Acoustic BPE for Speech Generation with Discrete Tokens.
Feiyu ShenYiwei GuoChenpeng DuXie ChenKai YuPublished in: CoRR (2023)
Keyphrases
- speech recognition systems
- acoustic features
- speech sounds
- speech recognition
- emotional speech
- hidden markov models
- automatic speech recognition
- prosodic features
- speech signal
- continuous variables
- generation process
- acoustic signal
- text to speech
- discrete space
- speaker verification
- speaker recognition
- spoken language
- audio visual
- multi stream
- visual speech
- discrete version
- finite number
- language acquisition
- emotion recognition
- line segments
- data sets