Combination of temporal domain SVD based speech enhancement and GMM based speech estimation for ASR in noise - evaluation on the AURORA2 task -.
Masakiyo FujimotoYasuo ArikiPublished in: INTERSPEECH (2003)
Keyphrases
- speech enhancement
- noisy environments
- speech signal
- speaker identification
- speech recognition
- automatic speech recognition
- temporal domain
- noise reduction
- noisy speech
- signal to noise ratio
- speaker recognition
- spectral subtraction
- background noise
- vocal tract
- linear prediction
- spatial domain
- single channel
- additive noise
- spatial and temporal
- hidden markov models
- spatio temporal
- wiener filter
- mel frequency cepstral coefficients
- gaussian mixture model
- optical flow
- temporal consistency
- image processing
- computational complexity
- edge detection
- acoustic features
- speech synthesis
- audio visual
- space time
- multiscale
- temporal information