The Role of Timbre in the Memorization of Microtonal

Intervals
Pro Gradu
Rafael Ferrer Flores
Department of Music
University of Jyväskylä
September 2007
JYVÄSKYLÄN YLIOPISTO
Tiedekunta - Faculty
Humanities
Laitos - Department
Music
Tekijä - Author
Rafael Ferrer Flores
Työn nimi - Title
The Role of Timre in the Memori!ation of Microtonal "nter#als
$ppiaine - %uject
Music& Mind and Technolo'y
Työn laji - Le#el
Master(s Thesis
Aika - Month and year
%eptemer )**+
%i#umäärä - ,umer of pa'es
-.
Tii#istelmä - Astract
The aim of this thesis /as to determine if timre has any effect in the memori!ation of
melodic inter#als0 For this purpose& a test /as de#eloped in /hich )1 sujects heard
an inter#al& and after + seconds of silence /ere played three options from /hich they
had to select the ori'inal inter#al0 The sound samples composin' each tar'et inter#al&
had one control and three de'rees of timral modification0 %uch modifications
consisted in alterin' the ori'inal partial structure of the sound samples0 %pectral
Modellin' and Additi#e %ynthesis techni2ues /ere used to reali!e these modifications0
Results su''est that is possile to enhance or impair the aility of e3tractin' cues for
memori!in' inter#als y alterin' timral structure0
Asiasanat - 4ey/ords
timre& memory& ear trainin'
%äilytyspaikka - Depository
Muita tietoja - Additional information
bstract
The aim of this thesis /as to determine if timre has any effect in the
memori!ation of melodic inter#als0 For this purpose& a test /as de#eloped in /hich )1
sujects heard an inter#al& and after + seconds of silence /ere played three options from
/hich they had to select the ori'inal inter#al0 The sound samples composin' each tar'et
inter#al& had one control and three de'rees of timral modification0 %uch modifications
consisted in alterin' the ori'inal partial structure of the sound samples0 %pectral
Modellin' and Additi#e %ynthesis techni2ues /ere used to reali!e these modifications0
Results su''est that is possile to enhance or impair the aility of e3tractin' cues for
memori!in' inter#als y alterin' timral structure0
Table of !ontents
" Intro#$ction%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%&
' Theoretical consi#erations%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%"(
)05 Music education0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000005)
)0505 6ar trainin'00000000000000000000000000000000000000000000000000000000000000000000000000000000000051
)050) "nter#als0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000057
)0) Timre00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000058
)0)05 Timre and scales000000000000000000000000000000000000000000000000000000000000000000000000005.
)0)0505%pectral shape0000000000000000000000000000000000000000000000000000000000000000000000000000005.
)0)050)9onsonance : Dissonance00000000000000000000000000000000000000000000000000000000000005;
)0)0501%cales0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000)*
)0)0507Relations et/een timre& consonance and scale000000000000000000000000)5
)01 Memory000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000)7
)0105 A three sta'es in one000000000000000000000000000000000000000000000000000000000000000000000)7
)010505%ensory memory00000000000000000000000000000000000000000000000000000000000000000000000000)7
)01050)%hort-term memory0000000000000000000000000000000000000000000000000000000000000000000000)8
)010501Lon'-term memory0000000000000000000000000000000000000000000000000000000000000000000000)-
)010507<orkin' memory0000000000000000000000000000000000000000000000000000000000000000000000000)-
)010) Memory for timre000000000000000000000000000000000000000000000000000000000000000000000000)-
) *m+irical a++roach%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%',
105 Method00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000).
10505 Desi'n0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000).
1050) =articipants000000000000000000000000000000000000000000000000000000000000000000000000000000000000);
10501 Materials00000000000000000000000000000000000000000000000000000000000000000000000000000000000000001*
1050105%timuli0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000001*
Selection..................................................................................................30
Discrimination.........................................................................................31
Analysis...................................................................................................31
Synthesis.................................................................................................35
105010)Apparatus000000000000000000000000000000000000000000000000000000000000000000000000000000000000075
10507 =rocedure0000000000000000000000000000000000000000000000000000000000000000000000000000000000000007)
10508 Results00000000000000000000000000000000000000000000000000000000000000000000000000000000000000000007)
10) Discussion0000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000007-
10)05 Adaptaility000000000000000000000000000000000000000000000000000000000000000000000000000000000007.
10)0) Timre descriptors00000000000000000000000000000000000000000000000000000000000000000000000008*
- !oncl$sions%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%.&
References..............................................................................................................59
Appendix A. Instructive of the experiment............................................................67
Appendix B. Questionnaire....................................................................................68
In#e/ of Ill$strations
Fi'ure 50 Representation of sensory dissonance y =lompt > Le#elt000000000000000000000000000)5
Fi'ure )0 Rou'hness and ratios00000000000000000000000000000000000000000000000000000000000000000000000000000000000000))
Fi'ure 10 63ample of dissonance cur#e computed /ith %ethares( al'orithm0000000000000000)1
Fi'ure 70 Desi'n of the e3periment0000000000000000000000000000000000000000000000000000000000000000000000000000000);
Fi'ure 80 Minima hits from +1 dissonance cur#es0000000000000000000000000000000000000000000000000000000001)
Fi'ure -0 Mean dissonance cur#e for a set of +1 different timres000000000000000000000000000000011
Fi'ure +0 Representation of timre in relation /ith the steps of different scales00000000001+
Fi'ure .0 63ample of adjustment of timral structure0000000000000000000000000000000000000000000000000001.
Fi'ure ;0 %ynthesis types000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000007*
Fi'ure 5*0 Appearance of the e3perimental interface000000000000000000000000000000000000000000000000000075
Fi'ure 550 Accuracy per inter#al00000000000000000000000000000000000000000000000000000000000000000000000000000000000077
Fi'ure 5)0 Accuracy per %ynthesis Type0000000000000000000000000000000000000000000000000000000000000000000000078
Fi'ure 510 Accuracy per "nstrument Type0000000000000000000000000000000000000000000000000000000000000000000007-
Fi'ure 570 Tendency of responses 0000000000000000000000000000000000000000000000000000000000000000000000000000000007+
Fi'ure 580 Histo'ram of responses000000000000000000000000000000000000000000000000000000000000000000000000000000007+
Fi'ure 5-0 Adaptaility per inter#al00000000000000000000000000000000000000000000000000000000000000000000000000000007;
Fi'ure 5+0 $#erall Adaptaility00000000000000000000000000000000000000000000000000000000000000000000000000000000000008*
Fi'ure 5.0 $ne-/ay A,$?A et/een timres of t/o synthesis 'roups (ad( > (s2(00000085
Fi'ure 5;0 $ne-/ay A,$?A et/een timres of t/o synthesis 'roups& t/o data sets81
Fi'ure )*0 9orrelation et/een eha#ioural and acoustical data000000000000000000000000000000000087
Fi'ure )50 %catter plots of rele#ant timral descriptors and eha#ioural data0000000000000088
In#e/ of Tables an# 0orm$las
Table 1. Convergence of ratios from different sources..................................................22
Table 2. Tabulated frequencies for interval selection....................................................34
Table 3. Example of selection of candidates..................................................................36
Table 4. Different combination of subsets to test syntheses type with one-way ANOVA
.........................................................................................................................................45
Formula 1. Computation of ratios for a 12 tone equally tempered system....................20
Formula 2. Relation of partials and steps of a scale......................................................36
7
" Intro#$ction
This /ork has een inspired y the need for hard e#idence o#er the effecti#eness
of traditional practices in music education& in the area of ear trainin' /ithin /estern
culture0 Althou'h the solution for this prolem is o#iously far more comple3 than the
approach that can e co#ered in a sin'le thesis like this& an atomistic approach makes
possile to focus the prolem on the concept of timre and the memory for it0
The concept of timre encloses more 2uestions than ans/ers for the time ein'&
proaly ecause the disentan'lement of its perceptual characteristics in a systematic
2uantitati#e form is a fairly recent /ork0 The modern conception of timre has a
predecessor 'estated durin' the French 6nli'htenment at the hands of @ean =hillipe
Rameau& /ho predicted the e3istence of music uni#ersals in the inner structure of a
sin'le tone0 ,e#ertheless& the possiility to otain some scientific proof for that
conception came some years later as the result of the influence that pro#oked the /ork
of personalities like Fourier& Helmholt!& and Arey& amon' others B/hom are not
mentioned here ecause the intention is to estalish a rief chain of ideas and not doin'
an e3hausti#e historical re#ie/C0
,o/adays in the analysis of timre it can e said that /e ha#e t/o main ranches in the
academic production& one is the approach that uses a #eral description for the soundin'
2ualities of a 'i#en timre& and the second consists in computin' a numerical
description for it0 The first option is the most common used amon' musicians and non-
musicians& and the second is the outcome of research in an area denominated Music
"nformation Retrie#al0 This former ranch pro#ides the niche in /hich this /ork /ould
rest& since /e are interested in searchin' ojecti#e uni#ersals more than particular
differences amon' indi#iduals or 'roups0
6ar trainin' is a #ery important acti#ity of formal musical trainin' in /hich asic skills
of music literacy are de#eloped& for e3ample the necessary ailities for readin':/ritin'
music0 <estern musical tradition has de#eloped a sustantial numer of forms to fulfil
the main ojecti#e of codin' music y cate'ori!in' the sounds& throu'h uildin'
8
minimal construction locks and creatin' a ta3onomy for its physical characteristics0
<ithin this mentioned forms /e picked one that can e presumed as 'eneric& ecause it
can e found in the asic steps of most methodolo'ies0 "t consists in the identification
of melodic inter#als& /hich can e defined as the perceptual distance et/een t/o
consecuti#e musical tones0
"n this task of identifyin' morpholo'ies such as a claimed distance et/een t/o fi3ed
points& the role of memory is the central matter& ecause just as another perceptual
cate'ories like the #isual stimuli e#oked y an sculpture and /hich has its dominion
o#er a constant space& the auditi#e stimuli resultin' from music e3ists only in the
ephemeral present0 "mplyin' /ith this that music has its dominion o#er time and
proposin' that memory can e seen as a human capaility meant to recreate past
e3periences and project them in the futureD just as it is the purpose of learnin'0
The topics mentioned ao#e 'i#es us a frame/ork to e3plore causalities et/een certain
timral characteristics of a 'i#en sound source and the performance of memory&
particularly in the task of recallin' musical dyads0 The main focus is the mnemonic
strate'ies of lon' lastin' effect& just after the short-term memory span0 To this effect /e
run an e3periment in /hich people are asked to memori!e an inter#al& and then after a
period of silence they ha#e to identify the same inter#al et/een three optionsD ein'
t/o of the three options de#iated in tunin' from the ori'inal& and possessin' all the
same timre per trial0 The /hole e3periment consists of -* trials and the construction
of them is three inter#alsE 18*& 88*& +8* cents
5
D fi#e different instrumental sourcesE alto
flute& french horn& ooe& synthetic piano B/hich /e call M"D" pianoC& and o/ed
#iraphoneD and three analysis:adjustment:synthesis methods plus the ori'inal sample0
Thus 'i#in' a cuic matri3 of dimensions 1 3 8 3 7 F -* trials0 The inter#als /ill e
referred as microtonal ecause they fall in a di#ision of the /estern half-tone& /hich is
the standard shortest inter#al traditionally used0 The methodolo'y employed to adjust
the timre is at the core of this /ork ecause in case of succeeded in the first premise
/hich /asE different timres lead to differential mnemonic capailities& the ne3t step
/as to speculate aout /hich kind of modifications /ould lead to those results and /hy0
