Live Caption language pack constraints and automatic switching behavior

On-Device Transcription Limits

Chrome's Live Caption feature relies on on-device machine learning models to provide real-time transcription of audio. To maintain privacy and offline functionality, the browser downloads specific language packs to handle the audio-to-text conversion process.

While the feature supports multiple languages, there is uncertainty regarding how the browser manages transitions between different languages within a single audio stream or across multiple open tabs playing different languages simultaneously.

• Whether the system can dynamically switch language packs without manual user intervention in Settings.
• The impact of concurrent audio streams in different languages on the active transcription model.

Does Chrome support automatic language detection for Live Caption, or must the user manually select the language pack for each session? How does the feature handle mixed-language audio streams?