- hy hmic ea u e using
combined 75% o m + 25% sha ed ph ases simila i y. I ish
and Galician gen es highligh ed in blue and g een.
in a dis ibu ion igh ly cen e ed a 1.00 (wi h a maximum
ou lie o 1.0099). The bes -pe o ming me hods eached
GSR alues be ween 1.17 and 1.21, signi ican ly exceed-
ing alues expec ed unde andom condi ions. These e-
sul s suppo he s a is ical signi icance o he gen e-le el
pa e ns in e ed by he p oposed app oach.
5. ETHNOMUSICOLOGICAL CASE STUDY
To illus a e he po en ial o ou me hodology in p ac i-
cal con ex s, we ocus on he hy hmic and ch oma ic-
hy hmic ep esen a ions, using a combined simila i y
measu e wi h a 75% o m and 25% sha ed ph ase weigh -
ing. These con igu a ions yielded he highes a e age GSR
alues (see Figu e 3), o e ing clea e in e -gen e sepa a-
ion and s onge in a-gen e cohesion. The selec ed ea-
u es hus p o ide a pa icula ly meaning ul basis o ex-
plo ing s ylis ic connec ions ac oss adi ions. This anal-
ysis showcases how an e hnomusicologis migh le e age
such ep esen a ions o unco e cul u ally g ounded ela-
ionships wi hin and be ween epe oi es. The sou ce code,
anno a ed da ase , and esul s a e a ailable a h ps:
//gi hub.com/h ome o elo/ olk oo (Docke
image p o ided a h ps://hub.docke .com/ /
h elo/ olk oo ).
5.1 Rhy hmic Pa e ns
A p elimina y examina ion o he gen e ee in Figu e
5, based on sha ed hy hmic pa e ns and hei empo al
alignmen , shows ha bo h adi ions a e clea ly di ided
in o wo main b anches. The hea map in Figu e 6 e eals
a s onge co ela ion among I ish gen es (e.g., “polka”,
“ eel”, “slide”, and “s a hspey”, which sha e a s ong e-
la ionship as dance unes in duple o quad uple me e —
wi h he excep ion o “slides”, ypically associa ed wi h a
e na y me e ).
Analysing he diagonal o he hea map, “polkas”,
“ eels”, “slides” and “s a hspey” a e gen es wi h a high
deg ee o simila i y be ween in-gen e pieces. These cases
appea o suppo he assump ion—as s a ed in Sec ion
P oceedings o he 26 h ISMIR Con e ence, Daejeon, Ko ea, Sep embe 21-25, 2025
194
Figu e 9. Example o a clus e o common ph ases om sco es o he same adi ion bu belonging o di e en gen es.
Thei co esponding ch oma ic and hy hmic alues a e shown below each ph ase in eal and ed colou , espec i ely.
3.2— ha pieces wi hin a gi en gen e sha e simila hy h-
mic and pi ch cha ac e is ics. In con as , he I ish “wal z”
om he The Session da ase exhibi s a maximum in a-
gen e dis ance (1.0), e lec ing comple e hy hmic dissim-
ila i y among i s own pieces. This ini ially unexpec ed
ou come p omp ed close examina ion. A e discussing
his case wi h T easa Ha kin o he I ish T adi ional Mu-
sic A chi e, i was con i med ha he “wal z” label in
The Session includes a he e ogeneous selec ion o ma-
e ial: alongside I ish wal zes, i comp ises unclassi ied
unes and Sco ish ai s, many o which di e s ylis ically.
Mo e b oadly, such ins ances o low wi hin-gen e simila -
i y sugges ha ce ain gen es may equi e a iche se o
musical desc ip o s—o e en con ex ual in o ma ion— o
become meaning ul poin s o e e ence o app oaches ha
explo e con en -based simila i y ac oss epe oi es.