1 A cent is defined as the hundredth part of a whole tone.
9
The main strate'y consisted in sutractin' the tar'et inter#als from the timral structure
of the tones& ut the entire process is e3plained in detail in section 10501050
The te3t is di#ided mainly in t/o parts& ein' the first the theoretical frame/ork /hich
pro#ides a 'eneral o#er#ie/ and in some cases a plain summary of some /orks related
/ith music education& timre and memory0 The second is a ri'orous empirical
approach& /hich e3plains step y step the e3periment done for this /ork& from the
desi'n& participants and materials to the procedure& results and discussion0
10
' Theoretical consi#erations
A sustantial amount of academic production in music analysis is focused in
/ritten music& and most musicolo'ists /ould a'ree that the in#ention of a set of
symols to code music represented the most important ad#ance in the history of /estern
music0 "t allo/ed the rapid e#olution of music ecause the separation of human and
epistemic ojects BGent > =ople& )**-C0 ,e#ertheless& that code has also a disad#anta'e&
ecause it is just a 'ross representation of the actual sound itself0
The un#eilin' of structure and form is the outcome of analysis& therefore is 'enerally
admitted that music structure relies on rhythm& pitch and timre0 $ther cate'ories can
e considered as the result of the comination of these three features& as it is the case for
melody and harmony0
The set of symols /ritten in a score fitted #ery /ell to study the co'niti#e processes
in#ol#ed in music in terms of rhythm and pitch& ut unfortunately that /as not the case
for music outside those symols& for e3ample the music that makes use of timre as the
main aesthetic resource BFales& )**)& p08+C0
%i'nal processin' technolo'y is related from its semantic ori'in to communication and
codification0 This technolo'y can e seen as a ne/ /ay to represent sounds in a more
accurate /ay& for instance& it is possile to #isuali!e timre chan'es as a function of
time& /hich has permitted speculation aout music structure focusin' on timre features
B<essel& 5;+;C& and /hich 'oes eyond the symolic representation of a sound in a
score& openin' the possiility to understand the role of timre in learnin' music0
<ith the use of ne/ technolo'ies in music education& it is common to find electronic
de#ices in the classroom producin' synthetic sounds& for e3ample in the ear trainin'
class in /hich some teachers also send their students to self-study practices /ith
soft/are made for that specific purpose0 At this point one can speculate aout the
effecti#eness of trainin' the ear /ith one synthetic sound or another& or to e more
specific to understand the characteristics of that sound in order to impro#e the
applications0 $n the other hand it is important to understand that the rele#ance of ear
11
trainin' relies on the fact that it is not the ear that it is actually trained& ut the co'niti#e
schema /hich takes the form of an ima'e in the rain of each indi#idual0 Thus to train
the ear can e e3plained as the impro#ement of precision in music ima'ery /ithin a
particular tonal schema& ecause efore readin'& /ritin'& performin' or analysin' music
one has to make an ima'e of it in the mind B=itt > 9ro/der& 5;;)D Aordon& 5;;+C0
The perception of different timres must lead to the ac2uisition of different ima'es& ut
is there a set of characteristics in timre that can imprint in mind etter than othersH 9an
/e find a 'ood set of timres to impro#e the ear trainin'H "s timre a rele#ant feature in
the optimi!ation of learnin' musicH "s there any especially commendale instrument to
teach the most asic aspects of tonal schemaH
Althou'h research in musical timre has a relati#ely short history& there is crescent
empirical e#idence of its rele#ance for the perception of music0 "n traditional music
education in the /estern culture& the most closely related topic to the study of timre
can e found in the courses of orchestration& ut the approach is ounded to its aesthetic
ori'ins& /hich undoutedly is the result of many years of e#olution in seekin' for
functionality for the senses BHelmholt!& 5;87C0 ,e#ertheless that ar'ued functionality
has een reached throu'h unconscious processes& and the stress in peda'o'ical
applications has een left unattended0 There are of course e3emplar efforts like the
/ork of @a2ues-Dalcro!e& 4odIly& and $rff& /hom specified in their peda'o'ical
methodolo'ies the timres to e used in the early sta'es of musical de#elopment
B9hoksy& Aramson& Aillespie& <oods& Jork& )**5C0 Thou'h this area remains
un2uestioned in a systematic /ay& since there is a lack of literature dealin' for instance&
/ith the differential musical ailities ac2uired y usin' only #oice or the instrumental
set su''ested y $rff0 Furthermore in the area of music education for adults it is ur'ent
to understand ho/ different instruments could optimi!e the musical learnin'& for
e3ample in those countries /ere the official curricula does not contemplate musical
trainin' as an indispensale part of education for the plenary de#elopment of the
indi#iduals0
"n the process of learnin' #ie/ed from the co'niti#e perspecti#e& memory has a central
role not as a container of information& ut as a dynamic unit capale of systemati!e the
perception B=ante#& )**5& p01*5C0 Auditory memory is di#ided in three different
12
cate'ories accordin' /ith the span of retention of percei#ed phenomena0 These three
cate'ories are not isolated locks of processin'& instead& they actually /ork to'ether at
any 'i#en instant& and this interaction recei#es the name of /orkin' memory0 The first
cate'ory is the echoic memory& and it is related /ith e#ents happenin' in no more than
fe/ seconds0 The second is referred to as short-term memory and studies the
phenomena happenin' after the echoic memory and /ithin a time span of appro3imately
. seconds0 And the third is the lon'-term memory& /hich can e considered the most
stale ecause the 2ualities of the information it KretainsL& that can e reconstruction of
e#ents that had happen in the past eyond the domain of short-term memory& or
conditions that had een rehearsed se#eral times0 Lon'-term memory is also related
/ith the idea of schema& /hich is a kind of superior learnin' controllin' the perception
of ne/ phenomena BLeman& 5;;8D %nyder& )***C0
2.1 Music education
Learnin' has t/o sidesD one is the iolo'ical fact of 'enetic herita'e& and the
other has to e /ith the sophistication of strate'ies to take ad#anta'e of that 'enetic
herita'e0 Music has these t/o sides /hen it is learned B=apouMek& )**1C& and for that
reason there is no need to recei#e any special instruction to understand music& althou'h
in order to communicate particular /ays to or'ani!e sounds /e actually do0
"n /estern culture for e3ample& the de'ree of systemati!ation of music has e#ol#ed in
schools speciali!ed in the teachin' of music& and music education is suject to certain
con#entions depicted in the contents of the curricula and in the role of music and
musicians in e#ery society0 Ho/e#er& amon' the o#ious differences et/een each
society& one idea has een preser#ed since the foundations of /estern cultureE not all the
indi#iduals in the society are supposed to ecome producers of music& therefore& it can
e e3pected that amon' the three main musical acti#itiesD composition& performance and
analysis B9hoksy& Aramson& Aillespie& <oods& Jork& )**5C& e#ery person must posses a
fair kno/led'e in at least one of them0 9onse2uently it can e presumed that
alphaeti!ed indi#iduals in /estern /orld ha#e a minimum skill to analyse music
B%erafine& 5;..C0
13
<estern culture and its idea of music education has spread in many parts of the /orld&
ut in each 'eo'raphical re'ion there is a particular set of instruments used& as /ell as
the time dedicated to learn music as a part of 'eneral education BHar'rea#es > ,orth&
)**5C0
Gy considerin' the immense di#ersity of musical sound sources& it is una#oidale to
2uestion if there is no other etter /ay to e educated musically than /ith the traditional
practices 'estated in central 6urope durin' the last millennium0 Moreo#er& it is
imperati#e to ackno/led'e that technolo'y is introducin' in a #ery fast manner ne/
sounds and aesthetic forms& and that /e must e prepared to understand ho/ they /ill
affect our perception of music in the near future0
'%"%" *ar trainin1
There are t/o main sources of music educationD the proper music education that
is reali!ed in the institutions dedicated specifically to education or as a complement of
'eneral education& and the music education implicit in the oral tradition of music0 "n the
institutionali!ed music education ear trainin' is at the core of the curricula for e3ample
in Aermany BHar'rea#es > ,orth& )**5& p07+C& "taly /here one of the main aims is Nto
sharpen percepti#e ailitiesO Bp0+;C and =oland Bp051;C0 A rele#ant data in the case of
Aermany is that 51P of the teachers sur#eyed in 5;;8 had e3perience /ith M"D"
technolo'y in the classroom0
9omputer assisted instruction is an approach in ear trainin'& that has een 'ro/in' for
more than a 2uarter of century& like AQ"D$& /hich /as presented as an "nteracti#e
9omputer-Gased system for "mpro#ement of "nstruction and Research in 6ar-Trainin'
BHofstetter& 5;+8C0 More recent methods /hich ein' printed material also include the
possiility to access throu'h internet to auditi#e media& like the Gen/ard > 4olosick
method B)**8C& use a recorded piano to play the e3ercises0 Also recent #ersions of
soft/are dedicated to ear trainin' use synthetic sounds reproduced /ith a M"D" capale
de#ice& for instance Ear Master 5 B)**+C0 Gut a'ain& the documentation related /ith the
2ualities of sound source or timre is poor or une3istent0
As stated y %looda B)**8C& ear trainin' is an area in /hich teachers are used to
14
discuss Nde#elopin' a 'ood earO& althou'h from the scientific point of #ie/ most
peopleLs ears function e3cellently& and there is nothin' one can do to enhance their
functionin'0 The idea is to find out /hat needs to happen in the rain to produce the
eha#iour that musicians /ould associate /ith a N'ood earO Bp05+-C0 For that effect& the
music learnin' theory de#eloped y 6d/in 60 Aordon B5;;+C fits #ery /ellD in his ook
entitled Learnin' %e2uences in Music the main focus of attention is in the concept of
audiation0 Aordon e3plains that audiation happens /hen /e assimilate music that /e
ha#e heard or performed& and also /hen /e assimilate and comprehend in our minds the
music that comes from a symolic representation of it Bp07CD throu'h this description of
the concept it is easy to understand that the term audiation refers the same phenomena
that 9ro/der > =itt B5;;)C descried as Nima'ery in musicO0 They attract the /ork of
He B9oncernin' "ma'ery& 5;-.C to e3plain that Nima'ery representation is the
acti#ation of the same central neural systems that played a role in the ori'inal e#ent& ut
this time in the asence of the ori'inal sensory acti#ityO Bp01*C0 "ma'ery is linked to
perception& and in the particular case of timre it has demonstrated a hi'h correlation y
usin' eha#ioural and neural data BHalpern& Ratorre& Gouffard > @ohnson& )**7C0
'%"%' Intervals
The plain definition of a musical inter#al is referred as the distance et/een t/o
tones& ut this definition rin's some comple3ities if one think aout the perceptual
meanin' of a distance in the /orld of sounds& and e#en more if /e in#ol#e the nature of
a tone0 %o in order to disentan'le a functional definition of inter#al that /ould e useful
for the present project it ecomes necessary to rin' the concept of pitch as a
Nmorphoporic mediumO B%hepard& )**5C0 Gy usin' this concept& and just as e3emplified
y %hepard& it can e ar'ued that the specific #isual idea /e ha#e of a trian'le does not
chan'e if this trian'le chan'es its position in spaceD /hich implies that #isual space is
also a morphoporic medium0 "n the same sense& the percei#ed pitch space has
morphophoric 2ualities in a manner that ideas of auditi#e forms& such as the trian'le&
can e sketched0 Furthermore& melodies could e re'arded as those forms& as could e
scales& /hich from a reductionist point of #ie/& are nothin' else than sets of inter#als0
"n this thesis& the aim is to in#esti'ate ho/ different timres affect the memori!ation of
15
inter#als& so it is useful to think aout inter#als as sound units /ith particular
morpholo'ical characteristics0
The most rele#ant issue in the perception of inter#als is related /ith the re-co'nition of
patterns& ut ho/ this phenomena takes place in#ol#es a physiolo'ical system and its
capailities0 A tone in a musical conte3t outside the controlled en#ironment of a
laoratory& should e concei#ed as the sum of multiple pitches& /hich e3cite the
hearin' system in a manner that makes it to con#ey an analysis of such pitches& as /ell
as a reduction of them into a sin'le most salient feature kno/n as pitch0 Further
e3planations on ho/ this analysis > reduction takes place had re#ealed that many areas
on the physiolo'ical and neurolo'ical domain are in#ol#ed& and these had een
e3tensi#ely studied durin' the past )* years BGurns& 5;;;C0 For instance& it is kno/n
that humans are ale to discriminate appro3imately 5&7** different fre2uencies& in
discrimination tasks that in#ol#e the comparison of sounds at t/o fre2uencies in
immediate succession BHandel& 5;.;C0
Accordin' to the model descried y Deutsch B5;;;C& pitch is only one sudi#ision of
the ar'ued analysis reali!ed y the hearin' system& /hich is processed and stored in
parallel areas /ith inter#al si!e and timre& amon' other patterns like loudness and
duration0 Furthermore& it is e3pected that these ar'ued sudi#isions ha#e interaction
et/een themD in fact some e#idence su''est that the perception of inter#al si!e tend to
e distorted dependin' on timral #ariations B<arrier& Ratorre& )**)D Russo& Thompson&
)**8C0
2.2 Timbre
Timre is still at the e'innin' of )5st century an elusi#e concept& perhaps
ecause the necessity for an accurate description of it can e seen as a fairly ne/ task0
"t mi'ht e possile that the 2uest for an accurate conception of timre is the result of
the separation of the sound from its source& /hich /as a conse2uence of the industrial
re#olution& and /ith it also the ecomin' of recordin' technolo'y B%chafer& 5;++C0 This
separation rou'ht ne/ sounds that /ere not necessarily the result of an acoustic source
ut sometimes a crude si'nal created /ith a /a#e 'enerator0 Furthermore& the rele#ance
16
of recordin' technolo'y for our recent conception of timre and musical meanin' is as
stron' as it is also unattended y consumers outside the circle of the e3pertise& and it
must e underlined that the sophistication of recordin' techni2ues pursues an aesthetic
ideal BRa'orski-Thomas& )**8C rather than ein' an e3pression of other means& like for
e3ample peda'o'ical efficiency0
The ad#anta'e of countin' /ith an acoustic source as reference& /as the possiility of
associate a certain timre /ith its source in terms of #isual or #eral domains& in such a
/ay that the description of the sound /as made easy& for instance the e3pressionE Nthis
sounds like a000 -somethin' you ha#e e3perienced efore-O0 This #eral e3pression
encloses se#eral cues for memory BRo'ers& )**8C& /hich y some means completes the
information that could satisfy the description of a sound0 $r not& in the case of those
sounds that are so stran'e and ne/ that a #isual reference can e only fictional0 Gut so
far& this kind of #eral descriptions has een the constant throu'h history& at least in the
/estern culture& e#en for music e3perts ein' them instrumentalists& composers or
musicolo'ists0 "n the case of these e3perts& it can e found a #ery elaorated lan'ua'e
to discuss for e3ample aout the desired soundin' result of a particular piece of music0
,e#ertheless this code shared y musicians is far to e homo'eneous& and in some cases
it can e also contradictory at a metaphorical le#el dependin' on the su-cultural
conte3t0
An etymolo'ical approach for the /ord timre re#eals its French ori'ins and accordin'
to Fales B)**8C& the concept in the sense of sound 2uality is the result of a process of
e#olution occurred durin' the ei'hteenth century0 "t /as implicit in se#eral acceptations
/hich /ere actual metaphors of the timre itself& such as consonance or unison& /hich
referred indirectly to a Nne/O separation of this 2ualia of sound0 The prolem since
then has een to find a descripti#e #ocaulary to parse a sound into perceptual
phenomena0
"n this intend there are some references that are /orth to mention& one of them /as
pulished in 5+-8& and it is included in the #olume S? of the Encyclopédie /ritten y
Rosseau in his article aout sound0 This is supposed to e the first definition of timre
in the modern sense& ut y jud'in' the similarities it has /ith the definition of the
"nternational %tandards $r'ani!ation pulished in 5;-*& /e can confirm that it has not
17
chan'ed too much in nearly )** yearsD althou'h the former includes a note /hich makes
an e3plicit reference to the spectrum of the sound& the similarities consist in that oth
descrie /hat timre is not& rather than postulatin' ojecti#e facts aout /hat timre
actually is0 Another 'roup of pulications correspond to those 2uoted y Huron B)**5C
as theoretical approaches& in reference to the /ork of %la/son B5;.8C and McAdams
B5;;8C
)
0 %chaeffer(s Traité des objets musicaux B5;-.C could also e considered as an
antecedent of this 'roup althou'h his /ork could e descried in 'eneral terms as a
ta3onomical approach created for music peda'o'y purposes0 A third 'roup constitutes
perhaps a foundation in the field of music co'nition due the disco#erin' of a
multidimensional perceptual space for timre BArey& 5;++D <essel& 5;+;D McAdams&
<einser'& Donadieu& De %oete& 4rimphoff& 5;;8C0
$li'ated to this effort& ecause /e rely on it& is to point to the /ork of Lartillot >
Toi#iainen B)**+C& /hom had een de#elopin' a computational approach that can e
re'arded as a compendium of 2uantitati#e timre descriptors& /hich are taken from the
/ork of many researches in the area of music co'nition0 The nature of the tool makes
possile the processin' of a sound y modules that emulate the physiolo'ical and
neurolo'ical particularities of the human auditory system& thus pro#idin' an accurate
idea of ho/ the sound is Nmakin' senseO to the rain0 These descriptors could ha#e any
#eral laels ut their 2uantifiale result makes them e3cellent in terms of reliaility
ecause the measurement /ill al/ays e the same for a particular sound sample0 These
measurements represent the est solution to make connections et/een data e3tracted
from an audio si'nal and eha#ioural or neural data 'athered from people& ecause they
pro#ide 2uantifiale 'round to formulate statistical inferences0 <ith this computational
tools /e elie#e that the old prolem of non homo'eneous and sujecti#e description of
timre is 'radually startin' to disappear& and is lea#in' in its place ne/ attracti#e
prolems related /ith the perceptual sutleties of timre0
2 the year of this reference has been changed from 1986, because it is not clear which is the specific
work Huron is pointing at.
