Comparative experiments to evaluate the use of auditory-based acoustic distinctive features and formant cues for robust automatic speech recognition in low-SNR car environments.
Hesham TolbaSid-Ahmed SelouaniDouglas D. O'ShaughnessyPublished in: INTERSPEECH (2003)
Keyphrases
- automatic speech recognition
- distinctive features
- low snr
- speech recognition
- speech signal
- acoustic features
- sound source
- speech sounds
- noisy environments
- speech recognizers
- speech segments
- hidden markov models
- acoustic models
- word error rate
- spoken words
- speech recognition systems
- conversational speech
- recognition errors
- broadcast news
- speech retrieval
- speech corpus
- speaker independent
- noise model
- low signal to noise ratio
- pattern recognition
- vocal tract
- word recognition
- speaker identification
- speech recognizer
- speech synthesis
- language model
- neural network
- visual information
- spontaneous speech
- non stationary
- speaker adaptation
- objective function