Dynamic features in the linear domain for robust automatic speech recognition in a reverberant environment.
Osamu IchikawaTakashi FukudaRyuki TachibanaMasafumi NishimuraPublished in: INTERSPEECH (2009)
Keyphrases
- automatic speech recognition
- speech signal
- noisy environments
- speech recognition
- linear prediction
- word error rate
- word recognition
- non stationary
- vocal tract
- broadcast news
- recognition errors
- hidden markov models
- speaker verification
- speech retrieval
- noise reduction
- acoustic features
- multiscale
- error rate
- spontaneous speech
- language model
- conversational speech
- neural network