18
'%'%" Timbre an# scales
'%'%"%" S+ectral sha+e
To concede that timre can e computed from a sound si'nal implies that /e
posses a special /ay to represent the sound phenomena /hich is useful& amon' other
thin's& to apply different forms of analysis on it0 This representation also kno/n as
di'ital sound is essentially an arran'ement of inary ciphers encodin' a sound /a#e0
Althou'h there is a standard principle& /hich consists in encodin' the situation of a
'roup of particles in small portions of time and space& there has een a considerale
increment in the amount of mathematical al'orithms that seek to achie#e this principle
in a more efficient /ay0
Gy ha#in' a representation of a sound /a#e in a i dimensional form& /e can fi'ure out
that the /ider the /a#e means more displacement of particles and so more ener'y
1
& and
the amount of repetitions of a /a#e in a fi3ed time span is related /ith the fre2uency of
that sound Bor pitchC0 Gut this analysis of periodicities of a /a#e only /orks in an ideal
scenario& ecause the truth of music relies in the comple3ity of its /a#es& in such a /ay
that if /e /ant to 'o eyond the superficial kno/led'e aout #olume or pitch& it
ecomes necessary a more detailed analysis& one that perhaps y lookin' for
periodicities in the superficial periodicities /ould re#eal a three or multidimensional
form to represent the sound0 This prolem of decomposin' a comple3 /a#e into the
sum of its simpler components /as sol#ed y @ean-Gaptiste @oseph Fourier B5+-.-
5.1*C& and this decomposition applied to sound is kno/n as spectral analysis0
"n a detailed analysis of the spectrum of a sound& it can e oser#ed that there are
certain components that has more po/er than others& such components are referred as
principal components of the spectrum& and the distriution and differentiated po/er of
these components are uni2ue for each sound& ut some part of this components remain
unchan'ed in sounds that come from the same source& thus dra/in' a certain shape0
Timre could e re'arded as this non-#ariant shape dra/n y the principal components
in the spectrum0 "n other /ords& timre can e found in the spectrum of a sound ut this
3 or volume of particles displaced
19
imply that the spectrum /ill not contain rele#ant information concernin' other 2ualities
of that sound& like pitch for e3ample0
This is actually a confirmation for the predictions of @ean =hillipe Rameau B5.-1-5+-7C&
/ho askedE N9an it really e that /e hear three sounds e#ery time /e hear oneHO& in
relation to the spectral components he could hear& and se#eral of his collea'ues could
not0 $r it /as not the case that they /ould not really hear them ut as a matter of
perceptual 'roupin' they heard not the principal components& ut the timre itself
BFales& )**8C0
'%'%"%' !onsonance 2 3issonance
The special se'mentation done y Rameau made him suspicious aout the role
of those components for the entire harmonic system as it /as concei#ed at his time& and
the /ay the rules of composition had een settled alon' years of e#olution as a
conse2uence of constant e3perimentation in the 2uest for an aesthetic ideal0 This
disco#erin' made him and his successors to ha#e a ne/ consciousness aout the inner
structure of a sound& thou'h this could not e pro#en y scientific methodolo'ies until
the time of Helmholt! B5.)5-5.;7C0
"n his /ork Die Lehre von den Tonenmfindungen, Helmol!t estalished the
mechanical and physiolo'ical asis for the concepts of consonance:dissonance& y
analysin' the phenomena of soundin' t/o tones simultaneously& keepin' one at a fi3ed
fre2uency& and soundin' the second at different fre2uencies0 "n that /ay he claimed that
the difference et/een consonant and dissonant phenomena /as related /ith the
difference of their fre2uency& ut specifically /ith the sound produced y the
comination of the spectral shape of one tone /ith the spectral shape of the otherD if the
principal spectral components
7
of one sound resemled the other under a certain
threshold B11 H!C& then the inter#al
8
could e considered dissonant& and if this
resemlance /as outside this threshold& and additionally had a numerical relation close
to an inte'er then inter#al could e considered as consonant0 This /as later studied and
rectified later y =lomp > Le#elt B5;-8C& /hom added that such threshold /as in fact a
4 also referred as overtones or partials.
5 see a definition in section 2.1.2.
20
cur#e /hich chan'es as a function of fre2uency& this cur#e is contained /ithin t/o
linear oundaries /hich they called critical and/idth0 The area et/een these lines is
/ider at lo/er fre2uencies and narro/er at hi'her fre2uenciesD N%imple-tone inter#als
are e#aluated as consonant for fre2uency differences e3ceedin' this and/idth0
<hereas the most dissonant inter#als correspond /ith fre2uency differences of aout a
2uarter of this and/idthO Bp0 87.C0
'%'%"%) Scales
There is one inter#al /hich is the most important ecause can e found in the
music of many cultures as a oundary for a scale B4rumhansl& 5;;*C& and in /estern
tradition has een considered the most consonant in second place just elo/ the unisonE
the octa#e BFu3& 5;--C0 "f t/o tones are soundin' in unison that means that the salient
fre2uencies of their spectrum are so similar that the relation of distance can e
e3pressed as 5E5& in other /ords& for the most salient fre2uency of the spectrum in one
tone& there is another soundin' at the same fre2uency in the other tone0 "n the case of
the octa#e the relation is e3pressed as )E5& ecause the salient fre2uency of the second
tone doules the salient fre2uency of the first0 This numerical relations are kno/n as
ratios& and this is a synonym of inter#al& in such a /ay that a scale could e seen as a set
of ratios of fre2uency /ithin an octa#e0 There mi'ht e as many scales as are lan'ua'es
in the /orld& ut the most popular scale used in /estern culture is one that di#ides the
inter#al of an octa#e in t/el#e e#en parts& its ratios can e otained y the formulaE
Ratio
i
=2

i
12

/ere i is an inde3 /ith #alues from * to 5) indicatin' the correspondin' sudi#ision of
the octa#e
-
& the traditional names for this t/el#e sudi#isions areE unison& minor second&
major second& minor third& major third& just fourth& tritone Bau'mented fourth or
diminished fifthC& just fifth& minor si3th& major si3th& minor se#en& major se#en and
octa#e0 This scale /ill e referred in this thesis as the t/el#e tone e2ually tempered B5)
tone e2ually temperedC scale& and for practical reasons this /ill e our reference scale0
6 Further subdivisions of the octave can be computed with the same formula by changing the number 12
for the desired number.
