| Presentation | 2014-12-15 Investigation of Deep Neural Network and Cross-adaptation for Voice Activity Detection in Meeting Speech Akihiro NAKADANI, Longbiao WANG, Atsuhiko KAI, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In voice activity detection(VAD), performance largely decreases under the influence of noise and reverberation. In this paper, we focus on a VAD technique with deep neural network(DNN) framework and propose the environmental adaptation methods of the VAD model. As for the unsupervised adaptation of such discriminative models, utilizing erroneous identification result as target signal often reproduces an error and degrades the performance. As for unsupervised adaptation techniques in ASR systems, cross adaptation method using different types of models, has been proposed. Our cross-adaptation method improves the VAD performance by using recognition output of GMM and SVM which are different from DNN in terms of error tendency for unsupervised adaptation and achieves a robust VAD system capable of adapting the noisy and reverberant environment. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Voice activity detection(VAD) / Deep neural network(DNN) / Cross-adaptation / Noisy and reverberant speech / Environmental adaptation |
| Paper # | SP2014-107 |
| Date of Issue |
| Conference Information | |
| Committee | SP |
|---|---|
| Conference Date | 2014/12/8(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | |
| Vice Chair | |
| Secretary | |
| Assistant | |
| Paper Information | |
| Registration To | Speech (SP) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Investigation of Deep Neural Network and Cross-adaptation for Voice Activity Detection in Meeting Speech |
| Sub Title (in English) | |
| Keyword(1) | Voice activity detection(VAD) |
| Keyword(2) | Deep neural network(DNN) |
| Keyword(3) | Cross-adaptation |
| Keyword(4) | Noisy and reverberant speech |
| Keyword(5) | Environmental adaptation |
| 1st Author's Name | Akihiro NAKADANI |
| 1st Author's Affiliation | Shizuoka University() |
| 2nd Author's Name | Longbiao WANG |
| 2nd Author's Affiliation | Nagaoka University of Technology |
| 3rd Author's Name | Atsuhiko KAI |
| 3rd Author's Affiliation | Shizuoka University |
| Date | 2014-12-15 |
| Paper # | SP2014-107 |
| Volume (vol) | vol.114 |
| Number (no) | 365 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |