OpenAI says GPT-Transcribe cuts WER to 19.27% on Common Voice across 22 languages
OpenAIDevs · x · 2026-07-29
OpenAI says GPT-Transcribe improves substantially on transcription benchmarks:
- On Common Voice across 22 languages, it posts a 19.27% WER, versus 40.37% for Whisper-1.
- On the Real-World Audio Recording benchmark across nine languages, it reaches 8.98% WER, compared with 15.21% for Whisper-1.
- On the Context Aware ASR benchmark, semantic accuracy rises from 41.6% without free-form context to 45.2% with it.
The post also highlights practical ways builders can improve live transcription: provide free-form context, keywords for names and domain terms, expected input languages, and earlier transcribed turns.
More from Models
- Gemini 4 Argon now available in AI Studio for Google AI Pro subscribers — CurieuxExplorer · 2026-10-17
- Generative UI 101: Building Live Interfaces with Agents on Opus 5.5 and Rivals — EXM7777 · 2026-10-17
- OpenAI's new $500 tier and $200 plan throttling are driving power users away — cybrstg · 2026-10-17
- DeepSeek shows rapid vision gains: agent self-checks screenshots to build 3D scene — cedric_chee · 2026-10-17
- Full FP8 n-gram table visibly improves local Qwen3.8-Flash-Next at zero speed cost — sillynoobhorse · 2026-10-17
- Altman says OpenAI shelved a ready-to-launch model as too risky, warning of AI 'escape' scenarios — Aiden_Tech_Ai · 2026-10-17