[Feature]: Experimental team radio transcription testing and feedback #851

Prerequisites

• All information in this issue is written in English.

Component

Integration

Problem to Solve

Team radio transcription is an experimental feature. It needs a single place where users can follow its progress, share real-world test results, and report problems before it can be considered stable.

Proposed Solution

Use this issue to track the experimental transcription action for the latest F1 team-radio clip. The feature uses a user-configured Google Gemini AI Task in Home Assistant, keeps the transcript in English, and can present anonymous speaker labels when a clip contains two voices.

Please add a comment with your Home Assistant version, F1 Sensor version or branch, Gemini model, a short description of the clip tested, the result, and any relevant logs with secrets removed. Report incorrect or missing transcripts, speaker-label issues, action setup problems, performance concerns, and unexpected provider costs here.

The feature is manual and experimental. A Google Gemini API key and any associated provider charges are the user's responsibility. Setup instructions, examples, and current limitations are in the beta documentation.

Alternatives

Users can continue to use the existing team-radio audio clips without transcription. Other AI Task providers may be evaluated later, but this experiment focuses on the Home Assistant Google Gemini integration.