Presentation | 2021-03-03 [Poster Presentation] A unified source-filter network for neural vocoder Reo Yoneyama, Yi-Chiao Wu, Tomoki Toda, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | In this paper, we propose a method to develop a neural vocoder using a single network based on the source-filter theory. A neural vocoder makes it possible to generate high-quality speech waveforms by applying a deep learning framework to direct speech waveform modeling. On the other hand, its controllability tends to be lower compared to that of a traditional vocoder due to the use of a totally data-driven framework. To alleviate this issue, there have been studied other neural vocoding frameworks consisting of a source excitation part and a resonance filtering part as in a traditional vocoding framework and applying a parametric model to one of these two parts. The use of a part of the traditional approximations is effective for improving controllability of neural vocoder. However, the resulting controllability is still insufficient, and this framework also causes an adverse effect on sound quality degradation compared to the totally data-driven framework. Towards the development of a better neural vocoder, we propose "a unified source-filter network" as a novel neural vocoding framework using a single network. The proposed network consists of cascaded two networks corresponding to the source excitation part and the resonance filtering part, making it possible to optimize all network parameters using a unified training criterion. Moreover, we try to optimize the source excitation network to generate reasonable source excitation signals by applying an additional constraint to its output. Our experimental results have demonstrated that the proposed method can improve $F_0$ controllability compared to the neural source-filter as one of the conventional neural vocoding methods. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | speech synthesis / source-filter model / neural vocoder |
Paper # | EA2020-69,SIP2020-100,SP2020-34 |
Date of Issue | 2021-02-24 (EA, SIP, SP) |
Conference Information | |
Committee | EA / US / SP / SIP / IPSJ-SLP |
---|---|
Conference Date | 2021/3/3(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Online |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | Speech, Engineering/Electro Acoustics, Signal Processing, Ultrasonics, and Related Topics |
Chair | Kenichi Furuya(Oita Univ.) / Hikaru Miura(Nihon Univ.) / Hisashi Kawai(NICT) / Kazunori Hayashi(Kyoto Univ.) / 北岡 教英(豊橋技科大) |
Vice Chair | Yoshinobu Kajikawa(Kansai Univ.) / Kentaro Matsui(NHK) / Jun Kondo(Shizuoka Univ.) / Yoshikazu Koike(Shibaura Inst. of Tech.) / / Yukihiro Bandou(NTT) / Toshihisa Tanaka(Tokyo Univ. Agri.&Tech.) |
Secretary | Yoshinobu Kajikawa(Univ. of Tokyo) / Kentaro Matsui(NTT) / Jun Kondo(Doshisha Univ.) / Yoshikazu Koike(Tohoku Univ.) / (Univ. of Tokyo) / Yukihiro Bandou(Waseda Univ.) / Toshihisa Tanaka(Hosei Univ.) / (Waseda Univ.) |
Assistant | Yukou Wakabayashi(Tokyo Metropolitan Univ.) / Tatsuya Komatsu(LINE) / Shinnosuke Hirata(Tokyo Inst. of Tech.) / Yusuke Ijima(NTT) / Yuichi Tanaka(Tokyo Univ. Agri.&Tech.) |
Paper Information | |
Registration To | Technical Committee on Engineering Acoustics / Technical Committee on Ultrasonics / Technical Committee on Speech / Technical Committee on Signal Processing / Special Interest Group on Spoken Language Processing |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | [Poster Presentation] A unified source-filter network for neural vocoder |
Sub Title (in English) | |
Keyword(1) | speech synthesis |
Keyword(2) | source-filter model |
Keyword(3) | neural vocoder |
1st Author's Name | Reo Yoneyama |
1st Author's Affiliation | Nagoya University(Nagoya Univ.) |
2nd Author's Name | Yi-Chiao Wu |
2nd Author's Affiliation | Nagoya University(Nagoya Univ.) |
3rd Author's Name | Tomoki Toda |
3rd Author's Affiliation | Nagoya University(Nagoya Univ.) |
Date | 2021-03-03 |
Paper # | EA2020-69,SIP2020-100,SP2020-34 |
Volume (vol) | vol.120 |
Number (no) | EA-397,SIP-398,SP-399 |
Page | pp.pp.57-62(EA), pp.57-62(SIP), pp.57-62(SP), |
#Pages | 6 |
Date of Issue | 2021-02-24 (EA, SIP, SP) |