The questions people ask us

A dubbing studio that runs in your browser. You pick a scene from a film, a series or an anime, the text scrolls under the picture on a rythmo band, and you perform the lines with your own voice. The result can be published and shared.

There is nothing to install. It works on a computer and on a phone.

No. It is your voice, recorded through your microphone. Nothing is synthesised, no voice cloning.

AI is used elsewhere, on the preparation: it separates the original voices from the music and effects, transcribes the dialogue and works out who is speaking. The acting stays entirely yours.

Dubbing made by fans, outside any professional pipeline. You take a scene you love and perform it with your own voice, often in a different language from the original, sometimes twisting it into something else.

It is not the same as fansub, which means subtitling. Fansub is written, fandub is performed.

Voice over sits on top of the picture without replacing anyone: a narrator, a commentary, an advert. Nothing forces it to match anyone's lips.

Dubbing replaces the voice of a character speaking on screen. You have to land on their pace and their lips. That is what Dubba does, and that is why there is a rythmo band.

The ribbon of text that scrolls under the picture, with a fixed reading line. When a word crosses that line, that is the moment to say it.

Professional dubbing studios have used it for decades. It removes the need to learn lines by heart or count seconds: you just read at the right pace.

Paste a YouTube link or upload a video from your device. Dubba separates the voices from the music, transcribes the dialogue and identifies the characters. This usually takes one to two minutes.

Then you pick the roles you want to play, hit the red button, and read the rythmo band. You can listen back, record as many takes as you want, then publish.

Five minutes. But the best dubs run between thirty seconds and two minutes.

It is not only about patience: on a short passage, voice separation and text splitting are far more accurate. A long scene with many characters always gives a rougher result.

On the settings screen that appears before you start, a "Microphone" block lets you pick your audio input. If you have not granted access yet, click "Choose my mic": browsers only reveal device names once permission is given.

You can change it any time during the session, in the Voice panel at the bottom left of the studio.

Yes, and a good share of published dubs come from a phone. Use earphones: without them the mic records the film's sound again and the mix turns muddy.

Landscape makes the rythmo band easier to read. Mute your notifications too, an incoming call cuts the take.

In the studio, click "Fix the text". The rythmo band turns into a timeline: tap a line to edit its text, or drag it to shift its timing if it lands too early or too late.

Text edits save as you go. A drag to re-time needs the "Save" button. The top bar always tells you whether everything is saved.

Right-click it, or long-press on a phone, then pick the right character from the menu. You can also select several lines at once and reassign them together.

This helps when two voices sound alike: automatic detection sometimes merges them, or swaps them on a single line.

Because it has been certified. Once someone has checked the number of characters, fixed the text and named the roles, the scene is locked for everyone.

The transcript is shared by everyone dubbing the same video. Without that lock, one careless edit would break everyone else's work, and that has already happened. If you spot a genuine mistake on a certified scene, report it and we will fix it.

Yes, in the Voice panel of the studio. Put the real names in, it makes the rythmo band far easier to read when several characters talk to each other.

A bonus: a named voice is recognised automatically in your next videos.

Give the exact number of characters and re-run the analysis: the engine separates far better when it knows how many voices to look for.

Do not claim more voices than there really are. The engine obeys and splits one character in two to reach the count, which is worse than where you started.

On the settings screen, choose "Open mic" instead of "Dub the text".

By default Dubba only keeps what you say during the lines and cleans the rest, breaths and noises included. With an open mic everything is kept from start to finish, including whatever you say outside the script. That is the right choice for a commentary, a reaction or improvisation, but nothing gets cleaned up.

On the "Who are you dubbing?" screen, hit "Dub everyone". Every role is assigned to you and you perform them in the same take.

You can also take only some roles and leave the others on their original voice: their lines will keep the film's voice in the final mix.

It is almost always the wrong microphone selected, or a mic muted at system level. Check the input in the Voice panel, and watch the level meter while you speak: the bar should move.

On a computer, another app holding the microphone can also stop it from opening. Close it and record again.

The mic is picking up too loud and the sound breaks as it is recorded. That cannot be fixed in the mix, the take has to be redone.

Move ten to fifteen centimetres away from the mic, or lower its input level in your device settings. While recording, keep an eye on the level meter: as long as the bar stays out of the red zone, you are fine.

Use the "Align my voice" control in the Voice panel. It shifts your take by fractions of a second, forward or back, and the preview updates immediately.

This offset often comes from Bluetooth headphones, which add latency. Wired, it is almost always zero.

Create a collaborative session: you pick the video, Dubba detects the characters, and everyone claims the role they want through a simple link.

Nobody needs to be available at the same time. Each person records on their own, whenever they can, and the creator assembles everything once all takes are in.

No. When finalising, you can leave empty roles on their original voice: their lines keep the film's voice, and the dub still comes out.

Two possible causes. Either the creator finalised the session while you were recording, or someone took your role in the meantime.

Either way your take is not lost: a button lets you download it to your device, to send it to the creator or reuse it elsewhere.

To your profile, with a link you can share. You choose its visibility: public, link-only, or private.

A public dub can appear in the discovery pages of the site.

Yes. The vertical export is built for those formats: the video is reframed and your dub is baked into the audio track.

One tip if you post on Instagram: do not add music from their library. It replaces the original audio track, so your dub disappears.

Dubba never hosts the original video. For a YouTube video it stays on YouTube and is only displayed. What we store is your voice and the reconstructed music and effects track.

Dubbing falls under transformative use, but platforms apply their own rules. A short scene usually passes where a whole film gets flagged.

During the beta everything is open and nothing is charged. Paid plans will exist later to raise the monthly dub count and unlock high quality export.

MP4, MOV, WebM, AVI and MKV, up to 500 MB. You can also paste a YouTube link directly, with nothing to download.

The report button at the bottom left of the site. The message reaches us with the technical context, which lets us reproduce the problem without asking you ten questions.

The project is in beta and most fixes come from those reports.