Cepstrum-domain acoustic feature compensation based on decomposition of speech and noise for ASR in noisy environments.
Hong Kook KimRichard C. RosePublished in: IEEE Trans. Speech Audio Process. (2003)
Keyphrases
- noisy environments
- automatic speech recognition
- speech recognition
- noisy speech
- speech enhancement
- speech signal
- speaker verification
- noise reduction
- acoustic features
- digit recognition
- background noise
- speech sounds
- speaker identification
- visual speech
- speech recognition systems
- speech recognizers
- hidden markov models
- voice activity detection
- spectral subtraction
- feature vectors
- pattern recognition
- speech recognizer
- speech synthesis
- linear prediction
- non stationary
- feature set
- formant frequencies
- vocal tract
- spontaneous speech
- broadcast news
- additive noise
- language model
- image processing