Christophe d’Alessandro

On the use of the derivative of electroglottographic signals for characterization of nonpathological phonation

Electroglottography is a common method for providing noninvasive measurements of glottal activity. The derivative of the electroglottographic signal, however, has not attracted much attention, although it yields reliable indicators of glottal closing instants. The purpose of this paper is to provide a guide to the usefulness of this signal. The main features that are to be found in this signal are presented on the basis of an extensive analysis of a database of items sung by 18 trained singers. Glottal opening and closing instants are related to peaks in the signal; the latter can be used to measure glottal parameters such as fundamental frequency and open quotient. In some cases, peaks are doubled or imprecise, which points to special (but by no means uncommon) glottal configurations. A correlation-based algorithm for the automatic measurement of fundamental frequency and open quotient using the derivative of electroglottographic signals is proposed. It is compared to three other electroglottographic-based methods with regard to the measurement of open quotient in inverse-filtered derived glottal flow. It is shown that agreement with the glottal-flow measurements is much better than most threshold-based measurements in the case of sustained sounds.

Glottal open quotient in singing: Measurements and correlation with laryngeal mechanisms, vocal intensity, and fundamental frequency

This article presents the results of glottal open-quotient measurements in the case of singing voice production. It explores the relationship between open quotient and laryngeal mechanisms, vocal intensity, and fundamental frequency. The audio and electroglottographic signals of 18 classically trained male and female singers were recorded and analyzed with regard to vocal intensity, fundamental frequency, and open quotient. Fundamental frequency and open quotient are derived from the differentiated electroglottographic signal, using the DECOM (DEgg Correlation-based Open quotient Measurement) method. As male and female phonation may differ in respect to vocal-fold vibratory properties, a distinction is made between two different glottal configurations, which are called laryngeal mechanisms: mechanism 1 (related to chest, modal, and male head register) and mechanism 2 (related to falsetto for male and head register for female). The results show that open quotient depends on the laryngeal mechanisms. It ranges from 0.3 to 0.8 in mechanism 1 and from 0.5 to 0.95 in mechanism 2. The open quotient is strongly related to vocal intensity in mechanism 1 and to fundamental frequency in mechanism 2.

The pitch of short‐duration vibrato tones

The pitch perceived for short vocal vibrato tones was measured using a method of adjustment. The stimuli were synthetic vocal tones, produced by a formant synthesizer. The main parameter under study was the tone duration, as a function of the fractional number of vibrato cycles. This parameter was examined in relation to (1) the vibrato extent (0, 50, 100, and 200 cents); (2) the vibrato rate (4, 6, and 8 Hz); (3) the tone nominal frequency (220, 440, 880, and 1500 Hz). Durations ranging from 1/2 cycle to 2 cycles were studied. Our results showed that for short tones, the pitch perceived does correspond to a weighted time average of the F0 pattern. A separate perception took place for the high and low parts of the vibrato cycles, for large vibrato extents or slow vibrato rates. This phenomenon was related consistently with the glissando threshold. A simple numerical model of weighted time averaging with threshold was proposed. It demonstrated a good agreement with our experimental data. Finally, the exper...

The pitch of short-duration fundamental frequency glissandos.

Pitch perception for short-duration fundamental frequency (F0) glissandos was studied. In the first part, new measurements using the method of adjustment are reported. Stimuli were F0 glissandos centered at 220 Hz. The parameters under study were: F0 glissando extents (0, 0.8, 1.5, 3, 6, and 12 semitones, i.e., 0, 10.17, 18.74, 38.17, 76.63, and 155.56 Hz), F0 glissando durations (50, 100, 200, and 300 ms), F0 glissando directions (rising or falling), and the extremity of F0 glissandos matched (beginning or end). In the second part, the main results are discussed: (1) perception seems to correspond to an average of the frequencies present in the vicinity of the extremity matched; (2) the higher extremities of the glissando seem more important; (3) adjustments at the end are closer to the extremities than adjustments at the beginning. In the third part, numerical models accounting for the experimental data are proposed: a time-average model and a weighted time-average model. Optimal parameters for these models are derived. The weighted time-average model achieves a 94% accurate prediction rate for the experimental data. The numerical model is successful in predicting the pitch of short-duration F0 glissandos.

REALTIME AND ACCURATE MUSICAL CONTROL OF EXPRESSION IN SINGING SYNTHESIS

In this paper, we describe a full computer-based musical instrument allowing realtime synthesis of expressive singing voice. The expression results from the continuous action of an interpreter through a gestural control interface. In this context, expressive features of voice are discussed. New real-time implementations of a spectral model of glottal flow (CALM) are described. These interactive modules are then used to identify and quantify voice quality dimensions. Experiments are conducted in order to develop a first framework for voice quality control. The representation of vocal tract and the control of several vocal tract movements are explained and a solution is proposed and integrated. Finally, some typical controllers are connected to the system and expressivity is evaluated.

