Using a pitch-synchronous residual codebook for hybrid HMM/frame selection speech synthesis.
Thomas DrugmanAlexis MoinetThierry DutoitGeoffrey WilfartPublished in: ICASSP (2009)
Keyphrases
- speech synthesis
- speech recognition
- hidden markov models
- text to speech
- vector quantization
- vocal tract
- speech signal
- pattern recognition
- prosodic features
- speech corpus
- visual words
- automatic speech recognition
- noisy environments
- frame rate
- gesture recognition
- image representation
- image compression
- selection algorithm
- language model
- multiresolution
- feature space
- bayesian networks