Google Launches Gemini 3.5 Transcribe Speech-to-Text Model

Google unveiled Gemini 3.5 Transcribe, an advanced speech-to-text model designed to enhance voice interaction by filtering background noise, managing complex jargon, and removing speech disfluencies. The model introduces multi-speaker recognition and word-level timestamps, offering significant performance gains over its predecessor, Chirp 3.

Data from Artificial Analysis indicates that the time required to finalize transcriptions has improved by 70%. The technology is now integrated into Gboard, Chrome, and the Gemini app. Developers can access the model through the Gemini API in Google AI Studio and the Gemini Enterprise Agent Platform, while platforms including Agora, LangChain, and Vercel can use the Gemini Live API to build voice-driven interfaces.

Early adopters, including Vivo and Intellitek Health, have highlighted the model's accuracy and low latency as primary benefits. Other organizations, such as Lingopal and the Indian Premier League, also praised the system for its language support and speed.