The secret that makes Seedance 2.5 speak Armenian

Everyone tried to make Seedance speak Armenian. It did not work. People decided it was impossible.

We found the secret. Give it audio, write the dialogue in latinatar in the prompt, and Seedance 2.5 speaks Armenian. Lips and voice lock.

The rule is one sentence: one dialogue (a person speaking) = one audio file + the same speech in latinatar in the prompt. Without that, Armenian does not work.

Seedance shoots one continuous take. You pick camera, light, motion, face, and place. Phone or cinema, both work.

The full guide on making high-quality Seedance content is separate: Seedance 2.5: the practical guide. Here we keep going with the Armenian speech secret.

Do it like this.

1. Audio. Record the Armenian dialogue yourself, or make it in ElevenLabs. Both are fine.
2. One dialogue = one audio. A person speaking is one file. New speech / new shot means new audio.
3. Latinatar in the prompt. Write the same speech in Latin letters, the way Armenians type spoken Armenian on a Latin keyboard. It must match the audio.
4. Attach the audio to that shot in Seedance 2.5 and generate.

Example latinatar

Barev, yes ai-ov em sarqel ays video-n

Meaning: “Hi, I made this video with AI.” Swap your speech. Keep audio and prompt identical.

Prompt skeleton

SINGLE CONTINUOUS TAKE. ONE UNBROKEN SHOT. NO CUTS.

@[Image 1](image_1) = the speaker. Exact face, hair, clothing. Ignore original background.
@[Image 2](image_2) = the location. Keep this place exact.

Camera: [phone handheld OR cinema camera - you choose as director].
Length: [8–12] seconds. Photorealistic. Face always readable.

The person speaks this dialogue (latinatar, must match the attached audio exactly):
Barev, yes ai-ov em sarqel ays video-n

Hard rules: lips sync to the audio; no on-screen fake text; product/logo stays sharp if present; identity locked to @[Image 1](image_1).

This guide gives you the secret: an Armenian-speaking shot. If you want to learn how to make high-quality Seedance video. directing, taste, prompts, full pieces. that is Vibe Directing.

The next group starts 14 September. Spots are limited. Registration and details are on the course page.

We built a special Claude skill. It writes professional, director-level Seedance prompts for you.

Do it like this.

1. Download the skill ZIP (the download link will sit with this guide).
2. Open Claude.ai or Claude Desktop.
3. In the left sidebar go to Customize, then Skills.
4. Click +, then Create skill, then Upload a skill.
5. Upload the ZIP and turn the skill on.
6. Then just tell Claude: write a Seedance prompt for [your video]. The skill kicks in on its own.

Details: .

1. Pick one short dialogue.
2. Record it or make it in ElevenLabs. One audio.
3. Write the same speech in latinatar.
4. Write the Seedance prompt: references + camera choice + hard rules.
5. Put the latinatar in the prompt, attach the audio, generate.
6. Check the lips. If wrong, fix the match.