Transcribe an audio file
Use Audio file for an existing recording. It has its own connection and language, with an independent model in manual mode. Voice and files using the same managed runtime share its selected transcription model. No microphone, recording shortcut, or completed Voice setup is required. Configure its Transcription options first, using a saved connection or local runtime.
| Before you start | Limit |
|---|---|
| File picker accepts | FLAC, MP3, MP4, MPEG, MPGA, M4A, OGG, WAV, or WebM |
| File size | At most 2 GiB; your server may accept less |
| Other recording work | Finish or cancel file transcription before using the microphone |
Your server must decode the selected format; Freehand does not transcode it. WAV is the conservative choice for whisper.cpp. Freehand does not split oversized files for upload. Selecting a file neither uploads nor modifies it.
Transcribe and copy
Section titled “Transcribe and copy”- Open Audio file and choose Choose to select a recording.
- Choose how results arrive. Enable Show text as it arrives when available, or leave it off for a completed result.
- Choose Transcribe. Follow progress and use Cancel to stop active work.
- Copy the completed transcript into your destination. File results never insert automatically because no destination was captured.
Choose Change for a different file or Again to transcribe the same file again. Optional Cleanup runs after a successful completed transcript. Listen is available when text to speech is enabled.
Choose a response mode
Section titled “Choose a response mode”Leave Show text as it arrives off to receive the completed transcript. Completed transcript means text updates are unavailable for the current selection. If you change or fix your server, Try streaming, when offered, enables updates for the next request; it does not upload the file itself.
Enable Show text as it arrives to review updates from a compatible server. The server may still return the complete transcript without intermediate updates. This is an uploaded-file request, separate from live microphone transcription.
If streaming fails, available partial text remains copyable and is not sent for cleanup. Check it for missing words. Some servers cannot signal whether a stream is complete; see file streaming compatibility.
Recover a failed request
Section titled “Recover a failed request”Open Details beside the error, correct the cause, then choose Retry. Copy any useful partial text before starting again.
| Symptom | Next action |
|---|---|
Upload rejected / 413 |
Choose a smaller file or adjust the server or proxy’s upload limit |
| Timed out | Review server load and the file request budget in Request settings |
| Partial or interrupted stream | Review the partial result and deliberately retry after fixing the connection |
Keep a result
Section titled “Keep a result”The current file result is available with history off. It clears when you clear or replace the file, start another transcription, or exit. Temporary request audio is deleted after success, failure, or cancellation; the original file is unchanged.