Index of content:
Volume 105, Issue 3, March 1999
- MUSIC AND MUSICAL INSTRUMENTS 
Computer identification of musical instruments using pattern recognition with cepstral coefficients as features105(1999); http://dx.doi.org/10.1121/1.426728View Description Hide Description
Cepstral coefficients based on a constant Q transform have been calculated for 28 short (1–2 s) oboesounds and 52 short saxophonesounds. These were used as features in a patternanalysis to determine for each of these sounds comprising the test set whether it belongs to the oboe or to the sax class. The training set consisted of longer sounds of 1 min or more for each of the instruments. A k-means algorithm was used to calculate clusters for the training data, and Gaussian probability density functions were formed from the mean and variance of each of the clusters. Each member of the test set was then analyzed to determine the probability that it belonged to each of the two classes; and a Bayes decision rule was invoked to assign it to one of the classes. Results have been extremely good and are compared to a human perception experiment identifying a subset of these same sounds.
Vibrational patterns and frequency responses of the free plates and box of a violin obtained by finite element analysis105(1999); http://dx.doi.org/10.1121/1.426729View Description Hide Description
By means of the finite-element method, a study of the vibrational behavior of the violin has been developed in two parts. In the first part, the analysis concerns the free violin plates and their tuning process, which has been accurately described by this numerical tool. In the second part, the whole violin box—except for the neck—has been modeled and its first ten eigenmodes have been calculated. In both parts, the calculated vibrational patterns and frequency responses have been compared with experimental measurements of other researchers.
105(1999); http://dx.doi.org/10.1121/1.426730View Description Hide Description
The multiconvolution algorithm [Martı́nez et al., J. Acoust. Soc. Am. 84, 1620–1627 (1988)] to calculate the impulse response or reflection function of a musical instrument air column has proved to be useful, but it has the limitation that the spacing between discontinuites is constrained to be some multiple of (for phase velocity c and time step ). This paper presents an improved method, the continuous-time interpolated multiconvolution (CTIM), where such a limitation has been removed. The response of an air column, modeled as an arbitrary one-dimensional acoustic waveguide constructed using cylindrical or conical bore segments with viscothermal damping and tone-hole discontinuities, is obtained through continuous-time convolutions between analytical reflection and transmission functions and discrete-time pressure signals. The arbitrary spacing between discontinuites is accounted for by interpolation of the discrete-time pressure signals. Many musical instrument air columns possess tone holes that are opened or closed so that tones of different pitches are produced. A time-domain calculation is presented of the acoustic responses of tone-hole discontinuities that may be open or closed. The resulting reflection and transmission functions are well suited for use in the CTIM.
A microcosm of musical expression: II. Quantitative analysis of pianists’ dynamics in the initial measures of Chopin’s Etude in E major105(1999); http://dx.doi.org/10.1121/1.426743View Description Hide Description
Patterns of expressive dynamics were measured in bars 1–5 of 115 commercially recorded performances of Chopin’s Etude in E major, op. 10, No. 3. The grand average pattern (or dynamic profile) was representative of many performances and highly similar to the average dynamic profile of a group of advanced student performances, which suggests a widely shared central norm of expressive dynamics. The individual dynamic profiles were subjected to principal components analysis, which yielded five Varimax-rotated components, each representing a different, nonstandard dynamic profile associated with a small subset of performances. Most performances had dynamic patterns resembling a mixture of several components, and no clustering of performances into distinct groups was apparent. Some weak relationships of dynamic profiles with sociocultural variables were found, most notably a tendency of female pianists to exhibit a greater dynamic range in the melody. Within the melody, there were no significant relationships between expressive timing [, J. Acoust. Soc. Am. 104, 1085–1100 (1998)] and expressive dynamics. These two important dimensions seem to be controlled independently at this local level and thus offer the artist many degrees of freedom in giving a melody expressive shape.