Exploiting low-dimensional structures to enhance DNN based acoustic modeling in speech recognition.
Pranay DigheGil LuyetAfsaneh AsaeiHervé BourlardPublished in: ICASSP (2016)
Keyphrases
- speech recognition
- speech recognition systems
- low dimensional
- speaker independent
- speech recognizers
- hidden markov models
- language model
- high dimensional
- speech recognizer
- acoustic models
- automatic speech recognition
- speech synthesis
- noisy speech
- speech recognition technology
- speech processing
- speech signal
- pattern recognition
- noisy environments
- speech understanding
- speaker identification
- dimensionality reduction
- speech retrieval
- audio visual speech recognition
- cepstral coefficients
- keyword spotting
- acoustic features
- speaker verification
- noise reduction
- principal component analysis