On combining vocal tract length normalisation and speaker adaptation for noise robust speech recognition.
Ramalingam HariharanOlli ViikkiPublished in: EUROSPEECH (1999)
Keyphrases
- speech recognition
- speaker adaptation
- noisy environments
- vocal tract
- speech signal
- speech synthesis
- speech enhancement
- automatic speech recognition
- hidden markov models
- background noise
- speaker independent
- language model
- speech recognizer
- speaker dependent
- maximum likelihood
- pattern recognition
- digit recognition
- speaker identification
- speech recognition systems
- linear prediction
- computer vision
- noise reduction
- signal to noise ratio
- spectral analysis
- text to speech
- em algorithm
- feature selection