Quasi-Periodic Parallel WaveGAN Vocoder: A Non-autoregressive Pitch-dependent Dilated Convolution Model for Parametric Speech Generation

18 May 2020  ·  Yi-Chiao Wu, Tomoki Hayashi, Takuma Okamoto, Hisashi Kawai, Tomoki Toda ·

In this paper, we propose a parallel WaveGAN (PWG)-like neural vocoder with a quasi-periodic (QP) architecture to improve the pitch controllability of PWG. PWG is a compact non-autoregressive (non-AR) speech generation model, whose generation speed is much faster than real-time. While PWG is taken as a vocoder to generate speech on the basis of acoustic features such as spectral and prosodic features, PWG-generated speech contains a high fidelity. However, because of the fixed and generic network structure, the pitch accuracy of PWG-generated speech degrades when the PWG vocoder conditioned on the acoustic features including unseen pitch features such as scaled pitches. The proposed QPPWG adopts a pitch-dependent dilated convolution network(PDCNN) module, which introduces the pitch information into PWG via the dynamically changed network architecture, to improve the pitch controllability and speech modeling capability of vanilla PWG. Both objective and subjective evaluation results confirm the higher pitch accuracy and comparable speech quality of QPPWG-generated speech, while the QPPWG model size is only 70% of that of vanilla PWG.

PDF Abstract


  Add Datasets introduced or used in this paper

Results from the Paper

  Submit results from this paper to get state-of-the-art GitHub badges and help the community compare results to other papers.


No methods listed for this paper. Add relevant methods here