YouTube started dubbing videos for every channel on 4 February 2026. The announcement said 27 languages.
In December 2025 more than 6 million people a day watched at least ten minutes of auto-dubbed video. That is YouTube's own figure, and it is why I looked at this seriously.
Then I opened the language table and counted the rows.
What the table says, read on 24 August 2026
The support page lists 30 original languages. That is what a video is allowed to be spoken in.
The second column is the part nobody reads. It says what each original language can be turned into.
For English, that column holds 20 entries:
- Arabic, Bengali, Dutch, French, German
- Hebrew, Hindi, Indonesian, Italian, Japanese
- Korean, Malayalam, Polish, Portuguese, Punjabi
- Russian, Spanish, Tamil, Telugu, Ukrainian
Nine of the 30 original languages never appear in that column. YouTube will listen to them. It will not speak them.
Those nine are Chinese, Chinese Traditional, Farsi, Romanian, Swahili, Thai, Turkish, Urdu and Vietnamese.
The feature points at English
I counted every row, not only my own. There are 29 original languages besides English.
Twenty-five of them dub into English and nothing else. One target, no second option.
Four have more. Hindi also reaches Bengali. Korean reaches Indonesian, Portuguese reaches Spanish, and Spanish reaches Arabic and Portuguese.
So the feature mostly moves the world's video into English. Going the other way, it stops at 20 languages.
What that does to the people I teach
The house style for this blog names my readers: Brazil, Iran, Poland, Indonesia, Turkey, India.
Four of those six are reachable from English. Portuguese, Polish, Indonesian and Hindi are all on the list.
India is served best of all. Six Indian languages sit in that column: Bengali, Hindi, Malayalam, Punjabi, Tamil and Telugu.
Iran and Turkey get nothing. A student in Tehran can have their own video dubbed into English. They cannot have mine dubbed into Farsi.
Expressive speech is eight languages
There is a second, shorter list, and it matters more than the first.
Expressive speech keeps the pitch and pacing of the original voice. Without it the dub is flat.
The table marks it with an asterisk. The asterisk sits on the target language, not on the source, which is easy to misread.
Count them and you get eight: English, French, German, Hindi, Indonesian, Italian, Portuguese and Spanish.
That is the same eight YouTube announced in February. Six months later the number has not moved.
So of the 20 languages I can reach from English, seven get the expressive voice. The other 13 get the flat one.
Two limits worth knowing before you plan
A video over 120 minutes is not eligible. Neither is one with very little speech, or one where the speaking is too fast to slow down.
Auto-dubbing is on by default for eligible channels, and dubs publish on their own.
In YouTube Studio, open Settings, then Channel, then Advanced settings. Tick the box for manual review before publishing.
What I have not tested
I have not run one of my lessons through it and scored the words.
That is the question which decides whether any of this is usable in teaching. Bake, cage, texel density, UV seam, normal map — five words carry a whole lesson.
I am not going to guess at an error rate. When I have counted it on a real clip, I will publish the count and the clip.
The arithmetic on my own catalogue
I have nine courses. Together they run 159 hours, all recorded in English.
The Substance Painter Master Course is 23 hours of that. The Unwrap Master Class is 50 hours on its own.
My own dubbing studio charges $3.50 per hour of dubbed video, per output language.
One course into one language is about $81. The whole catalogue into one language is about $557.
Running the transcription on your own GPU takes 30 per cent off that rate. The catalogue then costs about $390 per language.
What I am doing with mine
Free beats $557 whenever free reaches the person. For Portuguese, Indonesian, Polish and Hindi, it does.
For Farsi and Turkish there is no free option at all. Those two get paid dubbing, or they get nothing.
That is the whole decision. It is not a judgement about quality. It is a question of whether the row exists in the table.
Read the table yourself before you plan around it. It carries no date, and YouTube says it is still expanding.
