Powered cepstral normalization (p-CN) for robust features in speech recognition.
Chang-Wen HsuLin-Shan LeePublished in: INTERSPEECH (2006)
Keyphrases
- speech recognition
- noisy environments
- speech recognition systems
- automatic speech recognition
- hidden markov models
- speech signal
- language model
- speech synthesis
- speech recognition technology
- pattern recognition
- cepstral coefficients
- low level
- speech understanding
- speech recognizer
- speech processing
- speech recognizers
- keyword spotting
- speaker identification
- speaker independent
- speech retrieval
- feature vectors
- feature extraction
- speaker verification
- mel frequency cepstral coefficients
- classification accuracy
- probabilistic model
- speaker dependent
- feature space