Presentation | 2019-12-06 An evaluation of representation learning using phoneme posteriorgrams and data augmentation in speech emotion recognition Shintaro Okada, Atsushi Ando, Tomoki Toda, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | This paper presents a new speech emotion recognition method based on representation learning and data augmentation. To improve the robustness against unseen speech, the conventional representation learning-based emotion recognition method utilizes a latent variable extracted by an unsupervisedly-learned speech reconstruction model to train an emotion recognizer using a limited amount of supervised data. However, the latent variable is expected to include not only an informative factor for emotion recognition but also less informative factors, such as phonetic and speaker information. The proposed method alleviates the effects of these less informative factors on the latent variable. To reduce the effects of a phonetic factor, phonetic posteriorgram (PPG) is provided as an auxiliary input of the reconstruction model in representation learning. Moreover, the effects of a speaker factor is mitigated by data augmentation to generate utterances with various speaker characteristics by using a speech morphing technique. Experimental results show that the proposed representation learning method using PPG outperforms the conventional method. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | speech emotion recognition / representation learning / autoencoder / phoneme posteriorgrams / data augmentation |
Paper # | SP2019-43 |
Date of Issue | 2019-11-29 (SP) |
Conference Information | |
Committee | NLC / IPSJ-NL / SP / IPSJ-SLP |
---|---|
Conference Date | 2019/12/4(3days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | NHK Science & Technology Research Labs. |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | The 6th Natural Language Processing Symposium & The 21th Spoken Language Symposium |
Chair | Takeshi Sakaki(Hottolink) / / Hisashi Kawai(NICT) |
Vice Chair | Mitsuo Yoshida(Toyohashi Univ. of Tech.) / Kazutaka Shimada(Kyushu Inst. of Tech.) / / Akinobu Ri(Nagoya Inst. of Tech.) |
Secretary | Mitsuo Yoshida(Ryukoku Univ.) / Kazutaka Shimada(NTT) / / Akinobu Ri(Kyoto Univ.) / (Waseda Univ.) |
Assistant | Takeshi Kobayakawa(NHK) / Hiroki Sakaji(Univ. of Tokyo) / / Tomoki Koriyama(Univ. of Tokyo) / Yusuke Ijima(NTT) |
Paper Information | |
Registration To | Technical Committee on Natural Language Understanding and Models of Communication / Special Interest Group on Natural Language / Technical Committee on Speech / 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) | An evaluation of representation learning using phoneme posteriorgrams and data augmentation in speech emotion recognition |
Sub Title (in English) | |
Keyword(1) | speech emotion recognition |
Keyword(2) | representation learning |
Keyword(3) | autoencoder |
Keyword(4) | phoneme posteriorgrams |
Keyword(5) | data augmentation |
1st Author's Name | Shintaro Okada |
1st Author's Affiliation | Nagoya University(Nagoya Univ.) |
2nd Author's Name | Atsushi Ando |
2nd Author's Affiliation | Nagoya University/NTT(Nagoya Univ./NTT) |
3rd Author's Name | Tomoki Toda |
3rd Author's Affiliation | Nagoya University(Nagoya Univ.) |
Date | 2019-12-06 |
Paper # | SP2019-43 |
Volume (vol) | vol.119 |
Number (no) | SP-321 |
Page | pp.pp.91-96(SP), |
#Pages | 6 |
Date of Issue | 2019-11-29 (SP) |