Joint Optimization of Denoising Autoencoder and DNN Acoustic Model Based on Multi-Target Learning for Noisy Speech Recognition.
Masato MimuraShinsuke SakaiTatsuya KawaharaPublished in: INTERSPEECH (2016)
Keyphrases
- speech recognition
- joint optimization
- denoising
- speech recognition systems
- multi target
- noisy environments
- hidden markov models
- language model
- learning algorithm
- speech signal
- speech recognizers
- multi target tracking
- training process
- speaker identification
- sparse representation
- speech synthesis
- pattern recognition
- image processing