论文信息 - An autoencoder neural-network based low-dimensionality approach to excitation modeling for HMM-based text-to-speech

An autoencoder neural-network based low-dimensionality approach to excitation modeling for HMM-based text-to-speech

HMM-TTS synthesis is a popular approach toward flexible, low-footprint, data driven systems that produce highly intelligible speech. In spite of these strengths, speech generated by these systems exhibit some degradation in quality, attributable to an inadequacy in modeling the excitation signal that drives the parametric models of the vocal tract. This paper proposes a novel method for modeling the excitation as a low-dimensional set of coefficients, based on a non-linear map learned through an autoencoder. Through analysis-and-resynthesis experiments, and a formal listening test, we show that this model produces speech of higher perceptual quality compared to conventional pulse-excited speech signals at the p ≪ 0.01 significance level.

Bhuvana Ramabhadran | Srikanth Vishnubhotla | Raul Fernandez

[1] Junichi Yamagishi,et al. Towards an improved modeling of the glottal source in statistical parametric speech synthesis , 2007, SSW.

[2] Geoffrey E. Hinton,et al. Reducing the Dimensionality of Data with Neural Networks , 2006, Science.

[3] Takashi Saitoh,et al. An automatic pitch-marking method using wavelet transform , 2000, INTERSPEECH.

[4] Miguel Á. Carreira-Perpiñán,et al. On Contrastive Divergence Learning , 2005, AISTATS.

[5] Keiichi Tokuda,et al. Simultaneous modeling of spectrum, pitch and duration in HMM-based speech synthesis , 1999, EUROSPEECH.

[6] Thierry Dutoit,et al. Using a pitch-synchronous residual codebook for hybrid HMM/frame selection speech synthesis , 2009, 2009 IEEE International Conference on Acoustics, Speech and Signal Processing.

[7] Heiga Zen,et al. A trainable excitation model for HMM-based speech synthesis , 2007, INTERSPEECH.

[8] Heiga Zen,et al. Tying covariance matrices to reduce the footprint of HMM-based speech synthesis systems , 2009, INTERSPEECH.

[9] Keiichi Tokuda,et al. On the state definition for a trainable excitation model in HMM-based speech synthesis , 2008, 2008 IEEE International Conference on Acoustics, Speech and Signal Processing.