VALL-E 2: Neural Codec Language Models are Human Parity Zero-Shot Text to Speech Synthesizers.
Sanyuan ChenShujie LiuLong ZhouYanqing LiuXu TanJinyu LiSheng ZhaoYao QianFuru WeiPublished in: CoRR (2024)
Keyphrases
- language model
- text to speech
- language modeling
- speech recognition
- probabilistic model
- information retrieval
- document retrieval
- speech synthesis
- n gram
- test collection
- retrieval model
- query expansion
- statistical language models
- language modelling
- programming tool
- context sensitive
- language models for information retrieval
- prosodic features
- smoothing methods
- text to speech synthesis
- word processing
- translation model
- term dependencies
- document ranking
- vector space model
- pattern recognition
- ad hoc information retrieval
- machine learning
- query specific
- relevance model
- document length
- relevant documents
- video sequences