Whisper STT

Configure the built-in Whisper speech-to-text provider. Browser microphone device selection stays local to the Web UI and is managed from the plugin status panel.

Model Size

Whisper model variant loaded by the backend runtime.

TinyBaseSmallMediumLargeTurbo

Language

Language hint forwarded to Whisper. Use auto to let Whisper detect it.

Voice Message Handling

Choose whether final transcriptions are sent immediately or left in the composer for review.

Send immediatelyDraft in composer

Silence Threshold

Minimum detected signal before voice input switches from listening to recording.

Silence Duration

Milliseconds of silence required before the recording enters the waiting phase.

Waiting Timeout

Milliseconds to wait after silence before audio is sent for transcription.