Script then dub a faceless explainer | AI Video Factory

You want a publishable explainer and you do not want to appear. Write a timed talking-head script first, then synthesize on the same clock. The script tool does not need a video. Dubbing needs a picture about as long as the script — empty heads, a screen recording, or stock. Do not shoot yourself and paste copy afterwards.

Beats first, picture second

A whole essay with no in/out points is read end to end, and length is a guess. Do not reverse the order:

1. In the script tool, lock topic, style and duration (often 15 / 30 / 45 / 60 / 90 seconds).
2. After generation, fix the hook, split long beats, read them aloud.
3. Click Continue with dubbing so the current beats feed dubbing. Do not open a second project and type them in.
4. Drop in a local MP4 / MOV that is not shorter than the script.

Script only, then you appear on camera: script before the shoot. The file already has a voice: ten-minute captions — do not write a script and paste it back.

A faceless path

1. Open script in the studio. This step does not upload a video.
2. Make the topic specific. Style: explainer, compare, or tutorial. Generation: Auto. Edit the words, then Continue with dubbing.
3. Drop in a local picture. The file stays in the browser. Chinese uses 晓晴; do not lock an English voice on a Chinese script.
4. If the picture has no voice you want to keep, turn original audio off so a stock narrator does not stack. Leave it on only for useful room tone.
5. Speed follows each beat by default. If one line sounds thin and rushed, change that beat’s words or in/out.
6. Listen through, then export. To burn captions, use the same cues; mix first — SRT or burn-in.

The free plan includes one script run and one dubbing run per month. Styles, voices and limits are on Features and Pricing.

Check only three things

• First three seconds: the opening beat needs a hook. Kill “Hi everyone, today we talk about”.
• Picture long enough: if the file ends before the last beat, the tail is never spoken. Prefer a slightly long empty head over a short one.
• One breath per beat: two sentences in one beat make the voice inhale in the wrong place.

What usually breaks

Downloading TXT and typing it into dubbing drops the clock. Use Continue with dubbing. A talking-head stock file with original audio on stacks two voices. Export a cover in the platform ratio; do not use the first empty frame — covers from the clip.