all AI news
Post-decoder Biasing for End-to-End Speech Recognition of Multi-turn Medical Interview
March 4, 2024, 5:47 a.m. | Heyang Liu, Yu Wang, Yanfeng Wang
cs.CL updates on arXiv.org arxiv.org
Abstract: End-to-end (E2E) approach is gradually replacing hybrid models for automatic speech recognition (ASR) tasks. However, the optimization of E2E models lacks an intuitive method for handling decoding shifts, especially in scenarios with a large number of domain-specific rare words that hold specific important meanings. Furthermore, the absence of knowledge-intensive speech datasets in academia has been a significant limiting factor, and the commonly used speech corpora exhibit significant disparities with realistic conversation. To address these challenges, …
abstract arxiv asr automatic speech recognition cs.cl cs.sd decoder decoding domain e2e eess.as hybrid interview medical optimization recognition speech speech recognition tasks type words
More from arxiv.org / cs.CL updates on arXiv.org
Jobs in AI, ML, Big Data
AI Research Scientist
@ Vara | Berlin, Germany and Remote
Data Architect
@ University of Texas at Austin | Austin, TX
Data ETL Engineer
@ University of Texas at Austin | Austin, TX
Lead GNSS Data Scientist
@ Lurra Systems | Melbourne
Senior Machine Learning Engineer (MLOps)
@ Promaton | Remote, Europe
Data Analyst (Digital Business Analyst)
@ Activate Interactive Pte Ltd | Singapore, Central Singapore, Singapore