Building State-of-the-art Distant Speech Recognition Using the CHiME-4 Challenge with a Setup of Speech Enhancement Baseline.
Szu-Jui ChenAswin Shanmugam SubramanianHainan XuShinji WatanabePublished in: INTERSPEECH (2018)
Keyphrases
- speech enhancement
- noisy environments
- speech signal
- noise reduction
- single channel
- signal to noise ratio
- vocal tract
- linear prediction
- speech recognition
- noisy speech
- multi channel
- wiener filter
- background noise
- additive noise
- hidden markov models
- automatic speech recognition
- smoothing algorithm
- edge preserving
- speaker identification
- low contrast
- high frequency
- maximum likelihood