Skip to content
Launch pricing: 1 credit per second until Oct 7, then 1.2 credits per second.See pricing
Guides

How to write lyrics and direction for an AI duet video

Add your own lyrics, performance directions, or both in the optional text box. You have up to 500 characters to shape the singing style, camera mood, and movement. Leaving the box blank gives you our standard performance.[1]

How to write your lyrics and direction

Four helpful details to include are who starts first, what each person does, how they deliver the lines, and the overall mood on camera.[2]

01

Who goes first

Name who sings or speaks the opening line. Setting the starting order helps create a natural back-and-forth flow across the scene.

Example textPerson on the left starts singing the opening line, then person on the right joins for the chorus.
02

What each person does

Describe actions across the scene, such as turning toward each other, pointing to the camera, or stepping through a small dance.

Example textTurn toward each other during the intro, point at the camera on the beat, then share a small dance.
03

How to sing it

Choose a vocal style for the lines, such as rap, slow melody, or a soft whisper to guide the sound.

Example textDeliver the first verse as a fast rap, followed by a slow whispered harmony on the final line.
04

Mood on camera

Set the emotional tone for the scene, from upbeat and energetic to calm, playful, or thoughtful throughout the performance.

Example textKeep the mood playful and energetic, with warm smiles and lively eye contact throughout the take.

What to leave out

  • Skip lyrics you lack rights to, and choose your own words to keep your video yours.
  • Skip names of real public figures so the focus stays on your two subjects.
  • Skip harmful or unsafe words to help the AI model process your scene smoothly.
  • Save lengthy paragraphs for longer scenes, matching brief lines to shorter clips.

Video length and credit costs

Select 5, 10, or 15 seconds for your video. Credits are charged per second of video, and shorter text naturally suits a shorter video. The full credit cost displays clearly before you generate your scene.

Quick feature facts

FactValueSource
Text lengthAdd up to 500 characters of lyrics and direction.Create
Text requirementOptional, with standard performance used when left blank.Create
Suggested topicsTurn order, dance moves, vocal style, and camera mood.This guide
Where words goWebsite through routing partner kie.ai to AI rendering model, skipping analytics tools.Privacy
Unused prompt creditsAll charged credits return to your balance automatically.Refund policy

Frequently asked questions

Do I have to write lyrics and direction?

Writing in the box is completely optional. If you leave it empty, your scene uses our standard performance with default movements and vocal delivery. Adding your own words is simply there whenever you wish to customize who sings first, how the characters move, or the overall mood.

Can I use lyrics from my favourite song?

We recommend writing your own words or choosing lyrics you have permission to use. Creating original lines keeps the video uniquely yours and respects songwriter rights. You can write rhymes, spoken dialogue, or a simple greeting that fits your two characters.

How closely will the video follow my words?

Your text guides the overall performance, and each video is a fresh creative take. After your video is ready, take a moment to check that the visuals and lyrics match what you had in mind. You can always adjust your wording and create another take to explore different results.

How long can my lyrics and direction text be?

You can write up to 500 characters in the text box. That provides plenty of space to outline who sings first, detail specific actions like dancing or pointing, specify a singing style, and set the camera mood for your scene.

What happens to my credits if the AI model cannot use my text?

If the AI model cannot use your added text, all credits charged for that video return to your balance automatically. Nothing else is charged, allowing you to revise your description and try again whenever you are ready.

Sources

  1. Seedance 2.0seed.bytedance.com

    SupportsSeedance models take image references together with a text prompt.

  2. Bytedance Seedance 2.0 Minidocs.kie.ai

    SupportsOfficial documentation explains that a text prompt guides the motion, delivery, and creative styling of the generated video.

Ready to bring your duet to life

Create your video