To illus a e he sys ema ic po en ial o he p oposed
me hodology o c oss-cul u al s udies, conside an e hno-
musicologis examining he ela ionship be ween wo gen-
es om di e en adi ions: he I ish “ba ndance” and he
Galician “ umba”. They could cons uc a phylogene ic
ee o he co esponding sco es and obse e eme ging
simila i ies. Figu e 7, o ins ance, displays wo ph ases—
one om each gen e—wi h a no iceable hy hmic esem-
blance. Despi e he “ba ndance” ypically being pe o med
a a as e empo (wi h accen s e e y wo bea s) and he
“ umba” a ou ing bea -le el emphasis, hei ph asing e-
sul s in compa able accen ual pa e ns, ein o cing he pe -
cep ion o high hy hmic simila i y.
5.2 Ch oma ic-Rhy hmic Pa e ns
The ch oma ic- hy hmic gen e ee (Figu e 8) closely e-
sembles he hy hmic one (Figu e 5) unde he combined
simila i y me ic (75% o m, 25% sha ed ph ases), wi h
co ela ion coe icien s o Pea son = 0.9985, Spea man =
0.9982, and Kendall = 0.9687. This e lec s s uc u ally
equi alen ph ases iden i ied ia clus e ing and o m-based
compa ison. In con as , global simila i y be ween ch o-
ma ic and hy hmic ea u es yields no ably lowe co e-
la ions (Pea son = 0.8739, Spea man = 0.7833, Kendall
= 0.6160), indica ing ha each ea u e cap u es dis inc
gen e-le el ai s. The app oach he eby o e s lexibili y
o explo ing s ylis ic ela ionships ac oss di e en ep e-
sen a ional and compa a i e le els.
Focusing on he Galician b anch, we s udied he ela-
ionship be ween “ umbas”, “pasodobles”, and “ alses”—
gen es ypically in simple me e . Among hese, he sco e
ee e eals a “pasodoble”, a “ umba”, and a “ alse” ha
sha e simila ph ases belonging o he same clus e , as il-
lus a ed in Figu e 9. While hei ch oma ic- hy hmic de-
sc ip ions a e shown sepa a ely o cla i y, p ocessing is
done join ly, al e na ing pi ch and hy hm alues (see Sec-
ion 3). All h ee happen o be in C majo , hough he
me hod is ansposi ion-in a ian . This example highligh s
he me hod’s obus ness in de ec ing descending hi ds by
dia onic s eps despi e di e ing melodic pa hs be ween an-
cho no es—an alignmen no ensu ed by global simila i y.
6. CONCLUSIONS
This a icle p oposes a no el me hodology o analysing
la ge da ase s o sco es ega ding hei pi ch and hy h-
mic musical cha ac e is ics. The p oposed me hodology
is ancho ed in he musical hie a chy, hus es ablishing a
no el model ha inco po a es he musical o m (i.e. s uc-
u e). This app oach has yielded imp o ed ou comes com-
pa ed o baseline pe -sco e global alignmen me hods. The
esul s demons a e ha inco po a ing he musical hie a -
chy h ough i s s uc u al elemen s and iden i ying com-
mon pa e ns leads o enhanced analysis o musical ela-
ionships. Wi hin he con ex o olk music, inco po a -
ing hy hmic and ch oma ic- hy hmic ea u es esul s in a
mo e comp ehensi e depic ion o he da ase . The indings
sugges ha his amewo k, conside ed in i s ull design,
has he po en ial o se e as a aluable ool o e hnomu-
sicologis s conduc ing c oss-cul u al s udies. Fu he mo e,
i demons a es how bioin o ma ics and music can be com-
bined o ad ance and open up new a enues o esea ch.
The p esen pape es ablishes he ounda ions o u-
u e s udies, explo ing he use o o he phylogene ic algo-
i hms, ne wo k-based s a egies such as Neighbo Ne i-
sualisa ions, and me ics inspi ed by phylogene ic signal.
