Gemini 3.5 Transcribe: Google's Most Precise Speech AI Yet
Gemini 3.5 Transcribe is Google's new speech-to-text model, live in public preview with 2.6% word error rate and support for more than 85 languages.
2 articles
Gemini 3.5 Transcribe is Google's new speech-to-text model, live in public preview with 2.6% word error rate and support for more than 85 languages.
ElevenLabs Scribe v2 ships two models: Realtime at 150ms latency for voice agents, and Batch with 48-speaker diarization. 93.5% accuracy in 90+ languages.
