How to make an AI song from your own idea
Aqua Voice Studio lets you describe the music you want, choose an original AI singer or an instrumental, and render the result in your browser. Here is a first-song workflow you can follow, with a prompt you can adapt.
Free public beta · no card or payment required · capacity is limited
Watch the 36-second product walkthrough · an edited tour of the real public site, with synthetic music.
Make your first song in five steps
-
Start with a clear musical idea
Think about the scene or story, the feeling at the start, and where the music should end. In the studio, add a title if you want one, then put that direction in Describe your song. You can also fill in instruments, mood, and sounds to avoid. Those extra fields help guide the result; they are creative requests, not guarantees. Use the four-part prompt workbench if you want help turning those ideas into a brief.
-
Choose the length and an available engine
Select a duration and music engine from the choices shown in your account. For a first attempt, a shorter song with one clear musical arc is easier to evaluate. Leave Start with on A new composition if you are making music from scratch. Reference editing is for instrumentals you own or have permission to adapt.
-
Choose singing or instrumental
For an AI vocal song, set Vocal source to An original AI-generated singer. Choose Write lyrics from my idea, or select Use my reviewed lyrics and provide your own words. Adapt the eight-line starter lyric or set up your own authorised singing voice. Add a short vocal-delivery note if it matters to the song. For music without singing, choose Instrumental · no vocals and No lyrics; the instrumental guide has three different starting prompts.
-
Review the cost and generate
The studio shows the exact credit cost before you submit. After you choose Generate song, your request joins Aqua’s private render queue. A dedicated local worker handles jobs one at a time, so the wait depends on availability and the work ahead of yours. If a render fails, its reserved credits are returned.
-
Listen, refine, and keep the result
When the job succeeds, play the song in My songs. Download the WAV file, or MP3 when that export is available. If the musical direction missed your idea, revise one or two parts of the brief and try again. Review lyrics and audio carefully before sharing or publishing; generated results can vary.
A useful first prompt
Give the model a beginning, a change, and an ending. This example asks for an original vocal song without naming an artist or copying a recording:
“Make a reflective pop song about finding the courage to leave a familiar place. Begin with close, gentle piano and a warm bass pulse. Let a clear, intimate singer deliver a quiet verse, then lift into a hopeful chorus with light drums and layered harmonies. End with the piano motif returning on its own. Keep the sound warm and spacious; avoid harsh distortion and abrupt transitions.”
In the form, you could set Instruments & texture to “piano, warm bass, light drums,” choose an original AI-generated singer, and let Aqua write lyrics from the idea. If you already have lyrics you own, choose the reviewed-lyrics option and paste them instead. For a short render, simplify the brief to a single verse or chorus.
What the beta can and cannot promise
After email verification, a new beta account receives 30 credits. A standard generation currently costs 10 credits; the studio always shows the cost before submission. The current public-beta duration cap is 120 seconds. Registration may pause or require an invitation when capacity is reached, and payments are disabled.
The AI can follow a direction without reproducing every requested detail. Queue times and output quality vary. Aqua requires ownership or permission for any reference audio or voice profile; deceptive impersonation and use without consent are prohibited. Read the acceptable use policy before uploading someone else’s material.
Before you make your first AI song
Can an AI singer perform lyrics I wrote?
Yes. Choose an original AI-generated singer, select Use my reviewed lyrics, and paste words you wrote or have the right to use. Listen to the render before sharing it because sung phrasing may differ from the text.
Can I make an instrumental instead?
Yes. Choose Instrumental · no vocals and No lyrics. Describe the instruments, mood, and progression you want in the song prompt.
Can I use a real singer’s voice?
Only when the speaker has given clear permission for synthetic singing and your intended use. Aqua prohibits deceptive impersonation and false claims that synthetic audio is an authentic performance. If you do not have permission, choose an original AI-generated singer.