A new method for speech denoising and robust speech recognition using probabilistic models for clean speech and for noise.
Hagai AttiasLi DengAlex AceroJohn C. PlattPublished in: INTERSPEECH (2001)
Keyphrases
- probabilistic model
- noisy environments
- noisy speech
- speech enhancement
- denoising
- speech recognition
- speaker recognition
- computationally efficient
- hidden markov models
- text to speech
- image noise
- speech signal
- automatic speech recognition
- linear prediction
- gaussian noise
- signal to noise ratio
- spectral features
- error rate
- feature set
- speech synthesis
- edge detection
- fundamental frequency
- speech quality
- hybrid method
- missing data
- multiscale