Lesson 6 of 7
Sound comes too
7 min read
For a long time AI video came out silent — you'd add sound afterwards. Now the newest tools generate the sound along with the picture: footsteps that land on the step, a voice that fits the face. Where does a clip's audio come from?
Video used to be mute
Until recently, an AI video was a silent movie. The tool made the pictures; any sound — voices, footsteps, music — you added yourself in an editor afterwards. The newest models changed that: they generate matching audio at the same time as the video, so a wave crashing actually sounds like a wave, and a person's voice lines up with their face. Sound that's generated together with the picture is often called native audio.
Older tools made silent clips and you added sound later. Newer ones generate native audio — the sound is created together with the picture, already in sync.
A soundtrack is layers
Whether it's generated or added, the sound of a video is really several layers stacked together: the dialogue (people talking), the sound effects — footsteps, a door, a splash — and the ambient bed (room tone, wind, distant traffic) plus any music. Thinking in layers is powerful, because you can turn each one up, down, or off on its own.
A video's sound is a stack of layers — dialogue, sound effects, ambient/music — and each can be adjusted on its own.
Speaking every language
One layer is special: the dialogue. Swap it for a new recording in another language and re-sync the mouth — the lip-sync from the last lesson — and the very same video now speaks that language. That's dubbing, and AI makes it fast: one clip, many languages, mouths matching each.
When you dub, only the dialogue layer changes — the sound effects and ambient stay put. That's why thinking in layers matters: you translate the talking without disturbing the footsteps, the music, or the room.
The gist
- —Older AI clips were silent; newer tools generate native audio — sound made together with the picture, in sync
- —A soundtrack is layers — dialogue, sound effects, ambient/music — each adjustable on its own
- —Dubbing swaps just the dialogue layer for another language and re-syncs the mouth — one clip, many languages
You have a finished AI clip in English and want a Ukrainian version — same footage, same footsteps and music, but the person speaking Ukrainian with matching lips. What are you doing?
Continue in the app
Take the whole AI Video course — tracked
Get your personalized path, progress and streaks in the app — this lesson and every next one, in order.