Google has introduced Gemini 3.5 Transcribe, an AI audio model designed to convert informal spoken thoughts into cleaner, structured written text. The company says it could make dictation more useful across its products and Chrome, reducing the need to type messages, posts and prompts manually.

Google claims the model recognises more than 85 languages, removes filler words, handles self-corrections and can learn custom vocabulary and spellings. It can also identify up to three speakers in recorded audio with word-level timestamps, while supporting details such as order numbers and postcodes; planned integrations include Search Live, Gemini Live, Docs, Keep, Gmail and developer APIs.