21
'%'%"%- Relations bet4een timbre5 consonance an# scale%
The concept of sensory consonance and dissonance is important to understand
ho/ timre is related /ith a musical scale& for instance the e3periment aout sensory
dissonance conduced y =lomp > Le#elt B5;-8C& that used of a pair of sinusoidal /a#es
one at a fi3ed fre2uency and the second increasin' the fre2uency& thus dra/in' a cur#e
represented in fi'ure 50 "n this fi'ure it is possile to oser#e that the peak of
dissonance happens /hen the inter#al is close to the minor second& and then slo/ly
decreases0 "t can e also oser#ed a lael for the different sensations that are e#oked in
this kind of phenomenaE t/o sinusoidal that are #ery close in fre2uency /ill e heard as
eatin' each other& /hen the eatin' is too fast& then a ne/ sensation of rou'hness is
percei#ed& and /hen they are more distant& the rou'hness Bor dissonanceC disappears
and /e start to distin'uish t/o tones0
igure !" #epresentation of $ensory dissonance by %lompt & Levelt '$ethares, !((()"
Fi'ure 5 also informs that there are no harmonics present& meanin' that there are no
other components in the spectrum of this sound& so it can e ar'ued that this is a #ery
particular timre /hich mi'ht e difficult to find in a musical conte3t0 %o /e /ould
like to kno/ /hat /ould happen if the e3perimental tones are not simple sinusoidals&
ut more comple3 tones& for instance a tone composed of si3 main spectral components
/hich are inte'er multiples of the fundamental fre2uency B)8* H!0CD the 'raphical
representation of sensory dissonance for that e3periment also run y =lompt > Le#elt is
sho/n in fi'ure )0 Here the line of reference in the hori!ontal plane is not the 5) tone
22
e2ually tempered scale& ut the fre2uency #alues& and an e3tra information is pro#ided
in the points of minima of the cur#e that indicate ratios of fre2uencies0
igure *" #oughness and ratios 'adapted from $ethares, !((()
These ratios are coincident also /ith an old musical scale calculated y =ytha'oras
BT8.* - T8** G9C& and /hich are also coincident /ith some of the steps of the 5) tone
e2ually tempered0 "n Tale 5& these #alues are taulated to illustrate this coincidences0
Table !" +onvergence of ratios from different sources"
%ethares B5;;;C pro#ides an al'orithm to compute a cur#e of perceptual dissonance
from a sound si'nal& so it is e3pected that from this approach most of the components of
the spectral shape are included in the computation B%ee fi'ure 1C& thus pro#idin' a etter
accuracy in terms of the relation that a particular timre /ould ha#e /ith a scale0 "n this
cur#es of perceptual dissonance& /hich %ethares calls just dissonance cur#es& the same
principle of fi'ure ) can e applied& /ith the difference that the e3act #alues from the
minima points can e otained ecause they are estimated throu'h an al'orithm0 Fi'ure
1 sho/s in the x a3is the fre2uency ratios& and the 'rid in the ack'round sho/s the
position of the inter#als of the 5) tone e2ually tempered scale0 The numerical #alues in
the y a3is are aritrary& ut the idea is to sho/ a measure for the percei#ed dissonance0
Musical Names
Unison just fifth octave
1 1.06 1.12 1.19 1.26 1.33 1.41 1.5 1.59 1.68 1.78 1.89 2
1 1.2 1.25 1.33 1.5 1.67 2
1:1 5:6 4:5 3:4 2:3 5:3 1:2
minor
second
major
second
minor
third
major
third
just
fourth tritone
minor
sixth
major
sixth
minor
seventh
major
seventh
12 - tet
Pythagorean
ratios
23
igure ," Example of dissonance curve computed -ith $ethares. algorithm
This e#idence su''ests that in the e#olution of scales from the pure aesthetic
perspecti#e& it has een al/ays an intuiti#e and perhaps unconscious influence rou'ht
y the physiolo'ical attriutes of the auditory system0 Gy manipulatin' dissonance
cur#es /ith different settin's it is possile to understand that dissonance:consonance is
only one approach of many possiles& in fact in section 10501050 a numerical approach
used for our e3periment is e3plained0 The relations that can e estalished et/een
timre and scales are infinite& ut depend stron'ly in the le#el of resolutionD y
considerin' only fe/ components of the spectral shape& /e 'et only fe/ sudi#isions of
the octa#e& ut if our le#el of resolution is hi'h and /e use in the computation each
salient periodicity found in the spectrum of a 'i#en timre& the numer of consonant
points increases dramatically0 This mi'ht e 'ood reason for ans/erin' /hy there is a
sustantial amount of musical scales in the /orld& ecause the le#el of resolution for
e#ery culture is different& and perhaps this resolution is related /ith other areas of
auditory perception as for e3ample those related /ith human speech sounds B%ch/art!&
Ho/e& =ur#es& )**1C0
24
2.3 Memory
'%)%" three sta1es in one
"n this section is riefly re#ie/ed a three sta'es model presented y Go %nyder
B)**5C in his ook entitled NMusic and MemoryO0 This model di#ides the memory in
three main sections accordin' to their capailities to process information /ithin certain
time span0 These are the sensory& short-term and lon'-term memory0 This ta3onomical
approach does not imply that the rain is actually processin' information in three
different parts& ut rather those three con#er'e in one referred as /orkin' memory0
'%)%"%" Sensor6 memor6
"n the sensory memory Balso called the echoic memoryC& the auditory
information is or'ani!ed in a #ery asic /ay0 The input consists of impulses from ner#e
cells produced in the ear and each of the KfeaturesL like pitch and spectrum is e3tracted
y a 'roup of neurons that are speciali!ed in a iolo'ically si'nificant form to respond
to that particular specific feature0 These e3tractors may e estalished 'enetically
ecause they appear to e innate and no learned or different for different species0
%ensory memory is the most asic kind of auditory cate'orical perception B6#ans&
5;.)C0
At this point the information ecomes cate'orical& /hich means that it is no lon'er a
continuous sensory representation0 An auditory e#ent is a asic form of association& and
it occurs /hen particular features occur to'ether& for e3ample the perception of a note
duration& timre and pitch0 "n the echoic memory some asic non-#eral representation
Bima'in'C happens& and also the matchin' of lon'-term memory content to current
perceptual e3perience called pattern reco'nition0 Haituation is a special form of
reco'nition that occurs at a less conscious a/areness le#el0 "t is a phenomena that
makes the output of the neurons less acti#e o#er time& /hen their input is repeatedly an
identical stimulus B%nyder& )**5& p0 )7C0 This decreasin' of acti#ity is also called
adaptation responseD in other /ords& if a the same impulse like an amient noise is
25
repeated o#er and o#er a'ain& it is unlikely that e#ery ne3t repetition attracts our
attention0
'%)%"%' Short7term memor6
%hort-term memory lasts from three to fi#e seconds on a#era'e& dependin' on
the no#elty and comple3ity of the material to e rememered0 "t differs from the lon'-
term memory in that does not cause permanent anatomical or chemical chan'es in the
connections et/een neurons0 "t is a type of memory process /hit a certain de'ree of
speciali!ation& so there is proaly more than one short-term memory& for instanceE for
lan'ua'e& for #isual oject reco'nition& spatial relations& non- lin'uistic sounds and
physical mo#ement Bp0 7+C0
%hort-term memory has a close link /ith t/o conceptsE rehearsal and chunkin'0
Rehearsal is necessary to maintain information temporarily as short-term memory& it is
also necessary to store information into lon'-term memory and is a consciousness
process& tou'h it can happen also in a less conscious /ay0 "n 'eneral terms& any
repetition of elements in a pattern of e3perience constitutes a kind of rehearsal0
9hunkin' refers to small 'roupin's Bof elementsC associated /ith each other and
capale of formin' hi'her le#el unitsD it is the consolidation of small 'roups of
associated memory elements and leads to the creation of structured hierarchies of
associations0 %hort-term memory is associated /ith the le#el of melodic and rhythmic
'roupin'& this is the le#el at /hich the UlocalL order of music is percei#ed Bp0 8)-8-C0
This types of memory are our primary /ay of comprehendin' the time se2uences of
e#ents in our e3perience& althou'h the capacity if this memory is #ery smallD +V)
elements /ith conscious rehearsal& 1 or 7 elements /ithout conscious rehearsal& )8
elements /ith repetition patterns that fit time limits of short-term memory0 "n terms of
time span& it #aries from 1 to 8 seconds to 5* to 5)& dependin' on the type of
information Bp0 8*C0
26
'%)%"%) Lon17term memor6
Lon'-term memory mana'es patterns and relationships et/een e#ents on a
time scale lar'er than 1 to 8 seconds0 Memories need to e unconscious in order to
lea#e room for the notion of present& they are thou'hts that are formed /hen repeated
stimulation chan'es the stren'th of connections et/een simultaneously acti#ated
neurons0 The connections et/een 'roups of simultaneously acti#ated neurons are
called associations& and this process is referred to as cuein'0 There are three types of
cuein'E recollection /hen /e intentionally try to cue a memoryD remindin' /here an
e#ent in the en#ironment automatically cues an associated memory of somethin' elseD
and reco'nition& /here an e#ent in the en#ironment automatically acts as its o/n cue0
Lon'-term memory is not at all static& ut hi'hly dynamic0 The creation of lon'-term
memories is often refereed to as Ncodin'O& rememerin' is a process of reconstruction
rather than reproduction Bp0 -;-+5C0
'%)%"%- 8or9in1 memor6
The process in /hich semi-acti#ated lon'-term memory ecomes hi'hly
acti#ated and conscious& and short-term memory fadin' in from perception and out Bto
prime ne/ memory associationsC is called /orkin' memory0 The conceptual difference
consists in& /hile short-term memory can e understood as stora'e& /orkin' memory
must e considered as a process0 <orkin' memory deals /ith immediate perceptions
and related acti#ated lon'-term memories& also /ith conte3tual information that resides
semi-acti#ated& ut not in consciousness& and also /ith information that had just een in
consciousness Bp0 7+-7;C0
'%)%' Memor6 for timbre
Amon' other information discerned at the fe/ milliseconds after ha#in'
e3posed to a sound& there is the timre0 "t has een e3perimentally dissociated from
pitch& /hich is elie#ed to e processed independently in auditory short-term memory
B%emal& Demany& 5;;5D 5;;1C0 %ome research has een focused in reco'nition& /here it
has een disco#ered a differential capaility to reco'ni!e a 'i#en timre /hen some
27
aspects of the spectral shape had een altered BGer'er& 5;-7C& and also /hen the
e3perimental task demands the reco'nition of timre under different conte3tual
situations B4rumhansl& 5;;)C0
$ther approach that seems #ery promisin' to study the effects of timre on memory& is
the neurolo'ical& ecause it has pro#ided e#idence of auditory cortical enhancement as a
possile result of trainin' /ith an specific instrumental timre B=ante#& Roerts& %chul!&
6n'elien& Ross& )**5C& and plastic chan'es in the auditory corte3 in fre2uency
discrimination e3periments B=ante#& <ollrink& Roerts& 6n'elien& LWtkenhöner& 5;;;D
Mennin'& Roerts& =ante#& )***C0
28
) *m+irical a++roach
This chapter contains a detailed description of the different sta'es in /hich this project
/as in#ol#ed to/ards the 'atherin' of data and the conse2uent testin' of statistical
inferences related to the prolem of the effects of timre on the recallin' of microtonal
inter#als0
3.1 Method
)%"%" 3esi1n
The desi'n is oriented to/ards the comparison of performance in memori!in'-recallin'
inter#als played /ith different timres& so it can e descried as a Repeated Measures
desi'n& /here the independent #ariales are %ynthesis type& "nstrument name and
"nter#al len'th& and the dependent #ariale is the de'ree of accuracy in recallin' a 'i#en
inter#al performed y a suject& measured in numer of errors0 The main test in#ol#es
analysis of #ariance of different 'roups of timral characteristics and ori'ins& ut in the
otained data is also possile to oser#e the tendency of the response& and to speculate
aout a possile learnin' function0 The decision of studyin' timre effects y askin'
sujects to focus on another task& /as made /ith the intention of estalishin' an
homo'eneous status in the /ay /orkin' memory is controllin' the attention amon' all
sujects BHall > Glasko& )**8C0
"n fi'ure 7& there is a 'raphical e3planation of ho/ is desi'ned the e3periment0 "n the
lo/er left corner of the fi'ure& there is a s2uare se'mented in )* parts& each little s2uare
represents a sound sample /ith t/o uni2ue 2ualitiesE one instrumental ori'in and one
type of timral modification B8 "nstrument types and 1 %ynthesis types plus no
modificationC0 The ri'ht side of this dia'ram sho/s three shaded ack'rounds laelled
as Gase Fre2uency& Tar'et& and %econd > third choice& /hich purpose is to illustrate that
there are three inter#als per trial& one that is made et/een the ase fre2uency and the
tar'et& and the other t/o et/een the ase fre2uency and the second and third choice0
29
"nside of these shaded areas there are s2uares similar to the one that is in the lo/er left
