Voice Timbre Research Articles

Voice timbre in autophonic production compared with that in extraphonic production.

It is well known that voice timbre in autophonic production (hearing one's own voice during phonation) is different from that in extraphonic production (hearing one's voice after recording). The differences between the two productions, however, have not been evaluated quantitatively. This results naturally from the fact that there is no objective way to evaluate the voice timbre in autophonic production. The purpose of the present study was to show a method, named the delayed feedback method created by the author, in which voice timbre in autophonic production can be examined objectively, as a function of frequency, and then to evaluate the timbre in comparison with that in extraphonic production quantitatively, when male subjects phonated five Japanese vowels in the spoken mode. The experiment was conducted so as to simulate the sound through a loudspeaker as accurately as possible in its timbre and loudness to that perceived during phonation. It was found that, for all the vowels, the vocal sound was perceived relatively louder in the low frequency region (about 5 dB at 100 Hz) and softer in the high frequency region (about -5 dB at 4 kHz) than the reproduced sound after recording. This may result from the bone conduction elements during phonation. The findings obtained do also, in fact, coincide with experiences in everyday life.

Perceptual implications of selected aspects of vocal behavior

The perceptual dimensions derived from acoustical changes associated with vowel quality, voice quality, voice classification, and gender are not clearly defined. This project uses the trained singing voice as a source of controlled variation in parameters important to the timbre of both speech and voice. A database of vowels was collected from 18 singers representing six major voice classifications. Repetitions were made of three vowels at three or four fundamental frequencies and at three dynamic levels in normal register. Six subjects also attempted to match loudness (as indicated by a target of equal phons) for this isoparametric condition. Spectral measures (e.g., FFTs, STFTs, 1/3-octave band levels) were transformed in a manner intended to approximate perceptually relevant aspects of auditory processing (e.g., loudness and sharpness densities and loudness and sharpness over time). By comparing the spectral measures with their transformations, it may be possible to make inferences concerning which acoustical changes are perceptually relevant, and about the nature of their perceptual representation. [Work supported by the Institute for Perception Research, Eindhoven, The Netherlands.]

Speaking behavior and voice sound characteristics in depressive patients during recovery

Based on a sample of 30 depressive patients, we have investigated the time course of recovery from depression in so far as this time course was assessable through changes in psychopathology syndrome scores and through changes in speaking behavior and voice sound characteristics. Specifically, our study design provided 6 repeated assessments over 2 weeks and at a fixed time in the morning each Monday, Wednesday and Friday, plus a final assesment at the patients' releases from hospital. Thus, we were able to determine the degree to which single- parameter approaches to speaking behavior and voice sound characteristics reflect the individual time course of recovery from depression. In this context, we could rely upon a calibration sample with repeated assessments on 192 healthy volunteers which yielded all necessary information concerning reproductibility and sensitivity of speech parameters. Our analysis revealed several prominent features of speaking behavior and voice sound characteristics to be closely related to the time course of recovery from depression. In particular, the parameters “F0-amplitude”, “F0-6db-bandwidth” and “F0-contour” which assess important characteristics of a speaker's voice timbre, as well as the parameters “energy” and “dynamics” which assess a speaker's mean loudness and the variation of loudness over time, displayed consistently high correlations with depression syndromes. Moreover, the results of single-case analysis turned out to be in remarkable accordance with those of the cross-sectional one: in almost two-thirds of patients there existed a significant relationship over time between the global depression scores and major speech parameters. As to the remaining one-third of patients who did not fit the picture of high correlations between psychopathology and speech parameters, we found an overproportionally large number of non-improvers characterized by irregular patterns of slight improvement with subsequent deterioration, or of deterioration followed by slight improvement. In other words, one-third of patients displayed time courses of depression whose psychopathology is difficult to assess through standard exploration techniques. Accordingly, it is not clear whether the observed lack of correlation in these patients is due to insufficient data or to an actual disordance between the time development of psychopathology and that of speech parameters.

