BESTOW: Efficient and Streamable Speech Language Model with the Best of Two Worlds in GPT and T5.
Zhehuai ChenHe HuangOleksii HrinchukKrishna C. PuvvadaNithin Rao KoluguriPiotr ZelaskoJagadeesh BalamBoris GinsburgPublished in: CoRR (2024)
Keyphrases
- language model
- speech recognition
- language modeling
- word error rate
- n gram
- information retrieval
- document retrieval
- probabilistic model
- retrieval model
- mixture model
- language model for information retrieval
- test collection
- automatic speech recognition
- language modelling
- query terms
- context sensitive
- speech signal
- smoothing methods
- query expansion
- statistical language models
- ad hoc information retrieval
- retrieval effectiveness
- vector space model
- statistical machine translation
- natural language processing