Video subtitles
Recognise English speech, review subtitles over your video, and save a separate subtitle file.
Video stays on your device. Recognition runs in this browser using a local speech model, not Hosaka’s chat or image engines. No video edits or burned-in captions.
Video and subtitle preview
Playback: 0 seconds
Generate subtitles
English speech recognition · clips up to 30 minutes. First use loads about 60 MB of model/runtime files from Hosaka. Processing uses this device’s CPU and may take several minutes.
Optional separate audio track
If your browser cannot decode the video’s audio, select its matching WAV/MP3 audio. It must start at the same time as the video.
Automatic text and timestamps can be wrong, particularly with music, quiet speech or accents. Review them in the editor below.
Subtitle editor
YouTube export saves a plain UTF-8 SRT with numbered captions and millisecond timestamps. Styling and positioning are removed; wording and timing are kept. In YouTube Studio, open Subtitles, select your video and caption language, then Upload file → With timing. YouTube format guide. This downloads captions; it does not upload your video.
Edits update the separate subtitle track immediately. Export before leaving; the original video is never rewritten.
Speech recognition: Whisper tiny.en via Transformers.js. Runtime and model licences are included in the reusable template.