Google has launched Gemini 3.5 Transcribe, a dedicated tool within its Gemini Audio suite designed to clean up spoken audio by removing filler words like ‘ums’ and ‘ahs’. The system supports over 85 languages and handles interruptions or background noise without losing context. This new model sits alongside Gemini 3.5 Live and 3.5 Live Experimental to improve precision for voice-controlled features. It arrives while the broader Gemini 3.5 Pro model remains unavailable, having been delayed from its original June release date. The update focuses on practical transcription accuracy rather than generating new content or summarising text automatically.
The significance lies in the specific utility for note-taking and meeting documentation where clarity matters more than conversational flow. By stripping out verbal tics, the output becomes immediately usable for reading or archiving without requiring manual editing. This approach addresses a common frustration with current speech-to-text tools that often retain every hesitation a speaker makes. The result is a cleaner record of what was actually said, which is valuable for professional or personal use cases.
- Supports over 85 languages with automatic jargon detection
- Removes filler words and handles speech interruptions
- Functions independently of the delayed Gemini 3.5 Pro model
