ElevenLabs Scribe v3: speech-to-text with 27% fewer errors in 102 languages
Two new speech-to-text models — Scribe v3 for recorded audio and Scribe v3 Live for real-time streams — cutting word error rate 27% vs v2, with real-time speaker diarization, context-aware turn detection and a 10,000-term keyterm library.
Why it matters
A straight accuracy jump on the most-used transcription API in the voice-agent stack, plus a live variant that labels speakers in real time — the piece that makes multi-speaker meeting and call agents actually workable.
What you could build with it
A team building a meeting-notes or voice-agent product could swap in Scribe v3 Live to get cleaner transcripts with reliable speaker labels in real time, improving downstream summaries and action-item extraction without changing their pipeline.
Does it hold up?
Already the default transcriber in the ElevenLabs app and live in ElevenAgents and ElevenAPI, and Coval's independent measurement backs the accuracy claim (2.3% WER, 94 ms latency); a brief launch-day outage of Scribe v3 Live on 11 Nov was resolved in under two hours.
Built with ElevenLabs Scribe v3
- Scribe v3: More Accurate Transcription in 102 Languagesblog · Breakdown of the batch vs live models, the 27% WER cut, and the new diarization and keyterm features.
- ElevenLabs STT, TTS and S2S latency and accuracy benchmarksCoval · Independent daily benchmarks: Scribe v3 Live measured at 94 ms time-to-final-segment and 2.3% WER, 2nd of 32 STT systems.
- ElevenLabs launches Scribe v3 and Scribe v3 Live for messier real-world audioblog · Launch coverage of the two models, the 10,000-term library, and availability across ElevenAPI, ElevenAgents and the app.
Learn more
First spotted on article: source.
More AI models
VoiceStudio: open-source, fully-local ElevenLabs alternative for voice cloning, dubbing, and audiobooksdebpalash/VoiceStudio is a desktop app that runs voice cloning, voice design, video dubbing, dictation, transcription…models · JEV 0.75Edge0-145B: open 145B MoE that runs on a single consumer GPUEdge0 open-sourced 145B and 120B latent-MoE models derived from Qwen3.8-Flash-Next, plus an inference framework that…models · JEV 0.74Strata: run a 125B model on a gaming PCFree, MIT-licensed local inference engine that runs Qwen3.8-Flash-Next (125B MoE, about 6B active per token) on…models · JEV 0.73Google AI Edge Gallery: offline speech recognition and synthesis on-deviceGoogle added fully offline speech recognition and text-to-speech to AI Edge Gallery, with ML Kit APIs on Android and…models · JEV 0.73Soup: fine-tune an 8B LLM on a 4 GB laptop GPU from one YAML fileAn open-source CLI that fine-tunes LLMs from a single YAML file, with layer streaming that trains an 8B model on a 4…models · JEV 0.73Tencent Hunyuan Hy3: open 295B MoE with 21B active under Apache 2.0Tencent open-sourced Hy3, a 295B-parameter mixture-of-experts model with only 21B parameters active per token, under…models · JEV 0.73
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.