Evaluation of noisy speech recognition based on noise reduction and acoustic model adaptation on the Aurora2 tasks.
Masakiyo FujimotoYasuo ArikiPublished in: INTERSPEECH (2002)
Keyphrases
- noisy environments
- noise reduction
- speech recognition
- speech enhancement
- speech recognition systems
- noisy speech
- speaker independent
- speaker verification
- speaker identification
- signal to noise ratio
- edge preserving
- noise free
- speech recognizers
- edge detection
- automatic speech recognition
- speech signal
- background noise
- acoustic models
- language model
- speech processing
- speech synthesis
- hidden markov models
- noise level
- speech recognizer
- noise removal
- speech retrieval
- speech recognition technology
- wiener filter
- word recognition
- visual speech
- acoustic features
- speaker recognition
- machine learning
- median filter
- pattern recognition
- mel frequency cepstral coefficients
- audio signal
- spectral subtraction
- denoising