corner of the fi'ure& ut they are marked /ith numers that represent the tunin' of each
little s2uare0 %o for e3ample each little s2uare of the ase fre2uency is tuned at
)-50-1H!0& and )* of the -* little samples of the area shaded as tar'et are tuned at
)-50-1 H!0 plus 18* cents& other )* at )-50-1 H! plus 88* cents& and the rest at )-50-1
H!0 plus +8* cents0 The same lo'ics apply for the second > third choice area& in such a
/ay that there are )** different samples each one /ith a uni2ue 2uality0 The top left
area presents a summary of this0
igure /" Design of the experiment
)%"%' Partici+ants
A total of )1 sujects /ere recruited y three methodsE postin' in e-mail lists&
attachin' a paper to notice oards& and personal in#itations0 T/o tar'et 'roups /ere
considered durin' the composition of the in#itations& one /as the 'eneral pulic and it
depicted the rele#ance of hearin' skills in the appreciation of music& and the second /as
directed to musicians& in /hich /e formulated a delierate challen'e for the indi#iduals
ear trainin' ailities0 The case of the #eral in#itations included also a short
e3planation aout the test0 "n 'eneral terms& sujects /ere not informed aout the main
#ariale to e studied& /hich /as the timreD instead of that the test /as presented as a
pitch discrimination e3ercise0
30
There /ere . female and 58 males& /ith mean a'e of a'es et/een )* and 7* years old
BMean F ).0-& %D F 80)C0 They /ere asked to fill a 2uestionnaire after the e3periment
/hose purpose /as to collect information aout their musical ack'round and haits0
)%"%) Materials
)%"%)%" Stim$li
The desi'n of the stimuli in#ol#ed t/o main processesE selection:discrimination
of collections and samples& and analysis:synthesis0
Selection
For the selection of samples three sources /ere analysedE NMcAill Qni#ersity
Master %amples )0*O B$polko > <apnick& )**-C& N,ati#e "nstrumentsE 4ontakt )
%ample LiraryOBHa#er > %chmitt& )**-C& and a Microsoft M"D" /a#e-tale synthesi!er
played /ith a Realtek AL9)8; soft/are0 The ojecti#e /as to otain sounds /ith a
certain de'ree of ecolo'ical #alidity in terms of ho/ /idely are they used& so the first
one BMcAillC /as picked ecause it has een /idely used y researchers in
musicolo'ical studies& and also ecause durin' the de#elopment of this project /e /ere
in#ol#ed in another research that in#esti'ated the possile relations et/een timre and
emotions and used them as /ell0 The ,ati#e "nstruments option /as used ecause it
/as already a#ailale at the Music Department of the Qni#ersity of @y#äskylä& and also
ecause of the orientation of this product& /hich is directed to the home studio or semi
professional market& /hich means that samples are almost ready to e used for creati#e
purposes ecause of the e2uali!ation of loudness and classification of sounds& as /ell as
the soft/are interface to play themD this last 2ualities in contrast to the McAill collection
/hich needs a sustantial post processin' /ork efore it can e used0 The third option
/as selected ecause of its a#ailaility& popularity and use particularly in music trainin'
soft/are or /e sites that rely on it0
31
3iscrimination
For the discrimination of instrumental sets& the criterion /as first to find all
instrumental sets containin' the 51 samples of the central octa#e& M"D" -*-+) B97-98C0
"n second place and /ith the aid of M"RToolo3 BLartillot > Toi#iainen& )**+C a script
/as de#eloped in Matla /hich /e called NAnta'onist FinderO0 This script /as aimed
to analyse the samples& y computin' a 'eneral description of them& like duration&
amplitude and timral #alues0 This script helped to discriminate those samples that
/ere not /ell processed y the toolo3 ecause of particular comple3ities contained in
their structure0 For e3ample all the instrumental sets containin' samples /hose pitch
/ere under or ao#e estimated for more than a 5*P of error y the NmirpitchO al'orithm
Busin' default optionsC& /ere discarded0 Gy doin' this& the ojecti#e /as to find
realistic sounds /hich could e easily treated in posterior processin'0 Another strate'y
included in the NAnta'onist FinderO consisted in comparisons of specific timral
descriptors& to disco#er instruments /hose #alues are in the lo/er or hi'her ends of a
comparison tale0 For e3ample& in the initial sta'es of this project /e /ere particularly
interested in ri'htness& so y applyin' this strate'y on ri'htness& in a manner that a
pair of instrumental sets& one /ith a ri'ht sound 2uality and other /ith a dark sound
2uality could e included in the final e3perimental set0 This searchin' of opposites /as
also used /ith other descriptors& such as inharmonicity and rou'hness0
At the end& only fi#e instrumental sets /ere includedE French Horn& Alto Flute and
Go/ed ?iraphone from the McAill collection& an $oe from ,ati#e "nstruments and a
=iano from the Microsoft tale0 This final decision /as difficult to make& ecause /e
/ere lookin' for includin' at least one instrumental set per analysed collection& and
ecause for practical reasons in terms of duration of the e3periment& some timres that
had attracti#e and interestin' sounds had to e discarded0
nal6sis
"n his /ork& %ethares B5;;;C e3plains in detail a techni2ue to compute the ratios
of a ne/ scale that suits est to make music /ith etter consonance for a 'i#en timre0
"n the first part of this techni2ue is in#ol#ed the computation of the spectral shape of the
tar'et timre to e3tract the most rele#ant partials and its amplitudes0 Then& /ith that
32
information a dissonance cur#e is computed and it is ar'ued that the minimas of that
cur#e pinpoint the ratios /hich are components of that Nne/ scaleO0
Gy doin' these steps in a systematic form for a set of multiple timres& /e found that
the minimas for those timres are predominantly in the same ratios& thus 'i#in' 'round
for a 'eneric scale0 "n fi'ure 8& it is possile to oser#e a #isual pattern of the minima
hits for a set of +1 samples from the McAuill 9ollection0 6ach sample has a different
timre& all of them /ith a duration of 5 sec0 and tuned at DX7 T 155 H!0 The 'rid in the
ack'round depicts the location of the steps of the 5) tone e2ually tempered scale steps&
just for reference0
igure 5" Minima hits from 0, dissonance curves
The pattern of incidence of minima points for +1 dissonance cur#es on particular ratios
re#ealed in a #isual /ay in fi'ure 8& leads to the ne3t representation Bfi'ure -C& /hich is
the cur#e resultin' from computin' the mean of those +1 cur#es& and the 'reen circles
represent the steps chosen as ratios for a Nne/ scaleO that /ill e referred in this project
as the Dissonance 9ur#e %cale0 This mean cur#e sho/s a total of 5- points&
ne#ertheless some of them are not considered for further computations ecause in some
cases they /ere almost o#erlapped in terms of fre2uency0 ,ote that the lael in the
hori!ontal plane in fi'ure -0 contains the e3pression 5)-tet& /hich is an are#iation of
t/el#e tone e2ually tempered system0
33
igure 1" Mean dissonance curve for a set of 0, different timbres
At this point& y ha#in' the inter#als from the Ndissonance cur#e scaleO& and y
assumin' that these collection of radios represented a 'eneric scale computed from a
psychoacoustic relation /ith timre& the ne3t step /as to select /hich of these ratios
/ould e a 'ood asset to test memory /ith them0 Gut in this intend /e learned that
dissonance cur#e scale /as just one approach and possily not the est& and ao#e all&
/e needed some strate'y to discard most of the inter#als& ecause other/ise the
duration of the e3periment /ould e too lon'& thus not acceptale to our needs0
Durin' this selection of inter#als& /e aimed to discard all the inter#als that /ere #ery
close to the traditional 5) tone e2ually tempered& under the rationale that this
information /as already in the memory of our sujects& ecause of their cultural
ack'roundD the intended task /ould demand from them the memori!ation of some
inter#al they /ere not really familiari!ed /ith0 $ther ojecti#e of this selection process&
/as to assure the inclusion of those inter#als that /ere actually present in the spectral
shape0 %o another strate'y /as used& /hich consisted in calculatin' the ratios from
adjacent partials y simple di#ision of their fre2uenciesD this strate'y resulted
prolematic ecause the resultin' ratios for a particular timre& /ere #ery different from
the results for another& ne#ertheless three ratios appeared in a consistent form amon' the
34
same +1 timres used for the estimation of the mean dissonance cur#e& and /ere
considered useful for our purpose0 Gut from this former strate'y& 'ettin' only three
ratios /as not really satisfactory& so /e decided to incorporate a third set of inter#als&
/hich had already ein' used in music ut in a not so common or popular matter0 Gy
e3plorin' the music and ideas of authors like 9harles "#es B5.+7-5;87C and @uliIn
9arrillo B5.+8-5;-8C& amon' others& /e found that they had used a )7 tones e2ually
tempered scale& so this represented another set of inter#als /ith a hi'h de'ree of
ecolo'ical #alidity0
<ith these three sets of inter#als in mindE dissonance cur#e scale& adjacent partials
consistent ratios and 2uarter tone scale Bor )7 tones e2ually temperedC& the final decision
of /hich inter#als /ould e used& /as made y comparin' the taulated fre2uencies0 "n
the comparison it /as found that some inter#al distances /ere elo/ or #ery close to
the @ust ,oticeale Difference cur#e BJost > ,ielsen& 5;.8C0 The su-tales in tale )&
sho/ the #alues used to make these comparisons and those inter#als selected as tar'ets
appear hi'hli'hted0 Also in the lo/er left corner of the fi'ure there is a tale of
components of the t/el#e tone system and its M"D" #alues just for referenceD note that
all the fre2uencies in each scale are taulated for the central octa#e B97 to 98C0
Table *" Tabulated fre2uencies for interval selection
1 0 261.63 1 0 261.63 1 0 261.63
1 7.67 269.29 3 5.37 299.04 5 1.76 331.39
2 0 277.18 4 5.81 305.32 6 7.59 356.82
2 8.12 285.3 4 2.82 313.95 9 11.1 404.21
3 0 293.66 4 8.58 319.71 13 10.37 512.88
3 8.61 302.27 5 2.6 327.03
4 0 311.13 5 6.82 336.45
4 9.12 320.24 6 0.48 348.75
5 0 329.63 7 10.26 359.74
5 9.66 339.29 7 3.72 366.28
6 0 349.23 8 0.44 392.44
6 10.23 359.46 9 3.56 418.86 Steps MIDI Hz
7 0 369.99 10 3.87 436.13 1 60 261.63
7 10.84 380.84 11 8.32 457.84 2 61 277.18
8 0 392 11 4.76 470.93 3 62 293.66
8 11.49 403.48 13 0 523.25 4 63 311.13
9 0 415.3 5 64 329.63
9 12.17 427.47 6 65 349.23
10 0 440 7 66 369.99
10 12.89 452.89 Target Intervals 8 67 392
11 0 466.16 MIDI HZ 9 68 415.3
11 13.66 479.82 63.4855 319.98 10 69 440
12 0 493.88 65.4620 358.67 11 70 466.16
12 14.47 508.36 67.5155 403.84 12 71 493.88
13 0 523.25 68.8470 436.13 13 72 523.25
12tet ! 24tet 12tet ! "#ss $#n 12tet ! a"%. part#als
12tet
steps "#&&.
24tet
&re's.
12tet
step "#&&.
D#ss.
(re's
12tet
step "#&&
)"%a*.
+art#als
12tet
35
"nitially& four inter#als /ere chosenE three ecause of its incidence on the three scales
and the fourth one ecause /e oser#ed its importance for the dissonance scale0 This
fourth one& /as the third most consonant ratio in the dissonance cur#e scale Bas can e
oser#ed in the fi'ure -C& ut ecause of its closeness to the Major -th from the 5) tone
e2ually tempered scale& it /as discarded0 Furthermore& as it can e oser#ed in the little
tale located in the central lo/er ed'e of tale ) headed as NTar'et inter#alsO& /hich
contains the a#era'e of coincident #alues& the M"D" #alues re#ealed that the three first
#alues are almost in the middle of an inte'er M"D" #alue& /hich means that if the
distance from one M"D" inte'er to the follo/in' is half tone in the 5) tone e2ually
tempered system& three of our tar'ets /ere really close to the 2uarter of tone and one
not0 Thus the final decision of tar'et inter#als /ere threeE 18* cents BM"D"F -107;C&
88* cents BM"D"F-807-C and +8* cents BM"D"F-+08)C0
A minor compromise must e declared& /hich consisted in the roundin' of this #alues to
one decimal in a manner that at the end the shorter inter#al in#ol#ed in this research
/ould e the 2uarter of tone of the 5) tone e2ually tempered system0
S6nthesis
<ith the aid of the %pectral Modellin' Toolo3 B@ohnson& )**)C& a sound si'nal
is processed /ith a %hort Time Fourier Transform& and then the output is made of three
#ectors containin' the #alues of fre2uency& amplitude and phase of the sound si'nal as a
function of time0 From this operation it is possile to identify the peaks in the #alues of
fre2uency and their e#olution on time& /hich& accordin' to the idea of Sa#ier %erra
B5;;*C& can e accommodated in tracks0 Those tracks can e seen as the partials
+
of a
tone and their small fluctuations B/ithin a 'i#en thresholdC of pitch in the e#olution of
time0
"n our hypothesis the presence or asence of particular partials /ould lead to an
enhanced or impaired e3traction of cues for memory0 "n order to make those particular
partials to e NasentO& an adjustment must e done& /hich starts y assumin' that each
partial of a 'i#en tone has a numerical relation /ith each step of a 'i#en scale0 %o the
7 along the whole text the expressions: overtone and main spectral components, are used as a synonym
of partials.
