Presentation | 2020-01-22 [Poster Presentation] Speech features obtained from similarities between the input and output of a DNN-based VAD. Nozomi Shigaraki, Kei Yamamori, Suci Dwijayanti, Masato Miyoshi, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | We have been studying Voice activity detection (VAD) using a deep neural network (DNN). Log power spectra (LPS) and Speech period candidates (SPCs) are input into the DNN, where SPCs are numerically calculated from the 1st and 2nd derivative sequences of the LPS. We are now aiming to obtain speech features from similarities between SPCs and VAD decision. In this report, we will discuss Fundamental frequencies (F0) which are estimated from long term (each speech frames) and short term (japanese mora frames) similarities. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | Voice activity detection (VAD) / Speech period candidates / Log power spectra / Speech features / Fundamental frequency |
Paper # | EA2019-95 |
Date of Issue | 2020-01-15 (EA) |
Conference Information | |
Committee | EA / US |
---|---|
Conference Date | 2020/1/22(2days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | Doshisha Univ. |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | [Joint Meeting on Acoustics and Ultrasonics Subsociety] Engineering/Electro Acoustics, Ultrasonics, and Related Topics |
Chair | Kenichi Furuya(Oita Univ.) / Jun Kondo(Shizuoka Univ.) |
Vice Chair | Suehiro Shimauchi(Kanazawa Inst. of Tech.) / Shigeto Takeoka(Shizuoka Inst. of Science and Tech.) / Shinichi Takeuchi(Toin Univ. of Yokohama) / Hikaru Miura(Nihon Univ.) |
Secretary | Suehiro Shimauchi(NHK) / Shigeto Takeoka(Univ. of Tokyo) / Shinichi Takeuchi(Kanagawa Univ.) / Hikaru Miura(Doshisha Univ.) |
Assistant | Keisuke Imoto(Ritsumeikan Univ.) / Daisuke Morikawa(Toyama Pref Univ.) / Mototaka Arakawa(Tohoku Univ.) |
Paper Information | |
Registration To | Technical Committee on Engineering Acoustics / Technical Committee on Ultrasonics |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | [Poster Presentation] Speech features obtained from similarities between the input and output of a DNN-based VAD. |
Sub Title (in English) | |
Keyword(1) | Voice activity detection (VAD) |
Keyword(2) | Speech period candidates |
Keyword(3) | Log power spectra |
Keyword(4) | Speech features |
Keyword(5) | Fundamental frequency |
1st Author's Name | Nozomi Shigaraki |
1st Author's Affiliation | Kanazawa University(Kanazawa Univ.) |
2nd Author's Name | Kei Yamamori |
2nd Author's Affiliation | Kanazawa University(Kanazawa Univ.) |
3rd Author's Name | Suci Dwijayanti |
3rd Author's Affiliation | Sriwijaya University(Sriwijaya Univ.) |
4th Author's Name | Masato Miyoshi |
4th Author's Affiliation | Kanazawa University(Kanazawa Univ.) |
Date | 2020-01-22 |
Paper # | EA2019-95 |
Volume (vol) | vol.119 |
Number (no) | EA-375 |
Page | pp.pp.67-72(EA), |
#Pages | 6 |
Date of Issue | 2020-01-15 (EA) |