AI Singing Generator
Turn Your Lyrics Into Songs with AQUA Voice Studio
Create original AI-generated vocal songs. Write your lyrics, explore singing styles and hear your musical ideas come to life.
A$0 beta · 30 credits after email verification · 10 credits per standard song · adults 18+
Choose an original AI singer or your authorised voice profile. Songs start private; rendering takes minutes and uses a capacity-limited queue.
Rendering takes minutes. Your queue position appears after submission.
The River Keeps the Hours
A tender, reflective vocal song with piano, guitar and cello.
A real 26-second excerpt, already published by Aqua Studio. AI-generated music with Jed’s authorised synthetic vocal. It illustrates the saved-voice workflow; it is not a live human performance or a guarantee of another singer’s result.A whole song, not just spoken text
A singing generator combines a melody, synthetic vocals and a musical arrangement. Text-to-speech reads words aloud; voice conversion changes vocal character. In Aqua, a saved voice profile can be used during song generation, but uploading a reference does not promise a copy of its tune, timing or performance.
For a first try, choose an original generated singer and bring a few reviewed lines. You can set up your own authorised profile separately. Neither workflow requires pretending to be a named artist.
Two voice choices, different expectations
An original AI-generated singer
Choose An original AI-generated singer in Vocal source. Describe the delivery: intimate or strong, restrained or expressive, and how the verse should differ from the chorus. The result is a newly generated synthetic voice, not a named artist, a guaranteed vocal identity or a performance by a real singer.
Your authorised voice profile
If you have a saved profile, choose My authorised voice profile and select it. Voice conversion aims to carry the authorised vocal character into the generated performance. It can introduce pronunciation or timbre artefacts; similarity and expressive delivery vary. A recording and consent declaration are required before a profile can be used.
Use only your own voice or one for which the speaker has clearly permitted synthetic singing and your intended use. Profile creation is not permission to impersonate a singer or claim a synthetic track is authentic. Read the voice-consent requirements.
Make a song with your own voice
Start by creating a profile. After verifying your account email, open Voices and use Add an authorised voice.
- Prepare one clean 6–30 second reference of your own voice: one person, no backing music, overlapping voices or strong room noise. A singing reference near the intended range can help. Use WAV, FLAC, MP3, OGG, M4A or AAC, up to 32 MB.
- Enter a profile name and speaker name, choose your permission basis, and add any limits in the permission note. Confirm the synthetic-singing permission and Acceptable Use declarations, then select Save voice profile. This records your declaration; it does not verify identity or create permission.
- Once the profile shows Ready, return to Create, choose My authorised voice profile under Vocal source, and select it. Add your musical direction and reviewed lyrics, check the credit cost, then render and listen before sharing.
Bring words that tell one story
Decide who is speaking and what changed
Before writing, name the speaker, addressee and event. Choose a small number of concrete details. A verse should establish a situation; the next verse should add information rather than repeat sadness in different metaphors.
Choose one idea for the chorus
Use a short anchor line that means something in this particular story. Keep the words natural when repeated. Meaning matters more than forcing a rhyme.
Use reviewed lyrics in Aqua
Set Lyrics to Use my reviewed lyrics and paste words you wrote or have permission to use. Section labels such as [Verse] and [Chorus] help organize the request. For a short first render, use a compact verse and chorus rather than squeezing a long lyric sheet into a small duration.
Choose an arrangement that leaves space
Put the musical arc in Describe your song, then set instruments and vocal delivery. Review the displayed credit cost before Generate song. A full arrangement is rendered locally; a custom lyric sheet is a requested performance, so the final singing still needs checking.
A vocal prompt you can adapt
This is an original example of arrangement direction, not the exact prompt used for the demo. Pair it with your own concise lyrics about one specific event.
“An original reflective vocal song about finally packing away a former partner’s coat. Begin with close piano and a warm, intimate lead voice. Keep the verse sparse, then let cello, bass and restrained drums lift one memorable chorus. Use the supplied reviewed lyrics. Keep words clear and the melody singable; no dense choir, rap, exaggerated vibrato or sudden tempo changes.”
Instruments & texture: “piano, cello, warm bass, restrained drums.” Vocal delivery: “natural phrasing, gentle breath, clear words, restrained verses.” These details guide a probabilistic model; they do not guarantee crisp words or exact timing.
Try a complete starter lyric
This original fictional example is free to adapt for an Aqua experiment. It is a separate teaching example, not the lyric or render prompt behind the demo above. The speaker chooses to keep a friend company; each detail supports that decision.
Pair it with: “A quiet acoustic song about keeping a friend company. Fingerpicked guitar, warm piano and a restrained original singer. A sparse verse, a gently lifted chorus and a soft ending. Use the supplied lyrics; no extra verses or dense backing choir.”
The studio opens with this starter ready to review after signup and email verification. You choose whether to generate; this button does not submit a song or use credits.
Choose a short length that leaves room for these lines, review the displayed cost, and listen for every important word after rendering. The model can omit or alter words, so this is a starting request rather than a guaranteed performance. Follow the complete first-song steps.
Check the sung words before you share
- Compare the recording with the lyric sheet. Listen for omissions, changed words and a chorus that loses its meaning.
- Check the opening, loudest section and ending for audible artefacts or abrupt transitions.
- If phrases feel rushed, simplify the arrangement or give the lyric more time. Try another render rather than assuming the first result is finished.
- Keep clear synthetic-audio context when publishing, and confirm rights in any uploaded lyrics or voice.
Quality varies across lyrics, arrangements and voice profiles. Aqua cannot promise professional vocal intelligibility in every generation.
A limited free beta, a real queue
Standard new songs can be up to 120 seconds. Email verification activates 30 opening credits; a standard song costs 10. No card is required and billing is disabled. Daily grants and available capacity are limited.
Aqua’s local worker renders one song at a time. Vocal conversion can add processing, so expect minutes and a longer wait when the queue is busy. Failed jobs release reserved credits. Your songs remain private until you choose to create a public link.
AI singing questions
Can I turn my lyrics into a song using my own voice?
Yes, with an eligible authorised saved voice profile. Set up your voice, select it for your vocal song and bring lyrics you own or have permission to use. Review the credit cost before generation, then check the rendered words and voice quality before sharing. Pronunciation, word delivery and vocal similarity can vary. Explore voice choices in Studio Lab.
Is the AI singing generator free?
Aqua is a limited free beta. After email verification, the opening allowance is 30 credits and a standard generation costs 10. No card is required. Daily grants and the render queue are limited; this is not unlimited free generation.
Can Aqua write the lyrics for me?
The studio also offers Write lyrics from my idea with the local lyric writers available on the worker. Review generated words before publishing. You can avoid that writing step by supplying your own reviewed lyrics.
Does local mean it runs on my laptop?
No. You use the web studio; the models run on Aqua’s private workstation. Uploaded material is processed there, and the account library protects private results.
Can I make music without a voice?
Yes. Choose Instrumental · no vocals and No lyrics. See instrumental prompts and a real result.