Google's AI transcription removes filler words

Google has released an update for its Gemini Audio service, introducing new AI-driven transcription features. The new Gemini 3.5 Transcribe model is capable of automatically identifying specialized jargon and supports more than 85 different languages.

According to Google, this new transcription model represents a significant advancement compared to its previous Chirp 3 model. Improvements have been made particularly in multilingual performance and reducing word error rates. The service also allows for more natural editing using voice commands.

Gemini 3.5 Transcribe follows the recently launched Gemini 3.5 Live Translate feature. The company is still awaiting the release of the Gemini 3.5 Pro model, which was promised for June.

The technology enhances speech recognition accuracy and scope, reducing the need for users to manually correct automatically generated text.