Gemini 3.5 Transcribe targets more precise, intelligent speech-to-text
A new speech-to-text model, Gemini 3.5 Transcribe, is designed for real-time transcription and voice interactions. It converts raw audio into formatted text, handling background noise, specialized terminology, self-corrections and filler words. The model can also adapt to custom vocabulary and automatically detect and transcribe more than 85 languages, including regional accents and dialects.