Publications by Gregory J. Sandell

Note: publications marked are refereed.


G.J. Sandell and Christopher J. Darwin, "Recognition of concurrently-sounding musical instruments with different fundamental frequencies" (abstract). Journal of the Acoustical Society of America 100, 2683, 1996.

G.J. Sandell and M. Chronopoulos, "Identifying musical instruments from multiple versus single notes" (abstract). Journal of the Acoustical Society of America 100, 2752, 1996.

G.J. Sandell, review of Auditory Display; Sonification, Audification, and Auditory Interfaces, by Gregory Kramer (Ed.), Addison-Wesley, 1994. Music Perception 13/4, 583-591, 1996.

G.J. Sandell, "Roles for Spectral Centroid and Other Factors in Determining 'Blended' Instrument Pairings in Orchestration." Music Perception 13/2, 209-246, 1995.

G.J. Sandell and W.L. Martens, "Perceptual Evaluation of Principal Component-Based Synthesis of Musical Timbres." Journal of the Audio Engineering Society 43/12, 1013-1028, 1995.

C.J. Darwin and G.J. Sandell, "Absence of effect of coherent frequency modulation on grouping a mistuned harmonic with a vowel." Journal of the Acoustical Society of America 97, 3135-3138, 1995.

G.J. Sandell, "Analysis of concurrent timbres with an auditory model" (abstract). Journal of the Acoustical Society of America 95, 2958, 1994.

C.J. Darwin, V.Ciocca, and G.J. Sandell, "Effects of frequency and amplitude modulation on the pitch of a complex tone with a mistuned harmonic." Journal of the Acoustical Society of America 95, 2631-2636, 1994.

G.J. Sandell, "SHARC Timbre Database." Electronic publication: World Wide Web URL http://sparky.parmly.luc.edu/sharc, 1994.

G.J. Sandell, Concurrent Timbres in Orchestration: A Perceptual Study of Factors Determining 'Blend'. PhD Dissertation, Northwestern University, School of Music, 1991.

G.J. Sandell, "A library of orchestral instrument spectra." Proceedings of the 1991 International Computer Music Conference, 98-101, 1991.

G.J. Sandell, Reviews of Wayne Slawson's Sound Color and Robert Cogan's New Images of Musical Sound. Music Theory Spectrum 12/2, 255-261, 1990.

G.J. Sandell, "Perception of concurrent timbres and implications for orchestration." Proceedings of the 1989 International Computer Music Conference, 268-272, 1989.

G.J. Sandell, "Effect of spectrum and attack properties on the evaluation of concurrently sounding timbres" (abstract). Journal of the Acoustical Society of America 86 (supplement 1), p. S59, 1989.

C.L. Krumhansl, G.J. Sandell, and D. Sergeant. "The perception of tone hierarchies and mirror forms in twelve-tone serial music." Music Perception 5/1, 31-77, 1987.



"Roles for Spectral Centroid and Other Factors in Determining 'Blended' Instrument Pairings in Orchestration."

Abstract

Three perceptual experiments using natural-sounding instrument tones arranged in concurrently-sounding pairs investigate a problem of orchestration: what factors determine selection of instruments to achieve varying degrees of blend (fusion of multiple timbres into a single timbral image). The principle finding concerns the spectral centroid of the instruments (midpoint of the spectral energy distribution). Blend worsened as a function of the overall centroid height of the combination (the centroid of the composite spectrum of the pair), or as the amount of difference between the centroids of the two instruments increased. Slightly different results were found depending on whether the instruments were on the same pitch or separated by a minor third. For unisons, centroid height, attack similarity and loudness envelope correlation accounted for 51% of the variance of blend. For minor thirds, centroid difference, composite centroid, attack similarity, and synchrony of offset accounted for 63% of the variance of blend. In a third experiment instruments were manipulated to have different centroid levels to test if centroid made an independent contribution to blend. The results show that changes in centroid affect blend even when that is the only aspect of the sound that is changing. The findings create the potential for an approach to orchestration based on abstract properties of sound as a substitute for the traditional approach of teaching entirely by example.

