Details

Autor(en) / Beteiligte

• Tokuda, Keiichi
• Zen, Heiga

Titel

Directly modeling voiced and unvoiced components in speech waveforms by neural networks

Ist Teil von

• 2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2016, p.5640-5644

Ort / Verlag

IEEE

Erscheinungsjahr

2016

Quelle

IEEE Electronic Library Online

Beschreibungen/Notizen

• This paper proposes a novel acoustic model based on neural networks for statistical parametric speech synthesis. The neural network outputs parameters of a non-zero mean Gaussian process, which defines a probability density function of a speech waveform given linguistic features. The mean and covariance functions of the Gaussian process represent deterministic (voiced) and stochastic (unvoiced) components of a speech waveform, whereas the previous approach considered the unvoiced component only. Experimental results show that the proposed approach can generate speech waveforms approximating natural speech waveforms.