Kochanski Greg, Shih Chilin
Oxford University, Phonetics Laboratory, Oxford OX1 2JF, United Kingdom.
J Acoust Soc Am. 2003 Oct;114(4 Pt 1):2206-16. doi: 10.1121/1.1608964.
A novel, noninvasive experiment is proposed that reliably shows the strength of glottal oscillations. The quasi-glottogram (QGG) signal is generated from a microphone array that is trained to approximate the electroglottogram signal. The QGG may be useful to improve estimates of whether speech is voiced, to quantify partial voicing, and to reduce the phoneme effect when measuring the amplitude of speech signals. The technique is well adapted to the generation of text-to-speech systems, as it allows an estimate of the glottal flow during undisturbed, natural speech. For prosody studies, it can be used to provide an estimate of amplitude which is relatively unaffected by changes in phonemes, and is at least as reliable as standard estimators of amplitude.
提出了一种新颖的非侵入性实验,该实验能可靠地显示声门振荡的强度。准声门图(QGG)信号由经过训练以近似电子声门图信号的麦克风阵列生成。QGG可能有助于改进对语音是否浊音的估计、量化部分浊音以及在测量语音信号幅度时减少音素效应。该技术非常适合文本转语音系统的生成,因为它允许在无干扰的自然语音期间估计声门气流。对于韵律研究,它可用于提供相对不受音素变化影响的幅度估计,并且至少与标准幅度估计器一样可靠。