Google launches Gemini 3.5 Transcribe speech-to-text model

Google has released Gemini 3.5, a Transcribe speech-to-text model that automatically removes filler words like "um" and "uh," as well as self-correcting text, generating more fluent text. Compared to its predecessor, Chirp 3, it's approximately 70% faster, with a real-time speech error rate reduced to 5.5%. The model supports 85 languages and up to three pre-recorded audio tracks and is already used in the "Rambler" feature of the Gboard keyboard on the Pixel 11.