Combining spectral feature mapping and multi-channel model-based source separation for noise-robust automatic speech recognition.
Deblin BagchiMichael I. MandelZhongqiu WangYanzhang HeAndrew R. PlummerEric Fosler-LussierPublished in: ASRU (2015)
Keyphrases
- multi channel
- single channel
- source separation
- automatic speech recognition
- feature mapping
- speech recognition
- speech signal
- hidden markov models
- wiener filter
- signal processing
- independent component analysis
- prior information
- blind source separation
- noise reduction
- neural network
- image restoration
- image compression
- image processing