An mcmc approach to joint estimation of clean speech and noise for robust speech recognition.
Aleem MushtaqChin-Hui LeePublished in: ICASSP (2013)
Keyphrases
- speech recognition
- noisy speech
- noisy environments
- joint estimation
- speech signal
- spectral subtraction
- speech enhancement
- speech synthesis
- speech processing
- hidden markov models
- automatic speech recognition
- speech recognizer
- speaker identification
- language model
- background noise
- pattern recognition
- digit recognition
- speech recognition systems
- speech recognition technology
- speaker verification
- speaker independent
- word error rate
- recognition engine
- isolated word
- noise reduction
- speech recognizers
- speech recognition errors
- speaker dependent
- keyword spotting
- signal to noise ratio
- vocal tract
- noise level
- voice activity detection
- probabilistic model
- handwriting recognition
- linear prediction