Reconciling pronunciation differences between the front-end and the back-end in the IBM speech synthesis system.
Wael HamzaEllen EideRaimo BakisPublished in: INTERSPEECH (2004)
Keyphrases
- back end
- speech synthesis
- speech recognition
- user friendly
- vocal tract
- text to speech
- hidden markov models
- statistically significant
- speech signal
- data management
- prosodic features
- data types
- pattern recognition
- language model
- automatic speech recognition
- noisy environments
- version control
- neural network
- speech corpus
- building blocks
- control program
- nearest neighbor
- face recognition
- information retrieval
- detailed design
- data mining
- database