Perceptual Evaluation of Principal Component-Based Synthesis of Musical Timbres.

Abstract

Harmonic-based analysis and resynthesis of musical instrument tones, for example using the Phase Vocoder method, is a valuable technique but its data representation is very large. However, this dataset is usually highly redundant. Principal Components Analysis (PCA) can be used to encode such data into a smaller set of orthogonal basis vectors with minimal loss of information. Techniques for applying PCA to such data are explored, and the aural impact of the method on three tones (cello, trombone and clarinet) are studied in two perception experiments. Results show that nearly identical resyntheses can be produced with a 40-70% data reduction. A pre-processing step called Variable Duration Temporal Partitioning (VDTP) is introduced, which also affords a natural-sounding method for time-expansion and contraction of tones. An extension of the PCA technique is also introduced that implements a timbre space, or coordinate system for interpolation among a group of musical instruments.

Absence of effect of coherent frequency modulation on grouping a mistuned harmonic with a vowel

Abstract

When a single harmonic close to the first formant frequency is mistuned by about 8%, that harmonic makes a reduced contribution to the vowel's first formant frequency as measured by a shift in the phoneme boundary along an F1 continuum between /I/ and /E/ [C.J. Darwin and R.B. Gardner, j. Acoust. Soc. Am. 79, 838-45 (1986)]. In the present experiments, phoneme boundaries along an /I/-/E/ continuum were measured for vowels differing in F1 whose fourth harmonic (500 Hz) was mistuned by 0, ±3, ±6, or ±9%. All the harmonics of a vowel (including the mistuned one) were given either no FM or coherent FM at a rate of 6 Hz and modulation depth of ±5%. The results replicated the previous findings, but found no evidence for coherent FM preventing the segregation of the mistuned harmonic from the vowel.

Analysis of concurrent timbres with an auditory model

Abstract

Concurrently sounding timbres (two or more instruments playing on the same or different pitches) sometimes yield unexpected qualities of sound which are exploited in orchestration. A model simulating spectrum analysis by the human auditory periphery is used to explore the differing contributions of individual instruments to the sum sound. Signals are passed through a bank of overlapping bandpass filters whose center frequency spacing corresponds to equal physical distances along the basilar membrane. Thus sinusoidal components do not fall into single "bins" as in a Fourier analysis, but rather spread excitation over a large number of channels. Some partials are resolved in separate channels while other partials are processed in a single channel with others, leading to interactions that determine the overall timbre of the combination. Sampled musical instrument tones are combined into concurrent arrangements and analyzed with the method.

Effects of frequency and amplitude modulation on the pitch of a complex tone with a mistuned harmonic

Abstract

It has previously been found that when a single low-numbered harmonic of a complex tone is progressively mistuned, for mistunings to about 3%, the pitch of the complex changes in the direction of the mistuning but for larger mistunings (by about 8%) the pitch returns to its original value. The result is compatible with the operation of a mechanism such as a graded harmonic seive, which can reject from the calculation of pitch those frequency components that are implausibly distant from a harmonic frequency. The first experiment shows that the tolerance of such a sieve is increased when all the components of the complex tone (including the mistuned component) share a common pattern of frequency modulation at a rate of 6 Hz. The second experiment shows that the tolerance of the sieve is not increased whtn the components share a common pattern of amplitude modulation at 17 Hz. The third experiment replicates these findings and further shows that the increase in sieve tolerance for RM, but not AM, occurs at both 6 and 17 Hz.

