Subly's job is the accessibility deliverable: a caption track a viewer can use without the soundtrack. SDH is that job with a specific meaning — speech plus non-speech, [door slams], [music], speaker IDs — spelled out at /glossary/sdh. Auto captions are not that object. They transcribe what was said. They do not invent a sound description. Versely keeps the two passes separate on purpose.
How the speech layer is burned, and how the missing non-speech cues actually get onto the file, is below.
SDH is a spec, not a style
/glossary/sdh is the definition this page uses: a caption track that includes speech and non-speech audio so the file is usable without sound. Regular subtitles assume you can hear the mix. SDH assumes you cannot. The brackets are the whole difference.
Auto-transcribe will not invent those brackets. If the brief is SDH, someone still has to author the non-speech cues. That sentence is also in the glossary; this page does not pretend a preset closes the gap.
The speech layer is add-subtitles-automatically
add-subtitles-automatically is the accessibility-intent task on the same add_veed_captions engine: transcribe the spoken language, time each line, burn a readable BASIC look (simple, plain) rather than a hook treatment. Spot-check names, numbers and jargon — ASR is not a certified captioner on those.
The result is burned into the pixels, not a toggleable sidecar. Decide wording before the final render. A course player that wants a closed track is a different object; see /glossary/closed-captions.
Non-speech cues are authored overlays
[music], [applause], (Maya) when two voices share a frame — those lines are copy you write. add-timed-text-overlays-to-video places each one on its own start_sec / end_sec. Nothing in that tool listens to the file.
That is the SDH pass after the speech burn, not a second transcription. A mute viewer will not infer a knock that causes a cut unless the knock is on screen as text.
Burn-in first; a sidecar is a conversion
The generator writes captions into the MP4. Short-form feeds will not enable a track the viewer never asked for. When a client still wants a .srt or .vtt next to the file, sidecar conversion is a free-tool job (srt-to-vtt / vtt-to-srt), not a second caption engine.
How it works
1. Read the brief against /glossary/sdh
Speech only is subtitles. Speech plus non-speech and speaker IDs is SDH. Do not label the first as the second.
2. Burn the speech layer
add-subtitles-automatically, a plain readable preset, language matching what is actually spoken.
3. Author the missing cues
Timed overlays for [music], speaker IDs, and any sound that causes a cut or a joke.
4. Convert to a sidecar only if the destination needs one
The burn stays in the MP4. A .vtt or .srt is a conversion of the words, not a replacement for pixels on a muted feed.
Where this lives in Versely
Who this fits
- Course and training video that has to work with the sound off
- A talking-head whose off-screen sound still carries the beat
- Public-interest clips that need speaker IDs as well as dialogue
- A client who asked for SDH and must not be handed a speech-only burn labelled as such
Frequently asked questions
How does Versely compare to Subly?+
Versely burns the speech layer with add-subtitles-automatically, then authors non-speech SDH cues — [music], speaker IDs — as timed overlays. /glossary/sdh is the spec. Auto-transcribe does not invent those cues; that pass is written, not generated.
Does auto captioning produce SDH on its own?+
No. Auto captions transcribe speech. SDH adds non-speech descriptions and speaker IDs. Shipping a speech-only transcript and calling it SDH is just subtitles. The glossary names that pitfall on purpose.
Are these captions a file the viewer can switch off?+
Not the generator's output. add_veed_captions burns into the picture. A closed sidecar for YouTube long-form or a course player is a conversion step, documented at /glossary/closed-captions.
Which preset should an accessibility-first burn use?+
A BASIC look such as simple or plain, billed at the standard rate, so legibility is the point rather than motion. DYNAMIC looks are a different intent.
Other alternatives on Versely
Further reading
Try it inside Versely
The all-in-one AI studio for creators. 60+ models for video, image, voice, music and lipsync, in your browser or on your phone.
Free on the web and on iPhone, one account.
Reviewed September 4, 2026. Facts about Subly on this page are general, publicly known positioning, not pricing or feature claims — see /alternatives for how this page set is scoped. Versely capability links above are pulled from the same live data the rest of versely.studio uses, so they move when the product does.