YouTube real-time auto dubbing for live streams: what we know
YouTube says real-time auto dubbing is coming for live. Only the finished-video dubbing rules are published; here is the gap, and what Sume can and cannot do.
YouTube's Made on YouTube post says real-time auto dubbing is coming for live streams, and that is all it says in the text read: no languages, delay, eligibility or dates. The published rules belong to dubbing a finished video. Sume's English docs list no dubbing or live-translation model, so Sume cannot dub either; it can pull a finished video's audio track out for $0.01.
Sources: the Made on YouTube post and YouTube's help page Automatic dubbing, both read 2026-09-29, and the Sume docs pages cited below.
What has YouTube said about live dubbing?
Among its new viewer features, the post lists real-time auto dubbing for live. Nothing more was found on that page, so treat any language count or latency figure you see elsewhere as unconfirmed until YouTube publishes it.
How does dubbing a finished video differ?
For uploaded videos YouTube's help page gives concrete rules. A creator can turn automatic dubbing on or off in the YouTube Studio app under Settings, Content, and can tap Publish manually to review dubs first. Dubbed tracks are marked as auto-dubbed in the description.
| A video is ineligible if | Live stream |
|---|---|
| It exceeds 120 minutes | Not published |
| It has no speech, only music, or very little speech | Not published |
| It contains copyrighted content | Not published |
| The source language cannot be detected | Not published |
| The speech in the original audio is too fast | Not published |
Can Sume dub a live stream or a clip?
Not from what its docs list. The English model docs describe video, image, music, avatar and media tools, and no dubbing or translation model, and Sume works on finished files, not live feeds. If a listing appears later, the Models page is where it would show.
What can I do with a finished clip on Sume?
Audio detach takes one video in your workspace's media.sume.com storage and returns its audio as wav or mp3, at $0.01 per job. Use it to hand the original track to a translator or a separate tool. It does not translate anything itself.
curl -X POST https://api.sume.com/v1/audio-detach \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: detach-001" \
-d '{ "video_url": "https://media.sume.com/artifacts/artf_demo/talk.mp4" }'What about lip sync?
YouTube's help page describes lip sync as an automatic dubbing feature that alters the speaker's lip movement to align with the dubbed audio, and the earlier dubbing announcement calls it a pilot. Neither page says lip sync applies to live streams. Whether a speaker's mouth matches is a separate question from whether the words are translated, which is the subject of lip sync vs dubbing.
Which should I plan around?
For a back catalog, YouTube's automatic dubbing for finished videos is the documented path; check the eligibility list above first. For live, wait for YouTube to publish details. For a difference between dubbing and mouth matching, see lip sync vs dubbing.
Sources
Related posts
More in Use cases
- YouTube Shopping affiliate in 35 countries: video ideas to batch
YouTube says its shopping affiliate program reaches 35 countries by year-end. Five product video formats to batch, with the Sume length and shape limits.
- How to make a YouTube Shorts series: seasons and episodes
A YouTube Shorts series is a playlist of Shorts only, set up as a show with seasons and episodes. Rules from YouTube's page, and how to render episodes.
- Does upscaling video with AI need disclosure on YouTube?
No, YouTube lists video sharpening, upscaling and repair as not needing disclosure. The exception is an edit that also changes what a real event shows.
- YouTube video A/B test: how to make three hook cuts
YouTube says video A/B testing is coming for up to three cuts. Make three cuts of one take that share an ending and differ only in the hook, at $0.02 each.
Written by Sume