Auditory-based Acoustic Distinctive Features and Spectral Cues for Robust Automatic Speech Recognition in Low-SNR Car Environments.
Sid-Ahmed SelouaniHesham TolbaDouglas D. O'ShaughnessyPublished in: HLT-NAACL (2003)
Keyphrases
- distinctive features
- automatic speech recognition
- low snr
- speech sounds
- speech recognition
- acoustic features
- speech signal
- noisy environments
- speech recognizers
- speech retrieval
- word error rate
- speaker adaptation
- conversational speech
- hidden markov models
- speech segments
- broadcast news
- acoustic models
- noise model
- spoken words
- speaker independent
- sound source
- word recognition
- visual information
- higher order
- machine learning