Development and perceptual evaluation of amplitude-based F0 control in electrolarynx speech.

PURPOSE: Current electrolarynx (EL) devices produce a mechanical speech quality that has been largely attributed to the lack of natural fundamental frequency (F0) variation. In order to improve the quality of EL speech, in the present study the authors aimed to develop and evaluate an automatic F0 control scheme, in which F0 was modulated based on variations in the root-mean-square (RMS) amplitude of the EL speech signal. METHOD: Recordings of declarative sentences produced by 2 male participants before and after total laryngectomy were used to develop procedures for calculating F0 contours for EL speech. Specifically, the positive linear relationship between F0 and RMS amplitude observed in pre-laryngectomy speech was used as the basis for generating an F0 contour based on the amplitude variation of EL speech. An analysis-by-synthesis approach was used to modify the F0 contour, and a perceptual experiment was conducted to examine its impact on the quality of the EL speech. RESULTS: The results of perceptual experiments showed that modulating the F0 of EL speech using a linear relationship between amplitude and frequency made it significantly more natural sounding than EL speech with constant F0. CONCLUSIONS: The current study provides preliminary support for amplitude-based control of F0 in EL speech.

Development and testing of a portable vocal accumulator.

This research note describes the design and testing of a device for unobtrusive, long-term ambulatory monitoring of voice use, named the Portable Vocal Accumulator (PVA). The PVA contains a digital signal processor for analyzing input from a neck-placed miniature accelerometer. During its development, accelerometer recordings were obtained from 99 participants with normal or dysphonic voices. The recordings were used to (a) test the specifications and capabilities of the PVA for monitoring normal and dysphonic voices and (b) explore potentially useful displays for the large quantity of data generated by long-term monitoring. The current prototype PVA is pocket-sized (12 x 8.5 x 2 cm), lightweight (200 g), and capable of sampling 11 hr of voice-use data, including estimates of fundamental frequency, sound pressure level, and phonation duration.

A quasiarticulatory approach to controlling acoustic source parameters in a Klatt-type formant synthesizer using HLsyn.

The HLsyn speech synthesizer uses models of the vocal tract to map higher-level quasiarticulatory parameters to the acoustic parameters of a Klatt-type formant synthesizer. The benefits of this system are several. In addition to requiring a relatively small number of parameters, the HLsyn model includes constraints on source-filter relations that occur naturally during speech production. Such constraints help to prevent combinations of sources and filter that are impossible to achieve with the human vocal tract. Thus, HLsyn could lead to reductions in the complexity of formant synthesis and result in better quality synthesis. HLsyn can also be a useful tool for speech-science education and speech research. This paper focuses on the generation of acoustic sources in HLsyn. Described in detail are the equations and methods used to estimate Klatt-type source parameters from HLsyn parameters. Several examples illustrating the generation of source parameters for obstruents (voiced and voiceless) and sonorants are provided. Future papers will describe the filtering components of HLsyn.

Toward a model for lexical access based on acoustic landmarks and distinctive features.

This article describes a model in which the acoustic speech signal is processed to yield a discrete representation of the speech stream in terms of a sequence of segments, each of which is described by a set (or bundle) of binary distinctive features. These distinctive features specify the phonemic contrasts that are used in the language, such that a change in the value of a feature can potentially generate a new word. This model is a part of a more general model that derives a word sequence from this feature representation, the words being represented in a lexicon by sequences of feature bundles. The processing of the signal proceeds in three steps: (1) Detection of peaks, valleys, and discontinuities in particular frequency ranges of the signal leads to identification of acoustic landmarks. The type of landmark provides evidence for a subset of distinctive features called articulator-free features (e.g., [vowel], [consonant], [continuant]). (2) Acoustic parameters are derived from the signal near the landmarks to provide evidence for the actions of particular articulators, and acoustic cues are extracted by sampling selected attributes of these parameters in these regions. The selection of cues that are extracted depends on the type of landmark and on the environment in which it occurs. (3) The cues obtained in step (2) are combined, taking context into account, to provide estimates of "articulator-bound" features associated with each landmark (e.g., [lips], [high], [nasal]). These articulator-bound features, combined with the articulator-free features in (1), constitute the sequence of feature bundles that forms the output of the model. Examples of cues that are used, and justification for this selection, are given, as well as examples of the process of inferring the underlying features for a segment when there is variability in the signal due to enhancement gestures (recruited by a speaker to make a contrast more salient) or due to overlap of gestures from neighboring segments.