contextual-biasing

[ICASSP 2022] Improving End-to-End Contextual Speech Recognition with Fine-Grained Contextual Knowledge Selection

方言语音里的地址识别：给 Qwen3-ASR-1.7B 补上地址先验，零训练。多音系路由 + 加权拼音距离 + 分级地址库检索 + 带拒识的打分决策；配套分阶段归因评测（负样本 / oracle 上限 / 回归）与参数调参闭环（自动出提案，人 apply 才生效）

How much of Deepgram Nova-3's rare-word gain can keyterms pulled automatically from a company's 10-K recover? Measured on Rev's Earnings-21 against Rev's oracle list.

Local-first Vietnamese/English streaming speech-to-text PWA with private voice personalization