New framework boosts Indian language ASR and dialect identification

Researchers have developed a novel multimodal framework to simultaneously enhance Automatic Speech Recognition (ASR) and Dialect Identification (DID) for Indian languages. This approach utilizes a Bottleneck Encoder for dialectal features from Conformer speech representations and a RoBERTa encoder for ASR-generated embeddings, merging them via a gating mechanism. Tested across eight Indian languages and thirty-three dialects, the method achieved an 81.63% average DID accuracy and competitive ASR performance with CER and WER of 4.65% and 17.73%, respectively. AI

IMPACT This research could significantly improve accessibility and usability of speech technology for diverse language communities.

RANK_REASON The cluster contains an academic paper detailing a new methodology for ASR and Dialect Identification. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CL →

• arXiv
• Conformer
• Prasanta Ghosh
• Roberta
• speech recognition

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →