A face that talks

One photo and a voice become a character talking to camera, in Hebrew too.

What you give

A photo, and the words.

What you get

A video of the character saying exactly your words, no training and no waiting.

Price

Priced by length, shown first

The brief you paste in the chat

Turn the attached photo into a character talking to camera. The photo: [attached, of someone who agreed, or an invented character] The words, exactly: [what the character says] The voice: [my voice from an attached file / a voice you choose] Language: [Hebrew / English] Format: [9:16 / 16:9] The face stays exactly the same face, and the words are said exactly as I wrote them. Show me the plan and one itemized price first, and start only after I approve.

What Cadabra sends to the engine

cadabra_generate_avatar: image_url the attached photo, audio_url the voice read from the exact script, so the mouth follows every recorded word and the face is never redrawn.

How it works

Use it when you need someone talking to camera: a short explainer, a word to customers, a lesson or an announcement, without filming. One photo and a voice become a video of the character saying exactly your words, in Hebrew too, with no character training and no waiting for one.

How it works: attach a photo where a person and a face are clearly visible, and write the words exactly. The voice can be yours, from an attached file, or one chosen for you. The voice is made first from the exact script, then the mouth in the photo follows every recorded word, and the face itself is never redrawn.

A tip: one clip holds up to 35 seconds of voice. A longer script is best split into several parts, each made on its own. And use a photo only of someone who agreed, or of an invented character.

Questions and answers

How long can the character talk?

Up to 35 seconds of voice in one clip. A longer recording is stopped before any charge, and needs to be trimmed or split.

How is the price set?

By the real length of the voice, measured from the file itself. The price is shown before anything runs.

Which photo works?

One where a person and a face are clearly visible. A picture with no face detected is rejected. And always a photo of someone who agreed, or an invented character.

Do I need to train a character first?

No. One photo and a voice are enough, with no training and no waiting for one. The video is made straight from the photo and the voice.

Does it work in Hebrew?

Yes. The character says exactly the words you wrote, in Hebrew or English, and the mouth follows the recorded voice.

• All presets
• How to connect Cadabra