Week 7 Flashcards
Background in orchestration
The study of orchestration, relative to the history of western music, is relatively
recent. It was not until Berlioz’s “A Treatise of Modern Instrumentation and
Orchestration” (1844) that orchestration was presented as fundamental to the
structure of music. He attempted to provide a study of orchestration that was
independent and equal to studies of melody, harmony and rhythm. In the Romantic
Era, instrument designers established new methods for improving instruments, such as the addition of valves on horns and trumpets, and composers began to envision
novel combinations of instruments.
In the 20th century, Schoenberg’s concept of Klangfarbenmelodie referred to the
possibility of a succession of tone-colours related to one another in a way analogousto a relationship between the pitches in a melody. Klangfarbenmelodie inspired a
number of composers (including Stockhausen, Berio, and Schaeffer) to envision new possibilities for the realization of novel sounds in the electroacoustic medium.
Sandell article on orchestration
Good paper on timbre combinations
Gap in research opportunities. Lots to do
Orchestration treatise
A survey of major treatises reveals a persistent didactic tone with practical
considerations. Any kernels of what could be considered theory are hidden amongst
technical advice for scoring. The authors’ focus is on the “how” rather than the “why.”Further research in the pursuit of generalized principles in this area could benefit
composers and theorists alike.
Orchestration texts often amount to:
•
•
•
Factual observations about instruments (Didactic tone)
Prescriptions and prohibitions (practical)
Short musical examples from masters to be emulated
Need for organizing principles to achieve sonic goal
Study of orchestration
Disparity in scholarship
Lack of theoretical
F
Orchestration and auditory perception themes
Cssr Timbral blend, stream segregation Timbral intervals Role creating tension Solfege study
Role of blend
Mixing of timbres into perceptually fused event
Concurrent out ping and auditory image
Orchestrators combine instruments to sound simultaneously. This “melodic doubling”has been an important aspect of scoring instruments for ensembles.
The goal of blend is timbral fusion, which is related to principles of concurrent
grouping. Perceptual fusion gives rise to auditory events upon which timbre is
computed.
Multiple instruments can fuse to create a single auditory image, which is a
psychological representation of a sound entity exhibiting an internal coherence in its
acoustic behaviour. On a higher level, we can group sounds that form a composite
auditory image, a singular mental image from sounds that do not arise from a single
source (e.g., timpani strike plus a cello note and tuba note). A listener can determine
that these sounds come from multiple sources, but the musical context may lead the
listener to group these timbres into a single composite image.
Timbre blend and sonic goals, sandell paper
Three sonic goals
Sandell (1995) suggests that, in order to contribute to a theory of orchestration,
researchers should attempt to determine the sonic goal(s) orchestrators wish to
achieve, and then derive the principles that achieve that goal. He proposes 3
common sonic goals: timbral heterogeneity, timbral augmentation, and the creation
of an emergent timbre. (Last two deal with blend) These goals can be achieved through the presence or absence of blend.
Ligeti, 2nd Bagatelle for Wind Quintet – The use of different timbres allows one to segregate them perceptually and follow each one with ease (timbral
heterogeneity). Hear timbre independantly
Wagner, Parsifal (Overture) – The melody is presented by the cello, but the
cello timbre is augmented by instruments that appear at various times
throughout the melody in order to take advantage of their properties in differentregisters. One subservient timbre embellishes or highlights other
Ligeti, Atmosphères – The composer sculpts a large sound mass in timbre and register whose density thwarts attempts to recognize the constituent timbres.
Their combination thereby gives rise to an emergent timbral quality that
evolves over time
New timbre emerges from fusion of timbres
Studies of blend on musical contexts
Lower spectral centroid. Blend much better
Goodwin (1989) studied singers’ conscious efforts to blend with other choir singers.
He analyzed the spectra of singers when singing alone and when singing with a
recording of a choir where they attempted to make their own voice indiscernible
from the choir. The vowels sung in the group setting had stronger fundamental
frequencies (and first formants) and fewer and weaker upper partials (and weaker
upper formants) compared to vowels sung in the solo setting.
Kendall & Carterette (1993) studied the relationship between ratings of blend and the ability to identify woodwind sounds. When instruments were presented in concurrentpairs (dyads), listeners rated the degree of blend on a scale (oneness vs. twoness),
and in another study participants identified the instruments within the dyad. They
found an inverse relation between blend and ability to identify instrument sounds,
which suggests that the sounds that blend better are also more difficult to identify
separately in the mixture. Unisons blended significantly better than non-unisons, and identifications improved in the non-unison context.
Sandell blend studies
Sandell aimed to discover principles and generalized rules for selecting instruments that attain blend. He asked participants to rate the blend of all pairs of the 15
sounds from Grey (1977). He investigated the relationship between the acoustical properties of simultaneously sounding musical instruments and the psychological
judgment of blend.
Sandell used the blend ratings of all pairs of sounds from the Grey (1977)
timbre space as a measure of similarity and performed a multidimensional
scaling analysis. In the 2D solution, instruments that are close together blend
more than instruments that are far apart. If two instruments are close in space
(e.g., BN and S1), the degree of blend is rated as strong. If they are far apart (e.g. TP
and X2), the blending is weak and the sounds tend to be heard separately. The
dimensions of this “blend space” are moderately correlated with the attack time (x
axis) and strongly correlated with spectral centroid (y axis). This analysis
demonstrates the combined contribution of the two acoustic factors to
perceived blend, i.e. sounds that have similar centroids and similar attack
times blend better.
