Effect of Glottal Pulse Shape on the Quality of Nattural Vowels

Abstract

An experiment is described that investigates listener preferences for speech samples with varying glottal pulse-shape parameters. A male speaker recorded monosyllabic words and a continuous sentence. A pitch-synchronous analysis was carried out by a digital computer on the vowel portions of these samples. For every pitch period, the analysis provided: formant frequencies, waveform of the glottal excitation function, and an accurate pitch-period measurement. In each vowel, the natural glottal excitation function was replaced by a mathematical function with a shape not unlike that of natural glottal waves. The waveform of the artificial function could be modified by varying two parameters analogous to the “opening” and “closing” times of natural glottal pulses. Four opening times and four closing times (fixed relative to individual pitch-period lengths) were selected for experimentation. The artificial functions were substituted for the natural glottal pulses, and the speech wave was reconstituted period by period. Subjective evaluation of the reconstituted speech was carried out by means of computer-controlled paired-comparison tests. Sequential testing strategies were used to reveal the synthetic samples perceptually most similar to the original. The results indicate that an artificial glottal wave having the same analytical specification in every pitch period can yield reconstituted speech of quality comparable to natural speech.