Integrating articulatory data in deep neural network-based acoustic modeling
详细信息 查看全文
- 作者：Leonardo Badino ; leonardo.badino@iit.it" class="auth_mail" title="E-mail the corresponding authorAuthor Vitae ; Claudia CanevariAuthor Vitae ; Luciano FadigaAuthor Vitae ; Giorgio MettaAuthor Vitae
- 关键词：DNN-HMM ; Acoustic-to-articulatory mapping ; Deep neural networks ; Acoustic modeling ; Electromagnetic articulography ; Autoencoders
- 刊名：Computer Speech & Language
- 出版年：2016
- 出版时间：March 2016
- 年：2016
- 卷：36
- 期：Complete
- 页码：173-195
- 全文大小：1820 K
文摘
- •
We test strategies to exploit articulatory data in DNN-HMM phone recognition.
- •
Autoencoder-transformed articulatory features produce the best results.
- •
Pre-training of phone classifier DNNs driven by acoustic-to-articulatory mapping.
- •
Utility of articulatory information in noisy conditions and in cross-speaker settings.