SELM: Speech Enhancement using Discrete Tokens and Language Models.
Ziqian WangXinfa ZhuZihan ZhangYuanjun LvNing JiangGuoqing ZhaoLei XiePublished in: ICASSP (2024)
Keyphrases
- language model
- speech enhancement
- speech recognition
- speech signal
- language modeling
- noisy environments
- document retrieval
- noise reduction
- probabilistic model
- n gram
- retrieval model
- signal to noise ratio
- single channel
- information retrieval
- query expansion
- language modelling
- vocal tract
- smoothing methods
- speech synthesis
- statistical language models
- linear prediction
- language models for information retrieval
- additive noise
- test collection
- document ranking
- background noise
- wiener filter
- relevance model
- automatic speech recognition
- edge detection
- hidden markov models
- signal processing
- image segmentation