36
ojecti#e of the adjustments /as to take out of the partial structure those partials that
/ere numerically related to the tar'et inter#alsD in other /ords& to remo#e the inter#als
/hich /ere 'oin' to e used in the memory trials& from the inner structure of the tonesD
thus applyin' a modification on timre0
The method consisted of di#idin' the mean fre2uency of each track B p C y the
fre2uency of each step of the tar'et scale B f C0
ratio
ij
=
p
i
f
j
From this operation /e otain a matri3 of ratios /hich e3presses the pro3imity of each
pair of dataD the first ro/ and column /ould e3press the numerical relation et/een the
first partial and the first step of the scale& the second column of the same ro/ /ould
e3press the relation et/een the first partial and the second step of the scale& and so on0
Thus if each ro/ represents a numerical relation et/een that partial and all the steps of
a scale& the only thin' that remains is to identify the est candidate& /hich can e found
y lookin' in the fractional part of those numerical relations& for the #alue that is closer
to !ero0 For e3ample in the tale 1& /e can e3amine the first 5* partialsBtracksC of a
piano tone in 97T)-50-1H!0& and the numerical relations it has /ith the fre2uencies of a
5) tone e2ually tempered scaleD minimal remainders per ro/ are hi'hli'hted0
Table ," Example of selection of candidates
From this tale& /e can say that partials numers 5& )& 8 and + are related to the root
fre2uency& or in this case& to the first tone of the 5) tone e2ual tempered scale0 From
that data otained /e make a list of candidates to make the timral modifications on
them& /hich is usually ao#e 8*0 ,e#ertheless dependin' of ho/ ener'y is distriuted
&1 &2 &3 &4 &5 &6 &7 &8 &9 &10 &11 &12
p1 1.000 0.944 0.891 0.841 0.794 0.749 0.707 0.667 0.630 0.594 0.561 0.530
p2 2.999 2.831 2.672 2.522 2.380 2.247 2.121 2.002 1.889 1.783 1.683 1.589
p3 10.145 9.575 9.038 8.531 8.052 7.600 7.173 6.771 6.391 6.032 5.694 5.374
p4 14.438 13.628 12.863 12.141 11.460 10.817 10.209 9.636 9.096 8.585 8.103 7.648
p5 1.999 1.886 1.780 1.681 1.586 1.497 1.413 1.334 1.259 1.188 1.122 1.059
p6 5.002 4.722 4.457 4.206 3.970 3.747 3.537 3.339 3.151 2.974 2.807 2.650
p7 6.036 5.697 5.378 5.076 4.791 4.522 4.268 4.029 3.803 3.589 3.388 3.198
p8 12.254 11.566 10.917 10.304 9.726 9.180 8.665 8.179 7.720 7.286 6.877 6.491
p9 13.347 12.598 11.891 11.224 10.594 9.999 9.438 8.908 8.408 7.936 7.491 7.070
p10 4.004 3.779 3.567 3.367 3.178 2.999 2.831 2.672 2.522 2.381 2.247 2.121
37
in the spectrum it could reach a ma3imum of 8** Bthis threshold is specified in the
function that con#erts peaks to tracks from the %MToolo3C0 At this point is /orth to
mention that early intends on this matter of modifyin' timre& /ere aimed at displacin'
partials from its ori'inal positionD to stretch them y chan'in' their main fre2uency0
The idea /as aandoned ecause the audile result /as too far from the ori'inal sample
in terms of 2uality0
$nce the list of candidates has een otained& it is possile to ha#e a representation of
the partial structure& /hich comined /ith the amplitude #alues leads to another
#isuali!ation of the relation et/een timre structure of a 'i#en sound and a particular
scale0 Fi'ure + illustrates this idea& as the structure of this sample of a o/ed
#iraphone tuned in 97T)-50-1H!0& sho/s that its timre is al/ays related /ith the first
tone of these three scales B5) tones e2ually tempered& dissonance cur#e scale and )7
tones e2ually temperedC& and in second place /ith some step in the middle& /hich is the
inter#al kno/n as perfect fifth or )E1 ratio0 The scale in the #ertical a3is is aritrary ut
it is the result of addin' the #alue of amplitude of each partial that had an almost inte'er
relation /ith a 'i#en step of a scale Bcalled candidates in the upper para'raphC 0 The
fact is that y usin' more sudi#isions in one scale& that resolution sho/s more detail in
terms of relations et/een components of the spectral shape and the precise steps of a
'i#en scale0 This matter of resolution /as riefly discussed at the end of the section
)0)05070
igure 0" #epresentation of timbre in relation -ith the steps of different scales
38
"n the desi'n of the samples t/o synthesis techni2ues /ere usedE additi#e
synthesis and spectral modellin' synthesis0 For the first& t/o strate'ies of timral
modification /ere follo/ed& one doin' the synthesis usin' a copy of the spectral
information& fre2uencies and amplitudes from the ori'inal sample& and the second
discardin' those partials that /ere related numerically /ith the fre2uencies selected
from the mean dissonance cur#e Bcomputed for +1 different timres and e3plained in the
e'innin' of this section and also found hi'hli'hted in the su tales of Tale )C0 "n the
case of the spectral modellin' synthesis the strate'y /as different& ecause the timral
modification consisted in the cancellation of the desired partials y lo/erin' their
amplitudes0 This /as possile /ith one script from the %MToolo3 that calls an "n#erse
%hort Time Fourier Transform& that reincorporates the information aout fre2uencies&
amplitudes and phases into a sin'le time-fre2uency sound si'nal0 That ar'ued
cancellation Bor timral modificationC is easy to #isuali!e in the same kind of
representation used in fi'ure +& as demonstrated in fi'ure .0
igure 3" Example of adjustment of timbral structure
This particular e3ample sho/s the actual timral modification applied in the distriution
of partials of a o/ed #iraphone in 97T)-50-1H!00 Here is possile to oser#e three
39
NholesO at 8th& ;th and 51th steps& /hich correspond to the inter#als of 17;& 87- and +8)
cents of the scale e3tracted from the mean dissonance cur#e0
The final set is made of three different synthesis strate'ies plus the ori'inal sample&
/hich conform four sets of timres that in suse2uent sections /e refer as Nsynthesis
typeO #ariale0
All samples /ere e2uali!ed in loudness and con#erted from t/o to one channel& they
had a duration of 5)** ms and a linear en#elope for the first 5* ms0 and the last )**ms0&
/hich purpose is to take the amplitude from silence to the normali!ation le#el and then
ack0 For this processin' the soft/are named N%o3O y Ga'/ell B)**-C /as used0
The dia'ram of fi'ure ;& is a 'raphical summary of /hat has een /ritten aout the
desi'n of the stimuli in terms of ho/ a sound si'nal has een transformed in its timral
domain0 "n this fi'ure there are t/o shaded areas in the ack'round that represent the
t/o principal processes in#ol#ed& one that is the analysis of the timral structure& and
the second represents the synthesis of the different types of #ariale0 Dra/n o#er these
t/o main areas there are darker shaded rectan'les& /hich represent sta'es of those
process& and the inner o3es contained in them refer to computational al'orithms0 The
arro/s sho/ the direction of the flo/in' of informationD in a manner that from a sound
sample there are three outputs manipulated and one that is preser#ed as a copy of the
ori'inal0 The ellipses containin' the e3pressions aBtC& fBtC and pBtC are there to indicate
the e3istence of three #ectorsE amplitude& fre2uency and phase& as a function of time&
respecti#ely0 The are#iations Ns2O& NadO& NaoO and NorO are used as laels to simplify
the entire analysis:synthesis process0 Ho/e#er& the rationale eyond Ns2O is that y
usin' spectral modellin' synthesis& a timral modification in#ol#in' :uarters of tone
has een appliedD meanin' that the partials numerically related /ith the steps .& 5) and
5-th of a scale of )7 tones e2ually tempered& had een taken a/ay from the ori'inal
timral structure y lo/erin' to !ero the amplitudes of those specific partials0 The
are#iation NadO means that y usin' additi#e synthesis& a timral modification has
een done y usin' only those partials that are not related numerically /ith the steps 8&
; and 51th of the #issonance cur#e scale0 The are#iation NaoO means that the #ectors
40
of amplitude and fre2uency of the ori'inal sample had een copied to e added as
sinusoidals0 The lael NorO refers to a simple copy of the ori'inal sample0
igure (" $ynthesis types
"n order to otain the tar'et inter#als from the ori'inal samples /e used an audio
processin' utility named N%oundstretchO y =ar#iainen B)**8C& on the samples that /ere
closest to the tar'et0 For e3ample to otain the M"D"B-108CT1)*0)7H!0 /e used
NsoundstretchO on the M"D"B-1CT155051H! and M"D"B-7CT1);0-1H!0 to transpose up or
do/n respecti#ely& then they /ere compared y ear and one of them /as discarded y a
criteria of sujecti#e Nest 2ualityO et/een oth and also y lookin' for timral
homo'eneity /ithin all the instrumental set0
41
)%"%)%' ++arat$s
To 'ather the eha#ioural data a special patch /as desi'ned in the =ure Data
'raphical pro'rammin' en#ironment B=uckette& )**-C& runnin' on <indo/s S=0 <ith
the aid of this patch& /hich user #isual interface can e oser#ed in fi'ure 5*& the
responses of the sujects /ere collected0 They used a 'eneric mouse to click on some
choice o3es0 The stimuli /ere played throu'h the Di'idesi'n Mo3 ) sound interface
/ith A4A 575 %tudio headphones0
igure !4" 5ppearance of the experimental interface
<ith respect of the #isual interface& it must e pointed out that the desi'n /as aimed to
e simple to the user& and for that reason& it offered detailed instructions at the ottom of
the screen0 Furthermore& it had only one utton to operate the entire process& apart from
the s2uares of selection and the initiali!in' utton0 $ther aspect /hich recei#ed special
attention durin' the desi'n is related /ith the impossiility of 'ettin' errors in the
output data deri#ed from the e3periment& ecause all acti#e ojects had a special
eha#iour emedded in the pro'ram& for instance& the impossiility of continuin' /ith
the e3periment if the user had not made any choice or had taken more than one& as /ell
as for e3ample& tryin' to play a ne/ sound efore a special timer had sent a si'nal to
indicate that each trial had een completed successfully0
42
)%"%- Proce#$re
%ujects /ere asked to sit in front of a computer in a room desi'nated to make
#ideo conferences& /hich implies that the place has een conditioned to ha#e a relati#e
lo/ amient noise0 They /ere asked #erally if kne/ the meanin' of inter#al in a
musical conte3t& and if they do not& then recei#ed a rief e3planation aout it&
furthermore& they had to read an instructi#e Ba#ailale in Appendi3 AC& /hich purpose
/as to lo/er the possile an3iety& y in#ol#in' them in this e3perimental en#ironment
efore startin' the actual e3periment& and y 'i#in' also information in ad#ance aout
the proceedin'0
Durin' the e3periment& each e3ample inter#al /as follo/ed y other three /hich /ere
the selectale options0 There /as a silence et/een the presentation of the e3ample
inter#al and the three options of + seconds of duration0 6ach sound sample from the
dyad that formed the inter#als had a separation of 8* ms& and the separation et/een
one inter#al Bof the last threeC and the ne3t option /as +** ms0
The selectale inter#als /ere associated #isually to one 'reen circle that /as turned on
durin' the presentation of each option& to pinpoint an association to the o3 /here the
suject could make his:her choice0 The three options /ere al/ays presented #isually in
order from the left to the ri'ht& ut the audile reproduction /as in aleatory order& one
option /as a copy of the e3ample inter#al and the other t/o& /ere minus 8* cents or
plus 8* cents different from the tar'et0 %o& for instance& if the e3ample inter#al had a
measure of 18* cents& one of the options measured 18* cents& the second 7** cents& and
the third 1** cents 0 This #alues in cents represent al/ays a dyad /hich ase fre2uency
is 97TM"D"B-*CT)-50-1H!& for instance /hen /e refer 18* cents& the tones of the
inter#al are M"D"B-*CT)-50-1H!0 and M"D"B-108CT1)*0)7H!0
At the end of the e3periment& sujects /ere asked to fill a 2uestionnaire related /ith
their musical ack'round and 'eneral information as it is 'ender and a'e0
)%"%. Res$lts
9ollected data for each suject /as disposed in a matri3 of -* ro/s 3 +
43
columns& each ro/ contained the information for one trial& and the se#en columns /ere
dedicated as follo/sE the first three /ere used to record the characteristics of the
inter#als for the trial& in terms of its synthesis type& instrument name and inter#al len'th0
%ynthesis type could e one out of four options laelled asE (ad(& (ao(& (or( and (s2( Bthis is
e3plained in detail at the end of section 10501050CD instrument type could e one out of
fi#e options& laelled asE (afltn(& (frhrn(& (ooek(& (pnomi( and (#ir(& /hich means alto
flute& french horn& ooe& piano and o/ed #iraphone respecti#ely Bfurther e3planation
aout this instrumental sources could e found at the e'innin' of section 1050105CD and
inter#al len'th& /hich could e one out of three optionsE (18*(& (88*( or (+8*( cents0 The
fourth& fifth and si3th columns of each ro/ of the matri3& recorded the order of the
optional inter#al since they /ere al/ays presented randomly& so the numer * /as
desi'nated to the e3ample inter#al& the numer 5 to the e3ample inter#al de#iated y -8*
cents& and the numer ) to the e3ample inter#al de#iated y Y8* cents0 For e3ample if
columns fourth to si3 recorded the numers *& ) &5& that means that the first sample /as
the correct ans/er& the second /as the inter#al that /as one 2uarter of tone lon'er than
the e3ample inter#al& and the third one /as an inter#al shorter than the e3ample inter#al
for a 2uarter of tone0 The se#enth column recorded the option selected y the sujects&
this numer /as one out of three Bfrom * to )C representin' the location of the ans/er&
thus indicatin' if the suject had selected the inter#al represented in the columns 7& 8 or
- of the ro/0
"n a preliminary analysis of the data& four sujects scored out of the ran'e of the
inter2uartile& t/o of them elo/ the chance le#el and the other t/o ao#e the upper
decile0 Gecause of those reasons and in order to homo'eni!e the sample& those sujects
/ere considered as special cases and /ere discarded from the main sample0
The a#era'e accuracy for the main sample B5; sujectsC /as 8-08.P& sdF570)5P0 A
si'nificant correlation et/een years of musical in#ol#ement and accuracy /as found
BrF0-11)D p Z 0**8C0 ,e#ertheless this /as the only si'nificant relation that could e
estalished et/een the accuracy in responses and the 2uestionnaire data Bthis
2uestionnaire can e found in Appendi3 G0C0
%ince our interest /as focused in learnin' aout possile differences et/een the main
independent #ariales& /e ran a series of one-/ay A,$?A tests per independent
44
#ariale0 The case of the instrument type did not sho/ a si'nificant difference& ut in
the in the case of inter#al len'th FB)&8+CF107+D p Z 0*8& This result su''ests that is easier
to memori!e short inter#als than lon' ones Bsee fi'ure 55C0
igure !!" 5ccuracy per interval length
The case of synthesis type in#ol#es the main test of the e3periment ecause the central
hypothesis of this project ar'ued that y alterin' the structure of timre& the
performance of memory /ould sho/ a difference reflected in an impaired or enhanced
capaility to rememer the len'th of particular melodic dyads0 This /as tested /ith
satisfactory results& ecause the #ariale contained t/o susets that had not chan'es in
their partial structure& in other /ords t/o of the four susets could e considered as
Ncontrol setsO& ecause they maintained the ori'inal partial structure and in the other t/o
the spectral content suffered a modification in the timre y discardin' those particular
partials that had a numerical relation /ith the tar'et inter#als0 This /as re#ealed /hen
/e calculated a one-/ay A,$?A et/een the four susetsE (ad(& (ao(& (or( and (s2(0 and
otained an FB1&8-CF)0)+D pF*0*;*)& and then /hen the test /as ran for different
cominations0 Tale 7& sho/s these cominations and results0
45
Table /" Different combination of subsets to test synthesis type -ith one6-ay 57895
As sho/n in tale 7& the est proaility #alue /as otained /hen the comination (ad(&
(s2( /as tested FB5&).CF-0*1& pF*0*)*80 This result /as someho/ surprisin' ecause /e
/ere con#inced that y remo#in' some partials from the timre structure& the
memori!ation of inter#als /ould e impaired& ut althou'h oth synthesis outputs (s2(
and (ad( /ere aimed to/ards a remo#al of partials strate'y& one of them not only not
impaired the performance of memory ut on the contrary it enhanced the performance
in a si'nificant form /ith respect to the performance /ith the samples that preser#ed the
ori'inal partial structure0 The plot in fi'ure 5) sho/s means& #ariance and ma3imum
and minimum points for the results of accuracy of synthesis types0 Here is possile to
oser#e clearly that the means for the accuracy in the non-altered partial structures are
almost e2ual B(ao( and (or(C& /hile for those that had a modification in their timral
structure B(ad( and (s2(C sho/ different #alues0
igure !*" 5ccuracy per $ynthesis Type
ad or sq F p
, , 1-28 6.03 0.0205 .