These di ec ions may help add ess phenomena like ho -
izon al ansmission o s ylis ic hyb idisa ion, which ex-
ceed ee ep esen a ions. Fu he wo k wi h epe oi es
om o he adi ions may cla i y which ea u es and hie -
a chical le els a e mos in o ma i e. In pa allel, ongoing
collabo a ion wi h musicologis s aims o e alua e and e-
ine ph ase clus e ing, s eng hening he me hod’s po en ial
as a g ounded amewo k o compa a i e analysis.
P oceedings o he 26 h ISMIR Con e ence, Daejeon, Ko ea, Sep embe 21-25, 2025
195
7. ACKNOWLEDGMENTS
We would like o hank Folko eca Galega and Xulio Xesús
León Pé ez o p o iding access o hei ull a chi e o
Galician adi ional music. We also acknowledge he sup-
po o Pun oGal o he Folko eca p ojec , and all he col-
labo a o s who ha e con ibu ed sco es o he pla o m
(h ps:// olko ecagalega.gal/xen e).
Wo k unded by: CITIC, as a cen e acc edi ed o ex-
cellence wi hin he Galician Uni e si y Sys em and a mem-
be o he CIGUS Ne wo k, ecei es subsidies om he
Depa men o Educa ion, Science, Uni e si ies, and Voca-
ional T aining o he Xun a de Galicia. Addi ionally, i is
co- inanced by he EU h ough he FEDER Galicia 2021-
27 ope a ional p og am (Re . ED431G 2023/01); Xun a
de Galicia/FEDER-UE, Spain [g an numbe s ED481A
2023/006]; This esea ch has been pa ially unded by he
Eu opean Union’s Ho izon Eu ope esea ch and inno a-
ion p og amme unde he Ma ie Sklodowska-Cu ie g an
ag eemen no 101086338.
8. REFERENCES
[1] P. E. Sa age and Q. D. A kinson, “Au oma ic une
amily iden i ica ion by musical sequence alignmen ,”
in P oceedings o he 16 h In e na ional Socie y
o Music In o ma ion Re ie al Con e ence (ISMIR),
2015, pp. 162–168, licensed unde a C ea i e
Commons A ibu ion 4.0 In e na ional License (CC
BY 4.0). [Online]. A ailable: h p://dx.doi.o g/10.
6084/m9. igsha e.1468015
[2] P. E. Sa age, S. Passmo e, G. Chiba, T. E.
Cu ie, H. Suzuki, and Q. D. A kinson, “Sequence
alignmen o olk song melodies e eals c oss-cul u al
egula i ies o musical e olu ion,” Cu en Biology,
ol. 32, no. 6, pp. 1395–1402, 2022. [Online].
A ailable: h ps://doi.o g/10.1016/j.cub.2022.01.039
[3] S. L. Bomin, G. Lecoin e, and E. Heye , “The
e olu ion o musical di e si y: The key ole o
e ical ansmission,” PLoS ONE, ol. 11, no. 3,
p. e0151570, 2016. [Online]. A ailable: h ps:
//doi.o g/10.1371/jou nal.pone.0151570
[4] G. Rambold, D. Neubache , and S. Schießl, “Fin-
ge p in s, ba code sequences and quasi-phylogenies:
Tools o analysing polyphonic music,” PLoS ONE,
ol. 18, no. 3, p. e0280478, 2023. [Online]. A ailable:
h ps://doi.o g/10.1371/jou nal.pone.0280478
[5] J. H. J ., G. A. Ballen, K. H. Mühlo á, and
H. Vlho á-Wö ne , “Towa ds building a phylogeny o
g ego ian chan melodies,” in P oceedings o he 24 h
In e na ional Socie y o Music In o ma ion Re ie al
Con e ence (ISMIR), 2023, pp. 571–578. [Online].
A ailable: h ps://doi.o g/10.5281/zenodo.10265348
[6] J. M. Diaz-Banez, G. Fa igu, F. Gómez, D. Rappapo ,
and G. T. Toussain , “El compás lamenco: A
phylogene ic analysis,” in P oceedings o BRIDGES:
Ma hema ical Connec ions in A , Music, and Science,
2004, pp. 62–70. [Online]. A ailable: h ps://a chi e.
b idgesma ha .o g/2004/b idges2004-62.pd
[7] P. an K anenbu g, A. Volk, F. Wie ing, and R. C.
Vel kamp, “Musical models o olk-song melody
alignmen ,” in P oceedings o he In e na ional
Socie y o Music In o ma ion Re ie al (ISMIR),
2009, pp. 507–512. [Online]. A ailable: h ps:
//ismi 2009.ismi .ne /p oceedings/PS3-20.pd
[8] A. Volk, E. Chew, E. H. Ma gulis, and C. Anag-
nos opoulou, “Music simila i y: Concep s, cogni ion
and compu a ion,” Jou nal o New Music Resea ch,
ol. 45, no. 3, pp. 207–209, 2016. [Online]. A ailable:
h ps://doi.o g/10.1080/09298215.2016.1232412
[9] N. Ca alho, D. Diogo, and G. Be na des, “Compu-
a ional simila i y o po uguese olk melodies using
hie a chical educ ion,” in P oceedings o he 10 h
In e na ional Con e ence on Digi al Lib a ies o Mu-
sicology (DL M). Milan, I aly: ACM, 2023. [Online].
A ailable: h ps://doi.o g/10.1145/3625135.3625152
[10] B. Janssen, P. an K anenbu g, and A. Volk,
“A compa ison o symbolic simila i y measu es
o inding occu ences o melodic segmen s,”
in P oceedings o he 16 h In e na ional Socie y
o Music In o ma ion Re ie al Con e ence (IS-
MIR), 2015, pp. 622–628. [Online]. A ailable:
h ps://dspace.lib a y.uu.nl/bi s eam/handle/1874/
321370/Janssen_e al_P oceedings.pd ?sequence=1
[11] A. Volk and P. an K anenbu g, “Melodic simila i y
among olk songs: An anno a ion s udy on simila i y-
based ca ego iza ion in music,” Musicae Scien iae,
ol. 16, no. 3, pp. 317–339, 2012. [Online].
A ailable: h ps://www.s a .science.uu.nl/~ el k101/
publica ions/a /MusicaeScien iae2012.pd
[12] P. an K anenbu g, A. Volk, and F. Wie ing, “A
compa ison be ween global and local ea u es o
compu a ional classi ica ion o olk song melodies,”
Jou nal o New Music Resea ch, ol. 42, no. 1, pp.
1–18, 2013. [Online]. A ailable: h ps://doi.o g/10.
1080/09298215.2012.718790
[13] M. Gu ié ez-S einkamp, “The wo ld o he i ish / el
mundo i landés,” TSN: The Spanish Newsle e , no. 9,
pp. 77–89, Janua y–June 2020, smi hsonian Fellow &
Fulb igh Specialis (USA). [Online]. A ailable: h ps:
//dialne .uni ioja.es/desca ga/a iculo/7855918.pd
[14] E. A. F anch, “Ca los núñez y la música cel a,” Re is a
de Folklo e, no. 509, pp. 89–106, 2024. [Online].
A ailable: h ps:// unjdiaz.ne / olklo e/07 icha.php?
ID=5097&NUM=509
[15] X. M. S. Rei, “Language and music in galicia and
i eland in he ea ly 20 h cen u y,” Oceánide, ol. 13,
2020. [Online]. A ailable: h ps://oceanide.es/index.
php/012020/a icle/ iew/37
P oceedings o he 26 h ISMIR Con e ence, Daejeon, Ko ea, Sep embe 21-25, 2025
196
[16] A. F. Vásquez, “The galician iddle s yle,” in C ossing
O e : Fiddle and Dance S udies om a ound he
No h A lan ic 3, I. Russell and A. K. Guigné, Eds.
