LLM Digest

Story

arxiv_cs_cl · Nov 12, 2026 · paper

Source brief

Beyond Word Error Rate: A Failure Taxonomy of Streaming Dictation ASR

arxiv.orgNov 12, 2026
original source linked

In brief

Speech recognition (ASR) systems are usually compared by word error rate (WER) on recorded files, after punctuation, capital letters and number format are removed. In dictation, the audio is streamed live and the text goes into the user's message as written, so what matters is wh

Read the original at arxiv.org →Open in live feed Read that day’s brief

Beyond Word Error Rate: A Failure Taxonomy of Streaming Dictation ASR

Earlier in this thread 4 items

Measuring the Verifier's Floor: How Much Apparent Rare-Word Text-to-Speech Error Is ASR Error?

Shared Item-Level Error Limits the Aggregation of Small Language Models in Fermi Estimation

[AINews] Mercor $750m Series D at $20B valuation, $3B run rate

WordInferBench: Can Agents Infer How to Edit from Word Document Context?