Dubbing is not laying a voice over a video. It is replacing the original dialogue while keeping the music, the sound effects and the atmosphere. Here is how it actually works, and what separates a dub that lands from one that sounds fake.
Everyone's first attempt is the same: mute the video, record a voice on top. The result always disappoints, for a precise reason. Muting does not only remove the dialogue. It removes the score, the footsteps, the slamming doors, the wind, the explosions. The scene becomes a sonic desert with one voice floating in it.
Real dubbing keeps everything except the voices. In the industry that track is called the M&E, for music and effects. Studios receive it separately from the producer. For a video you found online, it does not exist. You have to rebuild it.
Thirty seconds to two minutes. Longer, and one mistake means redoing everything. Prefer clear dialogue with the music sitting back.
A separation model splits the mix in two: voices on one side, music and effects on the other. You keep the second and discard the first.
Knowing the words is not enough. You need to know the exact second each line starts and ends, otherwise you drift.
Headphones on, eyes on the actor's mouth, not on the text. Play the intention, do not read the line.
Your voice sits on the music and effects track. Two settings usually suffice: voice level, and a global offset if everything runs slightly early or late.
A limit worth knowing. Vocal separation is never perfect. On a heavily compressed clip where voices and music share the same frequencies, small artefacts remain in the high end. A dialogue scene with restrained music will always come out cleaner than a loud, saturated one.
The microphone question always comes first, and it is the wrong one. The room decides. A two hundred dollar microphone in an empty tiled living room sounds worse than a thirty dollar headset in a bedroom with a bed, curtains and a wardrobe. Soft surfaces absorb reverb, and reverb is what gives away an amateur recording. It cannot be removed afterwards.
Dubba handles steps 2, 3 and 5 automatically. You paste a link, it isolates the music and effects track, transcribes the dialogue with its timing, and shows a rythmo band scrolling under the video to tell you when to speak. You only have to perform.
Dub a video nowYes. An online studio chains separation, transcription and mixing for you.
The original stays under copyright. Amateur, non monetised dubbing is widely tolerated and often falls under parody exceptions, but this varies by country. For commercial use, get permission.
Start with what you have, in a furnished room. Upgrade only when the gear genuinely limits you.
Almost always because the video played through speakers. Use headphones. If a constant offset remains, it is your audio latency, and a global offset fixes it.
Fifteen to thirty minutes for a one minute scene, mostly spent on retakes.