[Feature]: Add Whisper transcription progress reporting with real-time percentage updates via Tauri events

[prince-pokharna]

Summary

Knowte runs Whisper transcription locally using whisper.cpp, which for a 1-hour lecture with larger models (medium, large) can take several minutes. The current UI shows a generic loading/processing state with no indication of transcription progress — no percentage complete, no estimated time remaining, no indication of which stage (loading model, processing audio, finalising transcript) the application is in. For long audio files this creates a poor and anxious user experience.

Problem

• The transcription process runs as an opaque background task with no progress feedback to the frontend.
• whisper.cpp supports progress callbacks via its C API (whisper_full_params.progress_callback), but this is not currently wired to the Tauri event system.
• Users processing long lectures have no way to distinguish between "still working" and "silently crashed".
• On machines where the model needs to be loaded from disk for the first time, the startup latency can be 15–30 seconds before any audio is processed, with zero feedback.

Impact

• Users cannot estimate how long to wait, leading to uncertainty and premature application restarts.
• Silent failures (e.g., OOM during large model loading) look identical to normal processing.
• The absence of progress feedback is a meaningful quality gap for a tool positioned against commercial transcription services.

Proposed Solution

1. Wire Whisper's progress callback in Rust: When invoking whisper_full(), set the progress_callback in whisper_full_params to a closure that emits a Tauri event:

params.progress_callback = Some(|_ctx, _state, progress, _user_data| {
    // progress is 0..=100
    app_handle.emit_all("transcription:progress", progress).ok();
});

2. Frontend listener: In the React component managing transcription state, listen for the transcription:progress event via appWindow.listen:

const unlisten = await appWindow.listen<number>('transcription:progress', (event) => {
  setTranscriptionProgress(event.payload);
});

3. Staged progress UI: Show a progress bar with labelled stages:

   • 0–5%: Loading Whisper model
   • 5–95%: Transcribing audio
   • 95–100%: Post-processing and saving transcript

4. Estimated time remaining: Calculate a simple ETA based on elapsed time and current percentage.

I am happy to implement this in both the Rust Tauri backend and the React frontend. Could you please assign this issue to me?

Labels: enhancement, UX, help wanted, GSSoC 2026

[prince-pokharna]

Hey @maintainers,
Please assign it to me under GSSoC 2026.
/claim /claim-assignment /assign