Minimum Generation Error Training for HMM-Based Speech Synthesis.
Yi-Jian WuRen-Hua WangPublished in: ICASSP (1) (2006)
Keyphrases
- speech synthesis
- speech recognition
- text to speech
- hidden markov models
- error rate
- training set
- prosodic features
- vocal tract
- supervised learning
- training process
- training samples
- training examples
- test set
- generation process
- minimum cost
- error bounds
- machine learning
- speech corpus
- feedforward neural networks
- automatic speech recognition
- training phase
- online learning
- bayesian networks
- computer vision