When vowels are synthesised by means of a source-filter model, a delta-pulse train is often used as a source signal. Although breathiness can to some extent be simulated by using a sophisticated glottal-source model, a more complete simulation of breathiness requires the addition of aspiration noise. When stationary noise is used, however, the noise is to a large extent perceived as coming from a separate sound source which hardly contributes to the breathy timbre of the vowel. It will be shown that this problem can be solved by using noise with a temporal envelope of the same periodicity as the pulse train. In a simple source-filter model, a combination of lowpass-filtered pulses and -synchronous highpass-filtered noise bursts of equal energy was used as source signal. In this way, the noise was no longer perceived as a separate sound, but integrated perceptually with the strictly periodic part of the signal, thus affecting a natural-sounding breathiness.
Cite as: Hermes, D.J. (1990) Synthesis of breathy vowels. Proc. ESCA Workshop on Speaker Characterization in Speech Technology, 121-126
@inproceedings{hermes90_scst,
author={Dik J. Hermes},
title={{Synthesis of breathy vowels}},
year=1990,
booktitle={Proc. ESCA Workshop on Speaker Characterization in Speech Technology},
pages={121--126}
}