SHARC Timbre Database

Abstract

SHARC is a public-domain, World Wide Web-available research database of musical timbre. It consists of steady state spectra taken from over 1300 different musical instrument tones, including complete chromatic runs from the standard playing range of the primary non-percussive instruments of the modern orchestra (39 instruments total). The source of the musical notes were the orchestral tones from the McGill University Master Samples (MUMS) Compact Discs, digital recordings of live musical performers. For each note, a short portion corresponding to the sustain or "steady state" portion of the tone was selected and analysed with a Fourier analysis. The database is available on the World Wide Web (
http://sparky.parmly.luc.edu/sharc). Users can download the entire database or "browse" it using SHARCs interactive plotting and soundfile features. A user can request a plot of any individual note for a given instrument, and explore ways of plotting it using various amplitude and frequency scales, synthesize a waveform of the note and play it over the users Web browser (e.g. NetScape). Some of the uses to which SHARC has been put are:

Concurrent Timbres in Orchestration: A Perceptual Study of Factors Determining 'Blend'

G.J. Sandell, PhD Dissertation, Northwestern University, School of Music, 1991.

Abstract

Orchestration often involves selecting instruments for concurrent presentation, as in melodic doubling or chords. One evaluation of the aural outcome of such choices is along the continuum of "blend": whether the instruments fuse into a single composite timbre, segregate into distinct timbral entities, or fall somewhere in between the two extremes. This study investigates, through perceptual experimentation, the acoustical correlates of blend for 15 natural-sounding orchestral instruments presented in concurrently-sounding pairs (e.g. flute- cello, trumpet-oboe, etc.).

Ratings of blend showed primary effects for centroid (the location of the midpoint of the spectral energy distribution) and duration of the onset for the tones. Lower average values of both centroid and onset duration for a pair of tones led to increased blends, as did closeness in value for the two factors. Blend decreased (instruments segregated) with higher average values or increased difference in value for the two factors. The musical interval of presentation slightly affected the relative importance of these two mechanisms, with unison intervals determined more by lower average centroid, and minor thirds determined more by closeness in centroid. The contribution of onset in general was slightly more pronounced in the unison conditions than in the minor third condition. Additional factors contributing to blend were correlation of amplitude and centroid envelopes (blend increased as temporal patterns rose and fell in synchrony) and similarity in the overall amount of fundamental frequency perturbation (decreased blend with increasing jitter from both tones).

To confirm the importance of centroid as an independent factor determining blend, pairs of tones including instruments with artificially changed centroids were rated for blend. Judgments for several versions of the same instrument pair showed that blend decreased as the altered instrument increased in centroid, corroborating the earlier experiments. Other factors manipulated were amplitude level and the degree of inharmonicity.

A survey of orchestration manuals showed many illustrations of "blending" combinations of instruments that were consistent with the results of these experiments. This study's acoustically-based guidelines for blend augment instance-based methods of traditional orchestration teaching, providing underlying abstractions helpful for evaluating the blend of arbitrary combinations of instruments.



The perception of tone hierarchies and mirror forms in twelve-tone serial music

Abstract

Four experiments are reported in which the materials are dervice from two 12-tone serial compositions (Schoenberg's Wind Quintet and String Quartet, No. 4). Two experiments use the probe tone method (Krumhansl & Shepard, 1979) to asess factors contributing to tone prominence in serial music. The contexts in Experiment 1 are musically neutral statements of the complete or incomplete tone rows; the context in Experiment 4 are excerpts from the two pieces. Two experiments iuse a classification task to evaluate whether the prime form of the row is perceived as similar to its mirror forms (inversion, retrograde, and retrodrage inversion). The materials are neutral presentations of the forms (Experiment 2) or excerpts from the pieces (Experiment 3). Large individual differences are found. A subgroup of listeners, with more music training on average, show the following effects in the probe tone experiments: low ratings for tones sounded more recently in the contexts and high ratings for tones not yet sounded; low ratings for tones fitting with local tonal implications; similar patterns for the neutral contexts and the musical excerpts. The remaining listeners show the opposite effects. Classification accuracy of mirror forms is above chance and is higher for the neutral sequences than the musicl excepts; performance is correlated with music training. The experiments show that some, but not all, listeners can percieve invariant structures in serial music depitre mirror transformations, octave transpositions of tones, and variations of rhythm and phrasing.

