AI Briefing
KO
Pick

Google Releases Gemini 3.5 Transcribe to Clean Up Background Noise and Stuttering

·2026.08.28 10:20

Key point

Google has released Gemini 3.5 Transcribe, a high-precision speech-to-text model that converts audio into clean text by handling background noise, technical jargon, and stuttering.

Details

Google has released a high-precision speech-to-text model that converts audio into clean text by handling background noise, technical jargon, stuttering, and self-corrections.

The model is available in two versions. gemini-3.5-transcribe-live supports bidirectional streaming with sub-1-second latency via the Live API, while gemini-3.5-transcribe is specialized for transcribing recorded audio.

googlegeminispeech-to-texttranscriptionlive-api
Read originalShare on Threads

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.

Related

Google AI BlogGoogle Releases Gemini 3.8 Flash TTS: Voice Customization via Natural Language PromptsGoogle AI BlogGoogle Integrates New Connected Apps Like Adobe and Peloton into GeminiHacker NewsGoogle Launches First Suncatcher Orbital AI Data Center Test on October 1ElevenLabsElevenLabs details Zoom transcription methods using native tools and Scribe v2 APITLDR AIGoogle Discloses Gemini Model Attempted to Hack Third-Party Systems After Escaping Test EnvironmentRedditGoogle Gemini AI Reported to Access Protected Systems