Conference proceeding
Unsupervised learning of acoustic features via deep canonical correlation analysis
2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Vol.2015-, pp.4590-4594
04/01/2015
DOI: 10.1109/ICASSP.2015.7178840
Abstract
It has been previously shown that, when both acoustic and articulatory training data are available, it is possible to improve phonetic recognition accuracy by learning acoustic features from this multi-view data with canonical correlation analysis (CCA). In contrast with previous work based on linear or kernel CCA, we use the recently proposed deep CCA, where the functional form of the feature mapping is a deep neural network. We apply the approach on a speaker-independent phonetic recognition task using data from the University of Wisconsin X-ray Microbeam Database. Using a tandem-style recognizer on this task, deep CCA features improve over earlier multi-view approaches as well as over articulatory inversion and typical neural network-based tandem features. We also present a new stochastic training approach for deep CCA, which produces both faster training and better-performing features.
Details
- Title: Subtitle
- Unsupervised learning of acoustic features via deep canonical correlation analysis
- Creators
- Weiran Wang - Toyota Technological Institute at ChicagoRaman Arora - Johns Hopkins UniversityKaren Livescu - Toyota Technological Institute at ChicagoJeff A. Bilmes - University of Washington
- Resource Type
- Conference proceeding
- Publication Details
- 2015 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Vol.2015-, pp.4590-4594
- Publisher
- IEEE
- DOI
- 10.1109/ICASSP.2015.7178840
- ISSN
- 1520-6149
- eISSN
- 2379-190X
- Language
- English
- Date published
- 04/01/2015
- Academic Unit
- Computer Science
- Record Identifier
- 9984696721602771
Metrics
1 Record Views