PARALLEL PERFORMANCE TRAITS IN MEDIEVAL AND MODERN TRADITIONAL MEDITERRANEAN SONG

Three quotations spark off the present paper. The first is by the philologist Paul Zumthor: ?Medieval literature can be more aptly compared with folklore... than with modern literatures?1. The second quotation is by the literary and musical scholar John Stevens with reference to song writing in the Middle Ages: ?In plain language, the musician did not set the words of the poem to music; he set its pattern?2. The implications of that observation were developed subsequently by its author3 and have sparked off some discussion, none of which was accessible in Ceau?escu's information-starved Romania to folklorist Speranza R?dulescu who mirrored Stevens's words in uncanny fashion, and unwittingly provided my third quotation, when in private conversation with me in 1987 on her country's traditional song she remarked how ?The words have their pattern projected on the music?. With that chance remark it finally dawned on me how much more there is yet to be played for in the admittedly difficult and dangerous game of drawing parallels between medieval and modern traditional performance practice. Published research has centred on the search for models?some more appropriate than others?for the use of instruments, improvisation and ornamentation techniques, and types of voice timbre in the performance of

Changes in voice characteristics after uvulopalatopharyngoplasty.

Previous reports have warned that tonsillectomy or uvulopalatopharyngoplasty (UPPP) may alter patients' speech by increasing the amount of nasal resonance as well as by changing voice timbre due to enlargement of the vocal tract. However, very few objective investigations, excluding nasality problems, have been carried out. We studied eight patients who underwent surgery for UPPP and recorded pre- and postoperative speech. The speech samples were then rated by seven experienced listeners, and acoustic spectra of two long vowels,/a/ and /e/, were analyzed using a computer program (MacSpeech Lab II). The listeners were unable to make a distinction between pre- and postoperative voice samples in the recordings studied. Acoustic analysis showed that the fundamental frequency as well as the first and second formants remained essentially unchanged. Present findings show that UPPP should not have a significant effect on voice characteristics as long as excessive nasality is not produced.

Formant frequency tuning in singing

<h2>Summary</h2> It is sometimes claimed that some singers tune their two lowest formant frequencies to harmonic partials in order to increase the audibility of the voice. Voice acoustics predicts that such tuning of formants should cause vowel quality to change. Using a newly constructed digital singing machine, the authors have explored the perceptual consequences of such tuning. Four different cases were represented, in which the two lowest formant frequencies were either constant or adapted to the fundamental frequency according to either of three different strategies. The resulting voice timbres were judged by an expert panel of singing teachers in a listening test consisting of descending chromatic scales. Constant formant frequencies were clearly preferred, presumably because formant tuning entails formant frequency shifts between adjacent tones so substantial that salient vowel quality shifts occur.

Physiological change, communication, and the playing of improvised music: Some proposals for research

In our music therapy research (Aldridge, 1988b; Aldridge & Vemey (1988) one of the areas we have considered is what happens when people improvise music together. Our intention is to be able to demonstrate to ourselves and other practitioners the influence we believe music therapy has on the body of the patient. We hope to demonstrate, in our later investigations, the mutual relationship of physical changes in the therapist and patient during the process of improvising music together. In trying to demonstrate this change we have looked for a simple physiological indicator. From a preliminary review of the literature about communication, suitable indicators emerged from the studies of cardiovascular change. The principal measures in such work were those concerned with changes in blood pressure and heart rate. Heart rate was chosen because it was a relatively easy parameter to observe and measure. Perhaps more significantly, it was a parameter acceptable to medical science with whom we, as a therapeutic discipline, are attempting to promote a dialogue. An earlier paper (Aldridge, 1988a) mentioned the important factors associated with both biological form and musical form-time, phrasing, pitch, rhythm, and melodic contour. Similar considerations apply for studies of communication. The basic preverbal fundamentals of human communication are called suprasegmentals-these are time, phrasing, rhythm, pitch, and voice tone (which would more accurately be called timbre). It is these qualities that are considered by music therapists when they assess tape-recorded sessions of improvised music therapy. The literature that first alerted us to these factors concerned chronic heart disease and Type A behavior (Dielman et al., 1987; Dimsdale, Stem, & Dillon, 1988; Friedmann, Thomas, Kulick-Ciuffo, Lynch, & Suginohara, 1982; Linden, 1987; Lynch, Long, Thomas, Malinow, & Katchor, 1981; Smith & RhodeWalt, 1986). Heart disease patients were described in terms that owed as much to a musical basis as they did to a physiological process. Loud, fast speech using a limited range of voice timbre, and speech patterns that interrupted the responses of a partner, appeared to reflect qualities music therapists heard in their descriptions of patients when they creatively improvised music (see Table 1). Although these qualitative descriptions may only be regarded as noise in the formal terms of grammar, they provide the essential expressive context for communication. It seemed important to compare apparently similar statements from two different theoretical backgrounds to see if there was any commonality between them. As heart disease is such an important problem in both mortality and morbidity throughout the Western world, it made sense to propose how a discipline like improvised music therapy could offer a tool for both assessment and treatment. The medium of improvised music offers possibilities for extremely varied communication and has a

