Missing feature theory and probabilistic estimation of clean speech components for robust speech recognition.
Philippe ReneveyAndrzej DrygajloPublished in: EUROSPEECH (1999)
Keyphrases
- speech recognition
- noisy environments
- noisy speech
- speech signal
- speech synthesis
- speech recognizer
- language model
- hidden markov models
- automatic speech recognition
- speaker identification
- speech processing
- pattern recognition
- cepstral coefficients
- speech recognition technology
- speech recognition systems
- handwriting recognition
- speech recognizers
- word error rate
- speech enhancement
- isolated word
- speech recognition errors
- speaker independent
- speaker recognition
- recognition engine
- probabilistic model
- acoustic models
- spectral subtraction
- voice activity detection
- neural network
- keyword spotting
- vocal tract
- speaker diarization
- signal to noise ratio
- bayesian networks
- machine learning