Speaker adaptation using regularization and network adaptation for hybrid MMI-NN/HMM speech recognition.
Jörg RottlandChristoph NeukirchenDaniel WillettGerhard RigollPublished in: EUROSPEECH (1999)
Keyphrases
- speaker adaptation
- speech recognition
- automatic speech recognition
- speaker independent
- speech recognizer
- hidden markov models
- language model
- maximum likelihood
- speech signal
- pattern recognition
- neural network
- speech synthesis
- noisy environments
- speaker dependent
- speaker identification
- keyword spotting
- broadcast news
- speech recognition systems
- vocal tract
- acoustic features
- probabilistic model
- machine learning