Subtitle Studio

About the Subtitle Studio

Subtitle Studio drops a video in, auto-transcribes the speech, lets you fine-tune the captions on a timeline, and exports them as SRT, VTT, or burned right into a new MP4. Style fonts, colors, outlines, and positioning live — what you see in the preview is what gets baked into the final file.

Everything happens inside your browser, including the AI transcription. Your video never leaves your device. The first run downloads a generic, open-source speech model (~80 MB) from Hugging Face over HTTPS, caches it permanently, and runs fully offline from then on — even the model file is the same one used for every voice, not anything tied to your audio.

What you can do

• Drop & auto-transcribe: MP4, WebM, MOV — drop a video and the built-in speech model generates timed cues automatically.
• Cue list editor: Edit, retime, or delete any individual cue. Add a new cue at the current playhead with one click.
• Shift all cues: Bump every caption forward or backward by 0.5s — useful when transcription is off by a consistent offset.
• Live preview overlay: Captions render on top of the video as you scrub, with the exact styling you've chosen.
• Caption styling: Font (Inter, Montserrat, Helvetica, Georgia, Courier), size as % of video height, text color, outline color & width, background fill (none, black at 50%/75%/solid), and position (bottom, middle, top).
• Export SRT or VTT: Save the captions as a sidecar file for YouTube, Vimeo, Premiere, Final Cut, or any video player that loads external subtitles.
• Burn-in to MP4: Re-encode the video with the captions baked into the pixels — perfect for social platforms that don't render external caption files.
• Fullscreen preview: Toggle fullscreen on the preview to check legibility on a bigger screen before exporting.

Common uses

• Adding captions to Instagram Reels, TikToks, YouTube Shorts, and other social videos where most viewers watch on mute.
• Making podcast episodes, interviews, and tutorials accessible.
• Generating SRT files for delivery to clients or distribution platforms.
• Localizing video content by editing the auto-transcribed text or replacing it with a translation.
• Adding stylized "title-card" captions to vlogs and product videos.
• Quickly transcribing a long meeting, lecture, or talk recording into editable text.

Tips for the best results

• Use the cleanest audio you can. The model handles background noise, but speech that's clearly above the noise floor transcribes far more accurately.
• For social videos shot vertically, keep caption size around 4–6% and use a black-50% or black-75% background so the text stays legible over busy footage.
• White text with a 2–4px black outline is the safest default — readable on any background.
• Use the "+0.5s all" / "−0.5s all" buttons if the captions feel uniformly late or early after transcription.
• Burn-in is the right choice for Instagram, TikTok, and any platform where the player won't load an external SRT. Use SRT/VTT export when uploading to YouTube or delivering to an editor.
• Keep the tab active during burn-in. Browsers throttle background tabs, which slows the encoder significantly.
• The model download only happens once. Future visits are fully offline — re-transcribing a second video skips straight to processing.

No accounts, no upload queue, no transcription credits — drop a video, edit the captions, export or burn-in. The output is yours to share, post, or hand off to an editor.

Related tools