SpeechT5: Unified-Modal Encoder-Decoder Pre-Training for Spoken Language Processing.
Junyi AoRui WangLong ZhouChengyi WangShuo RenYu WuShujie LiuTom KoQing LiYu ZhangZhihua WeiYao QianJinyu LiFuru WeiPublished in: ACL (1) (2022)
Keyphrases
- language processing
- language understanding
- spoken language
- natural language processing
- video codec
- human language technology
- decoding process
- low complexity
- natural language
- noisy channel
- error control
- distributed video coding
- human language
- machine translation
- rate distortion
- broadcast news
- wyner ziv video coding
- speech recognition
- successive approximation
- bit rate
- discriminative training
- motion estimation
- knowledge representation
- temporal correlation
- information retrieval
- modal logic
- artificial intelligence
- grammar induction
- video coding scheme
- motion compensated
- distributed source coding