AI Song Prompt Generator
Pick a genre, a mood and a tempo. You get three ready-to-paste prompts, built the way that works best in AI music apps: a specific scene, a real tempo, named instruments and a production texture.
Your prompts
- Variation 1
Late-night lo-fi hip-hop at 79 BPM, warm upright bass, muted jazz guitar, mellow saxophone, vinyl crackle and tape saturation, dreamy and hazy, instrumental, no vocals
- Variation 2
Dusty chillhop at 77 BPM, mellow saxophone, warm upright bass, muted jazz guitar, warm and slightly detuned, soft and floating, instrumental, no vocals
- Variation 3
Rainy-day lo-fi at 77 BPM, soft Rhodes electric piano, dusty boom bap drums, mellow saxophone, lo-fi filtered drums with soft swing, gentle and dreamlike, instrumental, no vocals
Runs entirely in your browser. Nothing you type is sent anywhere.
Turn your prompt into a finished song.
- 1Copy the variation you like.
- 2Open Muziko and choose Describe mode.
- 3Paste it in and generate. A full song takes about 8 to 15 seconds.
On a Mac with Handoff turned on, a prompt you copy here can be pasted straight into Muziko on your iPhone.
The six parts of a prompt that actually steers.
Every prompt the generator builds has these six parts. It’s the same structure we landed on after testing dozens of genres for the Muziko blog.
Genre plus a scene or era
"Soul" is vague. "Detroit 1966 Motown soul" points at one specific sound.
Berlin warehouse technoA tempo number
Without a number, tracks land on a generic middle tempo. Every genre has its own range.
at 112 BPMTwo or three named instruments
"Drums" gets you a preset. "Brushed drums, walking upright bass" gets you a band.
warm Rhodes, upright bass, muted trumpetProduction texture
How it is recorded matters as much as what is played: tape, reverb, room, grit.
vinyl crackle and tape saturationMood written as sound
"Sad" is a feeling. "Minor key, bittersweet" is something a model can play.
melancholic minor keyVocals, or no vocals
Say it either way. If you leave it out, the app picks for you.
instrumental, no vocalsMistakes the generator avoids for you.
Describing a feeling instead of a sound
"Calm ambient music" gets you a generic synth pad in a major key.
Read the ambient guide →Using the word "epic"
It brings in every trailer element at once, so the track has nowhere to build.
Read the cinematic guide →Not saying what you don’t want
Techno drifts into festival EDM unless the prompt says "no drop, no supersaw".
Read the techno guide →Not naming the signature instrument
Amapiano without the words "log drum" turns into jazzy house.
Read the amapiano guide →Go deeper on prompting
Frequently asked questions
What is an AI song prompt?
A short text description that tells an AI music app what to make: the genre, tempo, instruments, production texture, mood and whether there are vocals. The more specific it is, the closer the result lands to what you had in mind.
How long should an AI song prompt be?
One or two sentences, roughly 25 to 50 words. That is enough to name a genre and scene, a tempo, two or three instruments, a texture, a mood and the vocals. Past that, extra detail rarely helps and can pull the result in conflicting directions.
Do these prompts work in Suno, Udio or other AI music apps?
They are written for Muziko’s Describe mode, but they are plain-language descriptions, not commands tied to one app, so they work as a starting point in other text-to-music tools too.
Is the generator free? Do I need an account?
It is free, needs no account and has no usage limit. It runs entirely in your browser, so nothing you type is sent to a server.
Why does every prompt include a BPM number?
Without a tempo, AI music apps tend to settle on a generic middle tempo that doesn’t fully commit to any genre. Each genre in the generator has its own range, for example 110 to 115 BPM for amapiano and 128 to 140 for techno. The Slow, Medium and Fast buttons pick from within that range.
Can I use a generated prompt with my own lyrics?
Yes. In Muziko, use Write Lyrics mode for your words and use the generated prompt to set the style. Choose Female vocal or Male vocal in the generator so the prompt describes a sung track rather than an instrumental.