Batch Captions with Local Whisper

Batch Captions with Local Whisper

Drop a folder of recordings and get subtitle files back. Whisper runs locally on your machine: no API keys, no upload limits, no per-minute charges.

How It Works

1. Add files to the queue. Drag video or audio files in, or use the file picker. Supported formats: MP4, MKV, MOV, AVI, WebM, WAV, MP3, OGG, FLAC.

2. Pick a Whisper model. The same models available in the captions editor are available here: from tiny (fast, less accurate) to large (slow, most accurate). Quantized variants offer a good speed/accuracy trade-off.

3. Set the language. Choose auto-detect or pin to one of six languages (English, Spanish, French, German, Italian, Portuguese). Enable “Translate to English” if you want English output from non-English audio.

4. Choose output formats. Check any combination of SRT, VTT, TXT, and JSON. All selected formats are generated for each file. Output files are written next to the original with matching names (e.g., meeting.mp4 produces meeting.srt, meeting.vtt).

5. Adjust phrase length. The slider controls how many words appear per subtitle line. Short phrases (2-4 words) work well for vertical video captions. Longer phrases (8-12 words) suit documentation or accessibility subtitles.

6. Start the queue. Files process one at a time. Only the active file’s progress bar moves. The progress tracks two phases: audio extraction (quick, ~2%) and Whisper transcription (2-99%, with timestamps appearing in real time).

Key Details

One file at a time. Processing is sequential: Whisper uses significant CPU, and running multiple instances would slow everything down. The queue handles ordering automatically.

Cancel stops the current file, not the queue. If you cancel mid-transcription, the active file resets to pending. Remaining files stay queued. You can restart anytime.

Errors don’t break the queue. If a file fails (corrupt audio, unsupported codec), it’s marked as error and the queue moves to the next file. You can retry failed items later.

What’s Next

• Audio Extraction: extract audio tracks before transcription for faster processing
• Noise Removal: clean up noisy recordings before transcribing
• Batch Utilities Overview: all four batch processing tools