YouTube Live auto dubbing starts in 2027 — here is what you can hear tonight
On this page

YouTube announced Live auto dubbing on September 23, 2026. A pilot starts in early 2027. It translates a streamer's speech in real time, so viewers can listen in another language.
That sentence is being repeated everywhere, and it is easy to file under "YouTube already translates things." It does, in two older ways that do not cover the live stream you have open. The pilot is a third thing, and it is not on that stream tonight either.
More than 40% of live watch time already comes from outside a creator's home country. The gap is real. The announcement does not close it for a viewer this week.
Three features, one search
People search one question — "does YouTube translate live streams?" — and land on three different products.
| What you are looking at | YouTube | DubTab |
|---|---|---|
| Automatic captions on a live stream | English only, on some streams, and never translated. They need a normal-latency stream with captions turned on. | Listens to the audio itself. A dub and captions in your language, a few seconds behind the speaker. |
| Auto-dubbing on a video | Generated when an eligible video is uploaded. The creator can turn it off. A live stream is not an upload, so this track is not there. | On a normal video, dubs the one you are watching, including uploads that never got an official track. |
| Live auto dubbing | A pilot from early 2027. Speech, translated as it is spoken, for viewers on streams in the pilot. A viewer cannot switch it on for an arbitrary stream. | The same job, from your side of the screen, on the stream that is playing. The streamer sets up nothing. 60+ languages. |
Sources for the YouTube column: automatic live captions and auto-dubbing, plus the Live announcement. The language pair for the pilot is not in that post. A recap of YouTube's Creator Roundup describes an English and Spanish pilot with a small group of creators first. Treat that as a description of what was shown, not as a promise written into the blog post.
"YouTube already dubs videos"
It does, for some of them. Auto-dubbing runs when an eligible video is uploaded. Long videos, music-heavy audio, fast speech, and plenty of language pairs fall outside it. If the creator turns it off, the track is gone. If the video you opened never got one, you cannot request it.
When a video does have an official dub in your language, use it. It was made for that upload, and it is already attached.
A live stream has none of that. There is no file yet, so nothing was generated at upload. The caption track, when it exists, is English text on some streams. It is not a translation, and it is not a voice.
DubTab is the viewer-side version of the same wish. On a normal video it dubs the one playing, which is what the YouTube dubbing page is about. On a live stream it has to work with no track at all.
What early 2027 actually changes
The pilot translates speech as the streamer talks, and viewers listen in a language they pick. That is a real product, aimed at the right problem: a stream with an audience that does not share the streamer's language.
A few limits are already in the announcement, and a few are simply not promised.
It starts in early 2027, as a pilot. It is speech. A music stream or a stream with no words has nothing for it to translate. It shows up on streams in the pilot. Opening a random live URL and expecting a language menu is not what was announced. The viewer does not enable it. The streamer has to be in the pilot.
If the Creator Roundup description holds, the first pair is English and Spanish, with a small set of creators. A Japanese stream, a Korean stream, a Chinese stream, a Portuguese stream — none of that is in what has been shown. Even an English stream is out until that channel is in the pilot.
Until a stream you care about is actually in it, the official column for dubbed audio on live is still empty.
What you can hear tonight
DubTab listens to the stream's audio. Captions are not required, which matters because most live streams do not have usable ones. Pick a language once. The voice follows the speaker, a few seconds behind — it has to hear a phrase before it can say it. On a long stream the useful part is that the gap stays there: auto catch-up keeps the voice on the live edge instead of drifting minutes behind.
The streamer does nothing. No caption setting, no dub track, no knowledge that you are watching in another language. A stream spoken in Japanese, Spanish, Korean, or any of 60+ languages is the same case. YouTube's live captions still only exist for English. DubTab is not reading those captions.

That is the page for the workflow itself: translate a YouTube live stream. A replay after the stream ends is just a video, and the same account handles it.
The delay is the honest limit. This is not a zero-lag interpreter sitting in the room. If you need the joke at the exact frame it lands, you will feel the few seconds. If you need to follow a launch, a match, a lecture, or a creator who does not speak your language, those seconds are the whole method: hear a line, say it in your language, stay with the live edge.
What the pilot still will not be
Even after early 2027, the pilot is a creator-side track on some streams. Everything outside that set is the same problem as today.
A language the pilot does not speak. If the first wave is English and Spanish, every other source language is still on you.
A creator who is not in the pilot. Forty percent of live watch time from abroad does not mean forty percent of streams get a dub. A small group of creators is a small group of streams.
A stream that is not on YouTube. Twitch, Kick, and TikTok LIVE were not in this announcement. A live there still has no file and no official dub. DubTab hears those the same way when the audio is playing in the browser — the Twitch, Kick, and TikTok LIVE pages are the same idea on a different site.
Speech only, on both sides. YouTube's pilot translates speech. So does DubTab. A music bed with no words is not a translation problem either tool can solve by talking over it.