§ models · storyline
Google Gemini 3.5 Transcribe auto-corrects speech in 85 languages
Google launches Gemini 3.5 Transcribe, a speech-to-text model supporting 85 languages with filler-word removal, real-time correction, a 4% word error rate, and 70% lower latency than Chirp 3.
Google's new Gemini 3.5 Transcribe recognizes over 85 languages, strips filler words, and corrects slips of the tongue in real time. It hits a 4.0 percent word error rate in streaming mode, with 70 percent lower latency than its predecessor, Chirp 3.
Through function calling, the model can hand off tasks to other Gemini models. The article Google's Gemini 3.5 Transcribe turns speech to text in 85 languages while auto-correcting your verbal stumbles appeared first on The Decoder.
§ sources1 publication · timeline below