AI · Research · 2026-11-18 06:40
Researchers at the University of Washington introduced Murmur, a method using hierarchical chunking to optimize long-context ASR inference. Compared to WhisperX, Murmur significantly reduces error rates and achieves 2.5x faster transcription speeds for English and Chinese meetings.
Source: syfi.cs.washington.edu
Auto-generated summary for reference only. RSS