, , 1-28 3.3 0.0800
, , , , 3-56 2.27 0.0902
, , 1-28 2.8 0.1054
, , 1-28 1.1 0.3026
, , 1-28 0.55 0.4664
, , 1-28 0.08 0.78
. p / .05
ao df
46
The ar 'raph in fi'ure 51& sho/s the accuracy per instrument type0 This is only for
illustrati#e purposes ecause as mentioned ao#e this #ariale did not sho/ any
si'nificant difference in the one-/ay A,$?A test0 ,ote that y a3is has a non-
continuous scale0
igure !," 5ccuracy per :nstrument Type
3.2 Discussion
The results reported at this point ha#e only ac2uainted the de'ree of accuracy in the
ans/ers0 ,e#ertheless in a second e3ploration of the data /e found that y attachin'
#alues to the cate'ories& /e could oser#e a tendency in the responses0 The #alues
attached /ere -5 to the ans/er that corresponded to the tar'et len'th minus 8* cents& a
!ero to the correct ans/er& and a Y5 to the inter#al that corresponded to the tar'et len'th
plus 8* cents0 "n that /ay /e could kno/ not only the a#era'e error& ut also the
tendency of the error0 Fi'ure 57 sho/s the responses per suject& each ar represents
one suject0
47
igure !/" Tendency of responses
Fi'ure 58 sho/s a clear asymmetry to/ards the shortest inter#al& that could e due the
direction of the inter#als used throu'hout the e3periment& /hich /as from the lo/er to
the hi'her pitch0 Another e3planation for this result could e addressed to the Aestalt
principle of pro3imity& /hich implies that shorter inter#als are etter processed than
lon'er inter#als BDeutsch& 5;;;C0
igure !5" ;istogram of responses
Furthermore& this tendency scale /as used to find differences et/een synthesis types&
and the results for one-/ay A,$?A /ere impro#ed if compared to the analysis of
48
#ariance of only correct ans/ers BFB1&8-CF)0-)D pF*0*8;+C0 Also y editin' manually
this data to remo#e outliers& the result chan'ed to sho/ an acceptale le#el of
si'nificance FB1&8-CF107D pF*0*)1.0
Ho/e#er /e reali!ed this tendency data /as risky to formulate inferences& ecause it
contained emedded some randomnessD this effect /as due the desi'n of the interface to
'ather the data& ecause there /as no means of lea#in' any option lank0 Thus sujects
/ere oli'ated to make a choice& and as they declared #erally after their e3perience&
they sometimes had to make a choice they kne/ in full consciousness it /as a random
choice0 Also they declared that these choices had to e done like this ecause durin' the
memori!ation process they ecame distracted /ith en#ironmental noise or ecause they
/ondered ho/ lon' the e3periment /ould least0 Althou'h this approach of tendency
mi'ht represent an interestin' possiility to dra/ implications for our central
hypothesis& no further analysis /as carried out /ith it ecause /e considered that some
aspects of the interface ha#e to e impro#ed to make this measure useful0 For e3ample
like tryin' the same e3periment in a completely noise isolated room& or y addin' a
countdo/n of trials in the interface& and proaly a NlankO choice0
)%'%" #a+tabilit6
"n this section /e use the term NadaptailityO a term taken from the neurolo'ical
literature to refer to a function of learnin' as a result of repeated e3posure to a certain
impulse0 %ince this thesis is also inspired y searchin' for etter methodolo'ies in Near
trainin'O& it /as important to determine if people actually learned the inter#als after
ein' e3posed repeatedly to them0 %o /e accommodated the data in such a manner that
the correct responses per inter#al /ere summed 'i#in' the numer of times that a
particular inter#al /as played for all the sujectsD each inter#al /as repeated )* times0
There could not e found a si'nificant re'ression coefficient in any of the inter#als&
/hich /ould indicate that from the first to the last repetition there /ould e any
impro#ement& ne#ertheless the 'raphical results are sho/n in fi'ure 5-0
49
igure !1 5daptability per interval
Gut in any case a tendency is difficult to determine on #isuali!ations like the one in
fi'ure 5-0 ,e#ertheless if /e add all responses across inter#als and normali!e the scales
y di#idin' the #ectors y its standard de#iations& it is possile to oser#e a sli'ht
increment in the correctness of the ans/ers as a slope in a linear function& /hich 'i#es a
standardi!ed re'ression coefficient of *0180 This procedure is dra/n in fi'ure 5+& /hich
also re#eals other interestin' information0 "f /e oser#e the form of this #ector of
Nlearnin'O& it can e found a couple of peaks durin' the second part of the repetitions
and then a descent in the a#era'e of accuracy0 <e su''est that this eha#iour mi'ht e
due an actual adaptaility function& and the follo/in' descent as proale e#idence for
haituation Bsee )010505& p0)1C0 Ho/e#er& this seems #ery interestin' as a future
direction for further research& since this su''estions could e tested y separatin' the
'atherin' to reduce the effect of one inter#al len'th o#er the other and throu'hout the
'atherin' of more trials per suject0
50
igure !0" 8verall 5daptability
)%'%' Timbre #escri+tors
"n section 105080 /e found some si'nificant differences in the eha#ioural data& /hen
the #ariances of the data 'rouped y inter#al len'th and synthesis type /ere tested0 %o
the ne3t step /as& y assumin' the certainty of those differences& to disentan'le their
ori'in despite of the methods follo/ed durin' the desi'n of the sound samples& /hich
passed for a presumed timral modification0
For that effect and /ith the aid of the M"Rtoolo3& 57 timre descriptors /ere
computed for each sound sample in#ol#ed in the e3perimentE !ero-cross& centroid&
ri'htness& spread& ske/ness& kurtosis& roll-of at ;8P& roll-of at .8P& entropy& flatness&
irre'ularity& lo/-ener'y& rou'hness& and inharmonicity0 <e are a/are of the fact that
there are more /ays to descrie timre& and that many of those can e easily computed&
ne#ertheless this set /as selected ecause the output of each descriptor /as a sin'le
numeric #alue0 From this computation /e otained a matri3 of )**357D )** samples
resultin' from the comination of 7 synthesis types 38 types of instrument 35* different
tunin's Bmore detail of this can e found in section 10505C0
The first step /as to or'ani!e the )** samples in four 'roups accordin' to their
51
synthesis types& thus otainin' four sets of data of 8*357& representin' the timral
description in 57 measures for 'roups of 8* samplesD one 'roup for (ad(& other for (ao(&
one more for (or( and the last for (s2( Bfor reference on these laels see section 10505C0
$nce this 'roupin' /as done& the ne3t step consisted in computin' a series of one-/ay
A,$?As for pairs of descriptors& to re#eal /hich descriptor of timre /ere responsile
for the difference et/een 'roups0 %o for instance& the #ector resultin' from estimatin'
8* #alues for ri'htness in the samples contained in the 'roup (ad(& /as compared /ith
the #ector resultin' from the 8* #alues of (s2(D thus otainin' the proailities of
differences and leadin' to the fi'ure 5.& /here those #alues can e oser#ed0
Hori!ontal a3is in fi'ure 5. sho/s are#iations for each one of the 57 descriptors
mentioned in the upper para'raph0
igure !3" 8ne6-ay 57895 bet-een timbres of t-o synthesis groups .ad. & .s2.
"n fi'ure 5.& it can e oser#ed y #isual e3ploration that most of the 57 descriptors
sho/ed si'nificant difference et/een them& so this procedure pro#ided only a startin'
point ecause /e /ere lookin' for/ard to discriminate most of the descriptors as
possile candidates responsile of the effects on memory0 Althou'h different
cominations of synthesis 'roups /ere tested in the same /ay& they sho/ed the same
silhouette of ars& so /e decided to keep this comination B(ad( > (s2(C for further
52
analysis ecause it had demonstrated the most meanin'ful difference in the eha#ioural
results Bp Z *0*)& see tale 7C0
Ackno/led'ed that there /ere meanin'ful statistical differences et/een particular
timral descriptors computed from sound samples /ith specific le#els of modification&
the follo/in' 'atherin' /as to learn if such modifications e3pressed in acoustical data&
had some relation /ith the responses 'i#en y the sujects in the e3periment0 To reali!e
this comparison& a simple correlation et/een the responses of the sujects and each
one of the descriptors /ould fit0 ,e#ertheless there /as one prolem& ecause the
matri3 of responses /as composed of -* ro/s BtrialsC& and the #ectors of descriptors had
)** #alues& one per sample in#ol#ed in the e3periment0 To sol#e this prolem& the set of
)** samples 3 57 descriptors had to e reduced& so this method consisted in computin'
means for each timre measure amon' all the pitches that /ere in#ol#ed in one trial&
thus computin' means for 'roups of pitchesE M"D"B-*C-B-1C-B-108C-B-7C& M"D"B-*C-
B-8C-B-808C-B--C& and M"D"B-*C-B-+C-B-+08C-B-.C0 <ith that method of reduction from
5* pitches to 1& and then multiplied y the numer of instrument typesB8C and the
synthesis typesB7C 'a#e 57 #ectors of -* #alues& /hich could then e correlated0
Furthermore& to check the reliaility of this reduction /e computed a'ain series of one-
/ay A,$?As for pairs B(ad( > (s2( synthesis typesC of descriptors& ut /ith this set of -*
#alues& and the results sho/ed almost the same pattern& e3cept for the spectral roll-off at
.8P and the spectral entropy0 Gy operatin' a reduction of this nature& /e interpret that
effects of pitch /ere also reduced& thus re#ealin' statistical differences et/een 'roups
sli'htly out of that dimension0
"n the left side of fi'ure 5; it can e oser#ed the same results of fi'ure 5.& ut /ith the
scale of the #ertical a3is offerin' a !oom0 The ri'ht side of the same fi'ure B5;C sho/s
the results of the same computation of A,$?As for the 'roups (ad(-(s2(& ut /ithin the
set of #alues that had een reduced to correlate /ith sujects responses0
53
igure !(" 8ne6-ay 57895 bet-een timbres of t-o synthesis groups, t-o data sets
An interestin' issue could e e3tracted from the matri3 of responses from the sujects&
ecause it had 1 columns containin' the numer of hits for estimationsD the first
contained the numer of responses that estimated the tar'et inter#al elo/ its actual fit&
the second& the responses that /ere estimated correctly& and the third the responses that
estimated the inter#al ao#e the actual si!e0 %o /e could compute correlations of
timral descriptors not only for the correct ans/ers& ut also for the mistaken
estimations0 Fi'ure )* sho/s the results of these correlations in the upper part& and the
p-#alues in the lo/er partD each descriptor is correlated /ith each #ector of responses&
one for the trials estimated elo/ the correct si!e of the inter#al& other for the correct
ans/er and the third one for the ans/ers that indicated an estimation ao#e the correct
si!e Bas e3plained in the section 10505 the inter#al si!es de#iate y 8* cents from the
tar'et inter#alC0 ,ote that in the lo/er 'raph a line has een dra/n to sho/ the line of
pF*0*80
54
igure *4" +orrelation bet-een behavioural and acoustical data
The results otained in the analysis of this section su''est that there are at least three
timral descriptors /hich #alues ha#e a si'nificant relation /ith the responses of the
sujects& and that are also present in the analysis of #ariances as responsile for the
differences et/een synthesis types0 %o far& the spectral centroid& the spectral spread&
and the spectral roll-off are present in oth 'atherin's0 %pectral flatness attracts also our
attention ecause it correlated hi'h in the responses that underestimated the si!e of the
inter#al0 These timral descriptors are presented in scatter plots in fi'ure )5& that also
sho/s the eha#ioural data 'rouped y synthesis types0
55
igure *!" $catter plots of relevant timbral descriptors and behavioural data
%pectral centroid is a measure of percei#ed Nri'htnessO and represents the mean #alue
of the fre2uency distriution& the spectral spread is a measure of the #ariance of the
fre2uency distriution& roll-off is a measure of spectral shape and represent the
fre2uencies under /hich certain percenta'e of the spectral ener'y is distriuted
BAucouturier& )**-C0 These three descriptors and their ne'ati#e correlation could e
interpreted as the most important cues for memori!ation of inter#alsD /here lo/er
#alues of any of them /ould lead to a etter performance in memory& ut such
56
affirmation should e considered in a ne/ e3periment since too many #ariales are
in#ol#ed in this e3perimental set& and /ould e sources of statistical noise0
%pectral Flatness is defined as the radio of the 'eometric mean of the po/er spectrum to
the arithmetic mean of the po/er spectrum& and is a measure of NnoisinessO B@ohnston&
5;..C0 The hi'h correlation found in the under-estimation #ector /ould su''est that
hi'h le#els of noisiness measured /ith spectral flatness /ould proaly e a source of
memory impairment0 ,e#ertheless& more empirical e#idence is necessary to make such
a claim0
57
- !oncl$sions
%ome of the results presented here are only partial and deser#e more attention
and further analysis0 For e3ample /e could not estalish a connection for the effects of
e#ery compromise& such as the decision to round the inter#als for the desi'n of the
e3periment that /as not applied durin' the analysis:synthesis of the stimuliD this
disparity could e responsile of some distortion in the results0 Also the fact that
accuracy per instrument did not sho/ a si'nificant difference from 'roups of instrument
types could e e3plained as an inconsistency in the e3perimental desi'n due the lo/
numer of cases per instrument0 $ther e3planation could e that the timres selected
did not ha#e enou'h contrast in those timral aspects that resulted rele#ant for the cuin'
of memory for inter#als& thus su''estin' a failure in the discrimination sta'e&
particularly in the script that /as de#eloped for this purpose /hich /e called
NAnta'onist FinderO0 ,e#ertheless a positi#e aspect are the results otained for the
synthesis type #ariale& /hich could e re'arded as synthetic instruments& and their
timral characteristics /ere used to alance the other inconsistencies0 Althou'h these
synthetic timres /ere chosen from a palette of at least 5* different methodolo'ies in an
almost aritrary /ay& just y e3plorin' #isually the spectro'rams of #ery fe/ samples&
they pro#ed to e useful in the timral descriptors analysis0
The e3perimental settin' had some aspects that must e impro#ed& such as a counter in
the patch& a properly sound isolated place& and the recruitment of more sujects0 $ther
thin' that could e impro#ed is the 2uestionnaire aout the musical ack'round& to
include information on other auditi#e ailities such as numer of spoken lan'ua'es0
The entire desi'n of this project is an intend to link the perspecti#es and concepts of a
traditional /estern academic minded musician /ith the conceptuali!ation of the nature
of music from a scientific perspecti#e& and the matter of this link has een the definition
of timre0 This effort represent in our understandin' a conceptual rid'e et/een a
macro and a micro perspecti#e of the nature of musicE a sin'le tone and the uni#erse
contained in it0
58
A rele#ant issue e3plored in this project is that selection of timre to memori!e inter#als
should e considered carefully for music educators in ear trainin' e3ercises& ecause it
mi'ht e that certain timres /ould optimi!e the task0 Ho/e#er this su''estion must e
supported y more empirical e#idence ecause there are many #ariales that are not
considered& such as indi#idual preferences0
Future directions for this research /ould consist in applyin' more comple3 trials& such
as harmonic dyads and triads& and short se2uences of tones or melodies& as /ell as the
role of timre and emotions in memory performance0
59
References
Aucouturier, J. J. (2006). Ten experiments on the modelling of polyphonic timbre, from
http://www.jj-aucouturier.info/papers/PHD-2006.pdf
Baars, B. J. (1988). A cognitive theory of consciousness. Cambridge University Press.
