●SONNET — Claude Sonnet 5 is now the default model in Claude Code, with a native 1M-token context window and introductory pricing through August 31●CHROME — Claude in Chrome reaches general availability, letting you hand browser work directly to Claude●COWORK — Cowork expands to mobile and web so sessions and files follow you across devices, starting in beta for Max users●DATAVIZ — Claude Code adds a /dataviz skill offering guidance for designing charts and dashboards●AGENTS — Agent workflows gain background notifications, draft PR handoff, and improved failover●ENTERPRISE — Claude Enterprise adds richer admin analytics, model-level entitlements, and spend alerts●SONNET — Claude Sonnet 5 is now the default model in Claude Code, with a native 1M-token context window and introductory pricing through August 31●CHROME — Claude in Chrome reaches general availability, letting you hand browser work directly to Claude●COWORK — Cowork expands to mobile and web so sessions and files follow you across devices, starting in beta for Max users●DATAVIZ — Claude Code adds a /dataviz skill offering guidance for designing charts and dashboards●AGENTS — Agent workflows gain background notifications, draft PR handoff, and improved failover●ENTERPRISE — Claude Enterprise adds richer admin analytics, model-level entitlements, and spend alerts

TAG

Whisper

1 articles

← Back to all tags

Claude API¹ transcription¹ Japanese¹ voice AI¹ quality measurement¹

⬡ API & SDK/2026-07-11Advanced

When Whisper Wrote 'Thanks for Watching' Into the Silence — Measuring and Killing Japanese Transcription Hallucinations

Whisper inserts phantom boilerplate into silent or noisy segments of Japanese audio. Here is how to measure hallucinations with no_speech_prob and avg_logprob, strip silence with VAD, and have Claude validate what remains — with the implementation and the routine that pulled it back.