Comparing approaches to pitch contour stylization for speech synthesis

This chapter describes two approaches to pitch contour stylization as well as a perception experiment to evaluate and compare both methods. The first approach uses an automatic stylization procedure, based on perceptual criteria. It outputs the sequence of audible pitch events (static tones, dynamic tones, complex dynamic tones) in the utterance. Both the tonal perception model and the algorithm are described in some detail. The second approach, known as close-copy stylization, is a manual procedure in which a straight-line approximation of the pitch contour is obtained interactively, by resynthesis of the stylized contour and auditory comparison with the original. A perception experiment using synthetic stimuli with stylized contours was run in order to compare and evaluate both approaches. The stylized contours can hardly be distinguished from the natural contours. Tonal perception stylization gives slightly better results than straight-line stylization.

Human voice phoneme directivity pattern measurements

The application of directivity patterns to radiating sources into computer simulations and auralizations is common for loudspeaker models. Few applications include the directivity patterns of natural sources, partly due to the lack of sufficient data. This work presents the results of a detailed measurement study on human voice directivity in three‐dimensions. Unlike previous studies that have used average directivity data over read phrases, this work presents results that are measured for a number of sustained individual phonemes. Details of the measurement protocol and posttreatment processing are presented. Comparisons are made relative to phoneme, f0, spectral characteristics, and associated mouth geometry for several talker subjects. Studies have also been made on the directivity of the singing voice. Specifically, the variations in directivity relative to level (piano, fortissimo, etc.) and projection as controlled by the singer have been investigated. Results of this work are applicable to speech p...

On the dynamics of the clavichord: From tangent motion to sounda)

An experimental study of variations in the sound of clavichord notes at different dynamic levels is described. Radiated acoustic signal, tangent velocity and two tangent-string contact signals are synchronously measured for all 51 notes of an unfretted instrument. More than ten repeated measures are recorded in order to obtain as much variation in dynamic level as possible. The tangent motion, expressed in terms of velocity, is studied in the time and frequency domains. A model of the tangent-string contact point velocity is proposed. Then, three aspects of the sounded tones are analyzed: SPL and its relationship to tangent velocity, spectral slope, and pitch variations. These results indicate a linear relationship between sound pressure level and tangent peak log velocity. Spectral slope seems almost constant independent of tangent velocity and dynamic level. Both tangent velocity and finger pressure are shown to influence the fundamental frequency. In conclusion, controlling both finger velocity and finger pressure may prove challenging for the player, and this may explain why the sound quality of the clavichord depends so much on the players ability.

On the waveforms and spectra of glottal flow models

Several glottal flow models have been proposed for speech analysis and synthesis (e.g., LF, Rosenberg, R++, and Klatt). All these models do not use the same number of parameters, or the same name for similar parameters, and it appears difficult to compare their merits. Then, a unified framework for studying the time and frequency domain properties of glottal flow models is proposed. It is shown that all the models can be represented by a common set of five time‐domain parameters: three scale parameters (T0, peak amplitude, open quotient), a shape parameter (asymmetry quotient), and a closure continuity parameter. A generating function is computed for each model by normalization of the model with respect to scale parameters and closure continuity parameter. The specific features of each model are represented in its generating function. The spectrum of generating functions is low pass, and its derivative can be characterized by a spectral maximum, coined ‘‘glottal formant.’’ The closure continuity parameter...

Automatic intonation stylization using a model of pitch perception

Automatic analysis of intonation is a crucial step towards the long awaited automatic transcription of intonation. A new psychoacoustic model of pitch perception for short tones [d’Alessandro and Castellengo, J. Acoust. Soc. Am. 95, 1617–1630 (1994)] is applied to the problem of automatic F0 stylization for syllable‐sized units in French. The aim is to compute one or several psychologically justified tonal movements for each syllable. An important property of the model is that it models perception prior to any process of linguistic categorization. The model was implemented in a computer program for tonal analysis. This program contains several procedures: (1) F0 detection and voiced/unvoiced decision; (2) syllabic segmentation, based on loudness and zero‐crossings; (3) pitch integration using the model of pitch perception; (4) glissando rate computation, and static/dynamic tone decision; (5) differential glissando rate for complex pitch patterns computation. The analysis procedure described above has been...