(TM = muted trombone, C1 = bass clarinet, C2 = Eb clarinet, O2 = oboe, TP =
trumpet, BN = bassoon, FH = french horn, FL = flute, S1 = string, sul ponticello (on bridge), S2 = normal string, S3 = muted string, X1-X3 = saxophones, EH = english horn).
Mds is so great.
Sandell 1993 study
Spectral centroid imp.
Blend worsens as the composite spectral centroid is made higher and the difference in spectral centroid increases.
This figure shows the transformations of the flute tone. In the darkened version, theupper harmonics were attenuated relative to the lower harmonics. In the bright
version, the upper harmonics were amplified relative to the lower harmonics.
For all combinations of instruments, Sandell computed the spectral centroid of the
dyads (composite centroid) and the difference in centroid between the two
instruments. He correlated this data with the blend ratings.
When everything else is held constant (i.e., when spectral centroid is manipulated
independently), the blend was found to be a function of the composite centroid
and/or the centroid difference of the pair of instruments. In general, a bright
instrument blended worse than a dark instrument. The blend worsens as the
composite spectral centroid is made higher. A pair with maximally distant centroids
blended worse than a pair with maximally close centroids. These results show that
one factor (spectral centroid) makes an independent contribution to blend.
ImplicTions of sandell study
Multiple reversion
3
Overall composite spectral
Attack time, similar, blended bette
Changes in loudness, ver time
New orchestrations, using more blend able instruments
Sequential grouping cues auditory stream formation, source tracking)
Spectral continue unity
Intensity conrinuitt
Spatial continuity
Effects timbre differences on streaming
This sound example by David Wessel demonstrates how timbre can be used
as a sequential grouping factor. The upper melody is played with a single
timbre and is heard as ascending triplets. The lower melody is identical in pitchcontent, but the notes alternate between two instruments. When the notes are grouped perceptually according to timbre, two auditory streams result, each
one with a descending triplet pattern at half the tempo of the original
sequence. Therefore, re-orchestration can actually alter melody perception. A source’s acoustic properties are taken into account when forming auditory
streams.
Results from study
These are the results of the McAdams & Cunibile study. The comparisons are
shown on the bottom. Not all possible comparisons were presented due to the
lack of timbres at the required position in the McAdams et al (1995) space.
The y-axis represents the percentage of choices of the first sequence in the
pair. If people responded randomly, all of the bars would be at the 50% line.
Black bars are for nonmusicians and hashed bars are for composers of
electroacoustic music. Bars that are statistically higher than 50% are marked
with an asterisk. Note that most of the bars are higher than 50% indicating that the theory works to some extent and works better for composers than for
nonmusicians (with one exception).
One confounding factor is that the specificities (acoustic feature unique to
certain sounds) for some timbres were ignored, which distorts the vectors usedto choose the timbres. Timbral intervals as a part of musical discourse would
be difficult to achieve with complex and idiosyncratic sound sources. This
suggests caution in applying this theory to complex timbre spaces in which not all of the parameters are under a composer’s control.
Perception of timbral relations
McAdams investigated if some of the operations commonly used on pitch
sequences (such as transposition) could be used on timbre sequences. A
timbre melody in the McAdams et al (1995) timbre space is shown. Could one”transpose” the melody played starting on the “guitarnet,” so that it starts on
the bowed string and follows the same timbral trajectory?
Leads to timbral interavls
McAdams and Cunibile proposed a model of timbral intervals, which would
have properties similar to pitch intervals. A pitch interval is a relation along a
well-ordered dimension that retains a degree of invariance under certain kinds
of transformation, such as translation along the dimension, which is what
musicians call “transposition.” A timbral interval can be considered as a vector
in space connecting two timbres with a specific length (distance between the
timbres) and a specific orientation. This can be represented as an oriented
vector (or arrow) in the timbre space. To transpose a timbre interval, all you
have to do is slide the vector to a new position keeping its orientation and
length constant.
To test the notion of timbre intervals, timbre space is used as a mathematical
model. McAdams & Cunibile (1992) used an analogy task: Timbre A is to
Timbre B as Timbre C is to which Timbre D? Different pairs of comparison
intervals with different Timbre D’s were presented. D1 fell very close to the end point of the transposed AB vector. D2 fell at the end of a vector that was about the right length, but had the wrong orientation (off by at least 90°). D3 was in
about the right direction, but was too long by at least a factor of 1.8. D4 was offin terms of both orientation and length. In each trial, listeners heard A-B-C-D,
followed by A-B-C-D’ (where D and D’ were different and chosen from D1, D2,
D3 or D4). They had to answer which sequence best filled the analogy.
In class demonstration of a timbral analogy task. Note that no timbre exists
near the ideal endpoint of the transposed vector. The theory would predict thatpeople would choose the timbre closest to this ideal point as best filling the
analogy. “gtr” is usually chosen, although “spo” is a close second.
Segmental grouping
Lerdahl & Jackendoff proposed a certain number of grouping rules that
structure musical sequences. Two principals underlie the rules: proximity
(relating to temporal events where the boundary is set after a prolonged sound among other short ones) and similarity (so-called “change rules” where the
similarity of events is registered after hearing the first differentiating element
that initiates the following group). Most of the rules result from a change or
discontinuity in some auditory attribute, including timbre. The timbre rule states that sequences with significant changes in timbre will be segmented at the
change point. So the first sequence would have 2 notes separated from 3
notes, and the reverse would be true of the second sequence.
Deliège (1987) studied Lerdahl and Jackendoff’s grouping rules in musical
excerpts. She found timbre discontinuities to be among the most often
detected by listeners. So timbre changes can be an important structuring force in musical sequences. Group boundary perception.using timbre to segment.