Bagwell, C. (2001). Sox - sound eXchange, from http://sox.sourceforge.net/
Bent, I. D., & Pople, A. (2006). Analysis. In L. Macy (Ed.), Grove music online.
Retrieved May 13, 2006, from http://www.grovemusic.com
Benward, B., & Kolosick, J. T. (2005). Ear training: A technique for listening. McGraw-
Hill Higher Education.
Berger, K. W. (1964). Some factors in the recognition of timbre. The Journal of the
Acoustical Society of America, 36, 1888.
Burns, E. M. (1999). Intervals, scales, and tuning. (pp. 215–264)Academic Press.
Choksy, L., Abramson, R. Gillespie, A., Woods, D., & York, F. (2001). Teaching music
in the twenty-first century. Prentice Hall Upper Saddle River, NJ.
Crowder, R. G., & Pitt, M. A. (1992). Research on Memory/Imagery for musical timbre.
In D. E. D. T. Reisberg (Ed.), Auditory imagery Lawrence Erlbaum Assoc Inc.
60
Deutsch, D. (1999). The psychology of music (2nd. ed.)Academic Press.
Evans, E. F. (1982). Functional anatomy of the auditory system. In H. B. Barlow, & J. D.
Mollon (Eds.), The senses (pp. 251-306). Great Britain: Cambridge University
Press.
Fales, C. (2002). The paradox of timbre. Ethnomusicology, 46(1), 56-95.
Fales, C. (2005). Listening to timbre during the french enlightenment. Paper presented
at the Proceedings of the Conference on Interdisciplinary Musicology (CIM05),
Montréal (Québec) Canada. Retrieved April 24, 2006, from
http://www.oicm.umontreal.ca/cim05/cim05_articles/FALES_C_CIM05.pdf
Fux, J. J. (1966). Gradus ad parnassum. Broude.
Gordon, E. (1988). Learning sequences in music: Skill, content, and patterns (1997th
ed.). USA: GIA Publications.
Grey, J. M. (1977). Multidimensional perceptual scaling of musical timbres. The
Journal of the Acoustical Society of America, 61, 1270.
Hall, M. D., & Blasko, D. G. (2005). Attentional interference in judgements of musical
timbre: Individual differences in working memory. The Journal of general
psychology, 132(1), 94-112.
61
Halpern, A. R., Zatorre, R. J., Bouffard, M., & Johnson, J. A. (2004). Behavioral and
neural correlates of perceived and imagined musical timbre. Neuropsychologia,
42(9), 1281-1292.
Handel, S. (1989). Listening: An introduction to the perception of auditory events. MIT
Press.
Hargreaves, D. J., & North, A. C. (2001). Musical development and learning: The
international perspective. Great Britain: Colin Brock.
Haver, D., & Schmitt , S. (2006). Native instruments kontakt 2 sample library
Helmholtz, H. (1954). On the sensations of tone as a physiological basis for the theory
of music(A. J. Ellis Trans.).Dover Publications New York.
Hofstetter, F. T. (1975). GUIDO: An interactive computer-based system for improvement
of instruction and research in ear-training. Journal of Computer-Based Instruction,
1(4), 100-106.
Huron, D. (2001). Toward a theory of timbre. Paper presented at the Music Theory Mid-
West Conference. New Theories for 20th Century Music, Cincinnati, Ohio.
Retrieved May 14, 2006, from http://www.music-cog.ohio-
state.edu/Huron/Talks/SMTmidwest.2001/index.html
62
Johnson, M. (2002). Spectral modelling toolbox
Johnston, J. D. (1988). Transform coding of audio signals using perceptual noise
criteria. Selected Areas in Communications, IEEE Journal on, 6(2), 314-323.
Krumhansl, C. L. Cognitive foundations of musical pitch. Oxford University Press, New
York and Oxford (1990).
Krumhansl, C. L., & Iverson, P. (1992). Perceptual interactions between musical pitch
and timbre. Journal of experimental psychology. Human perception and
performance, 18(3), 739-751.
Lartillot, O., & Toiviainen, P. (2007). MIR toolbox. Jyväskylä University.
Leman, M. (1995). Music and schema theory: Cognitive foundations of systematic
musicology. Springer.
McAdams, S. (2006). Psychology of music, §II,3:Perception & cognition of timbre. In
L. Macy (Ed.), Grove music online . Retrieved May 10, 2006, from
http://www.grovemusic.com
McAdams, S., & Bigand, E. (1993). Thinking in sound: The cognitive psychology of
human audition. Oxford; New York: Clarendon Press; Oxford University Press.
McAdams, S., Winsberg, S., Donnadieu, S., De Soete, G., & Krimphoff, J. (1995).
63
Perceptual scaling of synthesized musical timbres: Common dimensions,
specificities, and latent subject classes. Psychological research, 58(3), 177-192.
Menning, H., Roberts, L. E., & Pantev, C. (2000). Plastic changes in the auditory cortex
induced by intensive frequency discrimination training. Neuroreport, 11(4), 817-
822.
Opolko, F., & Wapnick, J. (2006). McGuill university master samples
Pantev, C., Engelien, A., Candia, V., & Elbert, T. (2001). Representational cortex in
musicians. plastic alterations in response to musical practice. Ann NY Acad Sci,
930, 300-314.
Pantev, C., Roberts, L. E., Schulz, M., Engelien, A., & Ross, B. (2001). Timbre-specific
enhancement of auditory cortical representations in musicians. Neuroreport, 12(1),
169-174.
Pantev, C., Wollbrink, A., Roberts, L. E., Engelien, A., & Lutkenhoner, B. (1999).
Short-term plasticity of the human auditory cortex. Brain Res, 842(1), 192-199.
Papoušek, H. (1996). Musicality in infancy research: Biological and cultural origins of
early musicality. (2003rd ed., pp. 37-55). USA.
Parviainen, O. (2005). Soundstretch
64
Pitt, M. A., & Crowder, R. G. (1992). The role of spectral and dynamic cues in imagery
for musical timbre. J Exp Psychol Hum Percept Perform, 18(3), 728-738.
Plomp, R., & Levelt, W. J. (1965). Tonal consonance and critical bandwidth. The
Journal of the Acoustical Society of America, 38(4), 548-560.
Puckette, M. (2006). Pure data
Rogers, N. (2005). Verbal labels affect memory for musical timbre. Paper presented at
the Proceedings of the Conference on Interdisciplinary Musicology (CIM05),
Montréal (Québec) Canada. Retrieved April 20, 2006, from
http://www.oicm.umontreal.ca/cim05/cim05_articles/ROGERS_N_CIM05.pdf
Russo, F. A., & Thompson, W. F. (2005). An interval size illusion: The influence of
timbre on the perceived size of melodic intervals. Perception & psychophysics,
67(4), 559-568.
Schafer, R. M. (1977). The tuning of the world. Knopf.
Schwartz, D. A., Howe, C. Q., & Purves, D. (2003). The statistical structure of human
speech sounds predicts musical universals. Journal of Neuroscience, 23(18), 7160-
7168.
Semal, C., & Demany, L. (1991). Dissociation of pitch from timbre in auditory short-
65
term memory. The Journal of the Acoustical Society of America, 89(5), 2404-2410.
Semal, C., & Demany, L. (1993). Further evidence for an autonomous processing of
pitch in auditory short-term memory. The Journal of the Acoustical Society of
America, 94(3 Pt 1), 1315-1322.
Serafine, M. L. (1988). Music as cognition: The development of thought in sound.
Columbia University Press.
Serra, X., & Smith III, J. (1990). Spectral modelling synthesis: A sound
Analysis/Synthesis system based on a deterministic plus stochastic decomposition.
Computer Music Journal, 14(4), 12-24.
Sethares, W. A. (1999). Tuning, timbre, spectrum, scale. Great Britain: Springer.
Shepard, R. (2001). Cognitive psychology and music. Music, cognition and
computerized sound: An introduction to psychoacoustics (pp. 21-35). USA: Cook,
P.R.
Slawson, W. (1985). Sound color. USA: University of California Press Berkeley.
Sloboda, J. A. (2005). Exploring the musical mind: Cognition, emotion, ability, function
Oxford University Press
Snyder, B. (2001). Music and memory: An introduction. USA: MIT Press.
66
Warrier, C. M., & Zatorre, R. J. (2002). Influence of tonal context and timbral variation
on perception of pitch. Perception & psychophysics, 64(2), 198-207.
Wessel, D. (1979). Timbre space as a musical control structure. Computer Music
Journal, 3(2), 45–52.
Yost, W. A., & Nielsen, D. W. (1985). Fundamentals of hearing.Holt.
Zagorski-Thomas, S. (2005). Shouting quietly: Changing musical meaning by changing
timbre with recording technology. Paper presented at the Proceedings of the
Conference on Interdisciplinary Musicology (CIM05), Montréal (Québec) Canada.
Retrieved April 20, 2006, from
http://www.oicm.umontreal.ca/cim05/cim05_articles/ZAGORSKI-TS_CIM05.pdf
67
Appendix A. Instructive of the experiment
8elcome to this test on microtonal +itch #iscrimination;;
Gefore proceedin' /ith the e3periment& please read carefully the follo/in' instructions0
5C %it rela3ed in the most comfortale /ay
)CAfter readin' this instructi#e& /ear the headphones and locate the #olume control&
1C The e3ercise /ill consist in the follo/in' stepsE
aC Jou /ill e asked for your name0 Do click in the field that is desi'nated for it&
althou'h nothin' chan'es #isually& if you do click on it then is acti#ated and you can
type in0 Don(t for'et to press Z6,T6R[ after typin' your name
C Jou /ill find three main di#isions in the /indo/& the di#ision at the ottom
/ill tell you /hat to do durin' the /hole process0
cC Durin' the e3periment& you /ill hear one inter#al Bt/o consecuti#e tonesC
follo/ed y a silence and then other three inter#als0 Jour ojecti#e is to identify /hich
one of the last three inter#als sounds e3actly the same as the first presented0
dC The e3periment has -* trials& so it is a 'ood idea if you don(t delay your
ans/ers for a lon' time0
7C <hen you reach the end of the e3periment& a screen /ill tell you and then your
responses are recorded0 The data /ill e analysed and your score /ill e sent /ithin
one /eek
8C To e'in the test do click in the utton at the ottom of this screen0
68
Appendix B. uestionnaire
Ear Trainig Test
MMT Thesis Project
General Questionary ID___
Age:________________
Gender:______________
Musical Background
Do you have any previous studies of music?
Music Theory ( ) Instrumental practice ( )
Professional(lessons) ( ) Hobbie (self-study) ( )
For how many years?(aprox.)____________
Which instrument(s) do you play? and for how many years?
1._______________________________( )
2._______________________________( )
3._______________________________( )
If you want to make some extra comment about your musical background write in this
space:
______________________________________________________________________
______________________________________________________________________
______________________________________________________________________
Musical Preferences
How many hours per week you listen music?
Attentive listening ________________ Unattentive listening _______________
Do you have a list of preferred musical genres? Could you write those that you like
most?
_____________________ _____________________ _____________________
_____________________ _____________________ _____________________
_____________________ _____________________ _____________________
Hearing Particularities
Do you have absolute pitch?_______________
Do you have or ever had a hearing problem?_______________