Exploration of Multi-lingual Vocabulary for Context Biasing in SpeechLLM
Chaurasia, Arunima
Automatic speech recognition (AR) systems reliably transcribe common words but remain prone to errors on named entities and other rare words. This challenge is particularly pronounced in multilingual settings, where English entities frequently appear in German, Spanish, French, and Italian speech. Existing contextual...
- Fraunhofer-Gesellschaft
- English
- DOI: 10.24406/publica-10189