Abe deen, UK: The Elphins one Ins i u e, Uni e si y
o Abe deen, 2010, pp. 199–214. [Online]. A ail-
able: h ps://au a.abdn.ac.uk/bi s eam/handle/2164/
5048/C ossing_O e _2010_Ch._16_V_squez_.pd
[17] M. Albe o, “The cel ici y o galicia and he a i al o
he insula cel s,” in P oceedings o he Ha a d Cel ic
Colloquium, ol. 24/25, 2009, pp. 1–15. [Online].
A ailable: h ps://www.js o .o g/s able/40285178
[18] J. C. Cal o-So elo, “In he name o ossian: Cel ic gali-
cia and he ’b o he s om he no h’,” Social Iden i ies,
ol. 25, no. 6, pp. 828–842, 2019. [Online]. A ailable:
h ps://doi.o g/10.1080/13504630.2018.1564267
[19] T. Zhu, R. Fou nie -S’nieho a, P. Rigaux, and
N. T a e s, “A amewo k o con en -based sea ch
in la ge music collec ions,” Big Da a and Cogni i e
Compu ing, ol. 6, no. 1, 2022. [Online]. A ailable:
h ps://www.mdpi.com/2504-2289/6/1/23
[20] B. Janssen, P. an K anenbu g, and A. Volk, “Finding
occu ences o melodic segmen s in olk songs
employing symbolic simila i y measu es,” Jou nal o
New Music Resea ch, ol. 46, no. 2, pp. 118–134,
2017. [Online]. A ailable: h ps://doi.o g/10.1080/
09298215.2017.1293802
[21] S. B. Needleman and C. D. Wunsch, “A gene al
me hod applicable o he sea ch o simila i ies
in he amino acid sequence o wo p o eins,”
Jou nal o Molecula Biology, ol. 48, no. 3,
pp. 443–453, 1970. [Online]. A ailable: h ps:
//doi.o g/10.1016/0022-2836(70)90057-4
[22] L. J. Heye , S. K uglyak, and S. Yooseph, “Explo ing
exp ession da a: iden i ica ion and analysis o
coexp essed genes,” Genome Resea ch, ol. 9,
no. 11, pp. 1106–1115, 1999. [Online]. A ailable:
h ps://genome.cshlp.o g/con en /9/11/1106
[23] D. Me edi h, K. Lems öm, and G. A. Wiggins,
“Algo i hms o disco e ing epea ed pa e ns in
mul idimensional ep esen a ions o polyphonic mu-
sic,” Jou nal o New Music Resea ch, ol. 31,
no. 4, pp. 321–345, 2002. [Online]. A ailable:
h ps:// esea ch.gold.ac.uk/id/ep in /1008/
[24] N. Sai ou and M. Nei, “The neighbo -joining me hod:
a new me hod o econs uc ing phylogene ic ees,”
Molecula Biology and E olu ion, ol. 4, no. 4,
pp. 406–425, 1987. [Online]. A ailable: h ps:
//academic.oup.com/mbe/a icle/4/4/406/1029664
[25] X. Xie and G. Beni, “A alidi y measu e o uzzy
clus e ing,” IEEE T ansac ions on Pa e n Analysis
and Machine In elligence, ol. 13, no. 8, pp. 841–847,
1991. [Online]. A ailable: h ps://www.compu e .o g/
csdl/jou nal/ p/1991/08/i0841/13 RUynHuk4
[26] G. Tzane akis and P. Cook, “Musical gen e clas-
si ica ion o audio signals,” IEEE T ansac ions on
Speech and Audio P ocessing, ol. 10, no. 5,
pp. 293–302, 2002. [Online]. A ailable: h ps:
//www.cs.cmu.edu/~g zan/wo k/pubs/ sap02g zan.pd
P oceedings o he 26 h ISMIR Con e ence, Daejeon, Ko ea, Sep embe 21-25, 2025
197
