Convolutive Weighted Multichannel Wiener Filter Front-end for Distant Automatic Speech Recognition in Reverberant Multispeaker Scenarios.
Mieszko FrasMarcin WitkowskiKonrad KowalczykPublished in: INTERSPEECH (2022)
Keyphrases
- speech signal
- automatic speech recognition
- wiener filter
- additive noise
- single channel
- speech enhancement
- linear prediction
- speech recognition
- image restoration
- noisy environments
- noise reduction
- denoising
- vocal tract
- hidden markov models
- conversational speech
- point spread function
- broadcast news
- signal to noise ratio
- non stationary
- speech retrieval
- multi frame
- multi channel
- pattern recognition
- sound source
- noisy images
- probabilistic model