Improved training of excitation for HMM-based parametric speech synthesis.
Yoshinori ShigaTomoki TodaShinsuke SakaiHisashi KawaiPublished in: INTERSPEECH (2010)
Keyphrases
- speech synthesis
- speech recognition
- hidden markov models
- text to speech
- prosodic features
- training set
- data sets
- training algorithm
- vocal tract
- training process
- test set
- database
- online learning
- training samples
- supervised learning
- computer software
- back propagation
- feedforward neural networks
- training phase
- training examples
- recognition rate
- case study
- feature selection
- real time