Auto-generated captions

Long Play can automatically generate a caption track from your video's audio. This is completely optional - you can always upload your own caption file instead, or choose not to add captions at all.

How it works

Long Play transcribes the audio using Whisper, an open-source speech recognition model developed by OpenAI. It runs entirely on Long Play's own servers. Your video is never sent to any third-party AI service or external platform - the transcription happens locally, the same way your video is encoded.

The result is a Draft caption track. It is never shown to viewers until you have reviewed it, made any corrections, and clicked Save changes.

Starting caption generation

You can trigger it in two places:

• Right after upload - the upload confirmation page offers a "Generate captions" button while the video is still processing. Generation runs in parallel with encoding.
• Any time from the captions page - go to your video's edit page, open the Captions & Audio tab, and scroll down below the caption upload form. The "Auto-generate captions" section appears there as long as no auto-generated track already exists for that video.

How long it takes

Transcription time depends on the length of your video. A short clip may finish in minutes; a feature-length film can take a few hours. You do not need to stay on the page - you can close the tab and come back later. Your dashboard will show a notification when the draft is ready.

Language

You can optionally specify a language code (e.g. en, fr, ja). Specifying one improves accuracy. Leave it blank and the model will try to detect the language automatically.

Reviewing and editing

Once the track is ready, a notification appears on your dashboard with a direct link to the caption editor. Open it to review the transcript, fix any errors, and adjust timing if needed. The editor lets you edit cue text and timecodes, change positions, add or delete cues, and preview changes over the video in real time - see Editing caption tracks for the full guide.

When you're satisfied, click Save changes - the track becomes visible to viewers immediately.

Until you save, the draft track has no effect on playback.

One auto-generated track per video

The auto-generate option only appears if no auto-generated track already exists for a video. If you want to start fresh, delete the existing track first and the option will reappear.

Accuracy

Accuracy varies with audio quality, accent, background noise, and how clearly speech is separated from music. Tracks generated from clean, clear dialogue tend to be very accurate. Tracks from noisy audio, heavy music, or multiple overlapping speakers may need more editing. Always review before going live.

For information on uploading your own caption files, see Adding captions and subtitles.