Some perceptual correlates of phase in RELP coded speech

A series of experiments tested the contribution of phase information to the naturalness of residual excited linear prediction (RELP) coded speech, that is, how close a speaker's voice sounds under RELP coding to its unprocessed analog. Listeners were presented with a corpus of sentences uttered by eight speakers (four male and four female). Three classes of approximations of the residual phase spectra were employed on these sentences: quantized approximations for decreasing bit rates, noise‐corrupted approximations for increasing noise levels, and combinations of these over varying window sizes. Listeners classified these utterances on the basis of their proximity to unprocessed speech. Results indicate that residual phase spectrum provides information about voice timbre and quality and acts as a cue to speaker identity in coded speech.

Vertical laryngeal position—Research findings and their relationship to singing

Thomas Shipp: Our panelists will make a few statements, exchange some information and then we will open the discussion to the audience. Professor Sundberg. Johan Sundberg: Larynx height is a fascinating aspect of singing, in a physiological sense, and I would like to add a few comments. One is that I think that most speakers have the habit of raising their larynx with pitch, so that is the starting point for any voice training as far as I know. Larynx height has an acoustical effect on the voice organ, in that it shortens the resonance cavity in the pharynx, and this affects all format frequencies. If you change the format frequency, which you do by raising the larynx, then you change the voice timbre. I have a concern in regards to your findings that glissando was affiliated with larynx rises in professional singers. That would mean that when they sang glissando they would accept a change in voice timbre with pitch, and I don't think that is necessarily esthetically a great thing to do. I think that timbre should change, not for pitch reasons, but for expressive reasons. I would like to raise the question if you think that the effect of the glissando and larynx height was due to the fact that the glissando is a rather lovely practiced thing in singing music. My next point is that the larynx height has an acoust ical effect not only on the resonance chambers of the outer voice organ but also on the source itself. I have always thought that as larynx lowering is associated with inhaling, and inhaling is associated with glottal widening, then there is probably an abductory component in the larynx low-

Voice timbre in solo and choir singing: Is there a difference

Professional and amateur bass/baritone singers with experience both as soloists and choir singers heard their own voice mixed with a piano accompaniment while they performed a specially composed solo selection. Then the piano accompaniment was replaced by a recording of a choir made in the bass section during rehearsal and the subjects were asked to sing the bass part. In each case the subject's voices were recorded, and comparisons were made between similar vowels sung at the same pitch and sound pressure level. The output spectra are noticeably greater in the region of the singer's formant [J. Sundberg, J. Acoust. Soc. Am. 55, 838 (1977)] in the solo singing mode. However, differences in the source spectra, obtained by inverse filtering of the formants, are not consistent among the singer subjects.

Coding of the speech spectrum in three time-varying sinusoids.

Recent perceptual experiments with normal adult listeners show that phonetic information can readily be conveyed by sinewave replicas of speech signals. These tonal patterns are made of three sinusoids set equal in frequency and amplitude to the respective peaks of the first three formants of natural-speech utterances. Unlike natural and most synthetic speech, the spectrum of sinusoidal patterns contains neither harmonics nor broadband formants, and is identified as grossly unnatural in voice timbre. Despite this drastic recoding of the short-time speech spectrum, listeners perceive the phonetic content if the temporal properties of spectrum variation are preserved. These observations suggest that phonetic perception may depend on properties of coherent spectrum variation, a second-order property of the acoustic signal, rather than any particular set of acoustic elements present in speech signals.

Holographic investigations of skin vibrations

The paper describes preliminary results of the use of double-pulse holographic interferometry for vibration investigations of the skin surface of a singer's face (epiphysis), neck, larynx, and sternum. Holographic interferograms obtained in the experiment illustrate a dependence of the distribution of vibrations on the type of sound and on the fundamental frequency for the same person. The vibration interferograms of various persons are different even at the same sound and frequency. The method may become useful for diagnosis of voice timbre as well as in pathological cases.

