Make a song from a prompt

Describe the music, or bring your own lyrics, and get a full track with vocals or an instrumental. Four AI music models in one place, from 5 credits. No watermark, and you own what you make.

Make a song

Every music model, and what it costs

Two of these sing lyrics you write; two work from a description. Prices start at the figure shown and can rise with options such as length or the number of songs. The playground quotes the exact cost before anything runs, and runs that fail are refunded.

ModelPriceAboutYour lyricsGood for
Mureka V9from 5 creditsfrom $0.08 to $0.13YesFull songs from your own lyrics, with a chosen voice
Lyria 3 Profrom 8 creditsfrom $0.13 to $0.20NoGoogle, music tracks from a description
ElevenLabs Musicfrom 9 creditsfrom $0.15 to $0.23NoSongs or instrumentals, with the length you set
Music 2.6from 15 creditsfrom $0.25 to $0.38YesComplete songs with vocals and instruments

Credits work out at about 2 cents each depending on the plan. Plan prices are on the pricing page. Checked 27 September 2026.

How it works

  1. Step 1

    Describe the music

    Genre, mood, tempo and instruments. Specific words get you closer than a single adjective.

  2. Step 2

    Lyrics, or none

    Paste lyrics you wrote for a sung song, or leave them out for an instrumental.

  3. Step 3

    Try a few takes

    Music is the most variable output there is. Generate a few on a cheaper model and keep the one that works.

Which music model to use

  • A song from your own lyrics. Mureka V9 from 5 credits, or Music 2.6 for a fuller production.
  • Background music with no vocals. ElevenLabs Music with the instrumental setting, and the length you need.
  • A track from a description. Lyria 3 Pro, Google's music model.

Whose words and voice you may use

  • Write your own lyrics, or use ones you have the rights to.
  • Describe a genre or an era rather than naming a living artist, and do not ask for someone's voice.
  • You own what you generate under Percify's terms. It is not exclusive.
  • No watermark on any plan, including the free one.

Put the music on a video

A track is usually for something. Use it as the audio for a clip from the AI video generator, or have a face sing or speak it with a talking photo. For spoken words rather than singing, the AI voice generator is the right tool.

AI voice generatorAI video generatorMake a photo talkAI image generatorEvery model

Hear your idea as a song

Four music models, from 5 credits, no watermark, and failed runs refunded.

Make a song

Frequently asked questions

Everything you need to know before you start.

A model that writes and produces music from a description. You say what you want, a genre, a mood, the instruments, and it returns an audio track. Some models also take your own lyrics and sing them, which turns it into a song generator.

Yes. Mureka V9 and Music 2.6 both take lyrics you write and produce a sung vocal over the music. ElevenLabs Music and Lyria 3 Pro work from a description instead, and ElevenLabs Music can also make an instrumental.

From 5 credits on Percify, about 8 to 13 cents, up to from 15 credits for Music 2.6. The price can rise with options such as length or the number of songs, and the playground quotes the exact cost before anything runs. Credits work out at about 2 cents each depending on the plan, and failed runs are refunded.

Yes. ElevenLabs Music has an instrumental setting, and Music 2.6 can generate without vocals. Describe the instruments and the mood and leave the lyrics out.

Percify has a free plan with no watermark, and you start with free credits. Generating music costs credits on every plan, because every run costs us provider time. You can make a track before deciding whether to pay.

Percify terms say you own the AI generated content you create, subject to those terms, and output is not exclusive. Whether a track suits a specific commercial use, such as a broadcast or a store release, is worth checking against how you plan to publish it.

Describe the genre, the era and the mood rather than naming a living artist, and do not paste lyrics you did not write. You need the rights to anything you give the model, and a person’s voice and words are theirs.

Yes. Generate the track, then use it as the audio for a video. If you want a person to sing or speak it on screen, a talking photo lip syncs a portrait to any audio.