Recognition of concurrently-sounding musical instruments with different fundamental frequencies

Abstract

The "double vowel" paradigm [P. F. Assmann and Q. Summerfield, J. Acoust. Soc. Am. 88, 680--697 (1990)], in which recognition for two concurrently-sounding vowels is measured over various fundamental frequency (F0) separations, is applied to natural musical instrument sounds. The findings of the previous work, in which recognition reaches peak by 1 to 2 semitone difference, is replicated here, with about 35% greater accuracy overall for musically experienced over naive listeners. Attention is given to special cases of instruments that are frequently confused for one another even when F0 separations are wide (e.g., clarinet and English horn, or flute and French horn), since such combinations may have special value in musical orchestration. In another condition, concurrently sounding instruments are adjusted in level by listeners so the two sound equally salient. The double-timbre study is then run with variable relative levels between the two instruments. Translations between the effects of semitones separation and difference in level are explored.

Identifying musical instruments from multiple versus single notes

Abstract

These experiments investigated to what degree the identification of a musical instrument depends on change of timbre across total pitch range. Subjects were trained to recognize 12 orchestral musical instruments (natural recordings) by name. Each instrument was represented by a variety of pitches sampled regularly from the standard pitch range of that instrument. The sampling was four notes per octave, effectively forming (in musical terms) a diminished seventh chord arpeggio. First listeners identified the instruments from single note presentations taken from the arpeggios. Results showed that an instrument's identity is not equally salient across its range of notes, in ways that differ for each instrument. Next listeners were tested on a variable number of notes (N=1,2,...,7) sampled from the arpeggio presented during training in such a way that with increasing N, a greater amount of the arpeggio's total range was included. Results showed that on average, recognition improved as increasingly recognizable notes (experiment 1) were included in the arpeggio. The relationship between the rate of improvement with N and the recognizability of the single notes is explored. An auditory gestalt called "macrotimbre" is proposed to consider the role across-range timbre changes play in the identification of an instrument. [Work supported by NIH.]

Perception of concurrent timbres and implications for orchestration

Abstract

Candidates for perceptual correlates to timbre and orchestration are considered, and one of them (blend) is studied in an experiment using the Stanford collection of synthesized musical instruments. Results suggest that the blend judgment is inversely related to either the sum of the centroids of the two tones, or the sum of the attack times. The lower the overall centroid, or attack duration, the better the blend.

Effect of spectrum and attack properties on the evaluation of concurrently sounding timbres

Abstract

The perception of concurrently-sounding musical instruments underlies musical orchestration, but current knowledge about psychological attricutres of timbre emphasizes successive comparison (e.g., similarity, analogy). The psychological attribute "blend" was investigated by having eight listeners use a ten-point scale (a continuum of "separated" to "fused") to rate all possible simultaneous presentations of synthesized instrumental tones used in an earlier study [J. Greg, J. Acoust. Soc. Am. 61, 1270-1277 (1975)]. Following Grey's three-dimensional model, the data was analyzed for the effects of harmonic synchrony, spectrum content (centroid), and attack characteristics. The latter two attributes showed a significant realtionship to blend: in general, "dark" instruments and instruments with quick, clean attacks tended to blend well. The amount of blend reported for such instruments decreased as their pairs increased in brightness or attack length, and pairs of bright or noisy-attack instruments did not blend well. The results found here corroborate some of the recommendations for desirable blends in traditional orchestration manuals.

A library of orchestral instrument spectra

Abstract

Standard orchestral instruments and some early music instruments were analyzed for steady-state spectra. Examining the spectra from the entire playing range of an instrument reveals patterns of change that may be essential to its timbral character. The change in the centroid over the range of the instrument is looked at in detail.