Variable frequency generator for polyphonic electronic music system

Simultaneously played notes on a keyboard are automatically provided with different voice timbres. The assignment of voices to notes is by relative frequency-scale position within the chord. If four or more notes are played, the notes are scanned along the musical scale and the three lowest notes played and the highest note are sounded in different voices with the solo voice assigned to the highest note. Assignment of voices occurs automatically when fewer than four keys are played. One form of the system combines time multiplexed keying with priority coupling of the keying information to a plurality of voltage controlled oscillators through sample-and-hold memory circuits. Each oscillator has its own tone voicing circuits. In another preferred all-digital embodiment, the time division multiplex digital logic signals representative of the played keys are applied instead to a memory circuit and a comparator circuit. If the information stored in the memory from the previous scan cycle is the same as the information of the present cycle, thereby indicating no change in played keys, the comparator provides an enabling signal to a monostable and enable logic circuit. A priority selector circuit provides logic information representative of the three lowest keys played and the highest key played to the monostable and enable logic circuit. The monostable and enable logic circuit provides on four outputs information representative of the three lowest keys played and the highest key played and also information representative of the total number of keys played. This information is combined with the octave information and the note time slot information in four programmable counter circuits which divide four input frequencies from a rate scaler frequency generator to produce four tone signals having fundamental output frequencies f1 - fH corresponding to the three lowest notes and single highest note played. Also, the four programmable counter circuits provide control voltages corresponding to the played notes. The tone signals and control voltages are applied to voltage controlled gates and filter circuits which provide the necessary voicing to produce different orchestral effects on different notes. The resulting orchestral effect can be enhanced from an ensemble standpoint by controlled detuning of the four output frequencies, both within the chordal relationship and relative to other tones of corresponding pitch nomenclature under control of the same master oscillator. The outputs of the voltage controlled gate and filter circuits are applied to at least 2 output channels to avoid electrical cancellations when two programmable counters are assigned to the same note but slightly detuned.

Acoustic properties of voice timbre types and their influence on voice classification

An investigation to determine the significance of formant frequencies, pitch, and source-spectrum slope on voice classification is reported. Eight professional singers sang five common vowels on four common pitches, and, in a forced-choice test, vocal pedagogues classified the vocalization as tenor, baritone, or bass. Formant frequency and pitch differences accounted for most of the jury- evaluation results. To evaluate more accurately the significance of formant frequencies, pitch, and voicing and source-spectral slope on voice classification, vowel synthesis was employed. Measurements of spoken vowel vocalizations revealed higher formant frequencies in tenor timbre type and lower formant frequencies in bass timbre type. Long-time-average spectra seem to afford a good estimate of voice classification. Formant-frequency percentage differences between basses and tenors were similar to those found between males and females, suggesting that the mouth and pharynx lengths, known to differentiate females and males, probably also differentiate tenors and basses.

Some Results of the Analysis of Singing Voice

Oscillographic records of some masculine voices were made by means of high quality electroacoustic devices. Vowels u(oo), o, a (ah), e, i (ee) of various pitch from 94 cps to 490 cps were recorded. The records lasting 0.1 sec were made at the moment of the stationary, prolonged sounding of the vowel. The harmonic analysis of different periods of the voice of an artistic singer showed a sharply defined increase of harmonic components in two narrow bands of the spectrum: about 400–600 cps and about 2200–2800 cps. Doubtless in well-trained, good voices there are characteristic “singing formants,” practically the same for all vowels. For the vowels “u” (oo) and “i” (ee), the lower formant sometimes gets as low as 250–300 cps. The higher formant is characteristic for a “metallic” timbre of voice. In the voice of an inexperienced singer (baritone) with a strong but sharp timbre of voice and noticeable fall of loudness higher than mi♭ (325 cps), these sharply defined singing formants are not observed; the whole region of harmonies higher than 1500 cps shows a general increase. It is suggested that the singer arranges formant regions characteristic of a vowel and necessary for good articulation only in the beginning of a vowel (in the period of attach) after which he quickly rearranges the vocal apparatus to the singing position, which is almost the same for all the vowels. The quick and distinct rearrangement to the singing position is a necessary quality of a trained voice. The above-mentioned results agree with our former investigations of 1927 and with the investigations of Bartholomew who found a high singing formant in the region about 2800–3200 cps.