Start with function, not fame

Begin with what the audio is for: a background cue, an original vocal demo, a practice loop or a short intro. A purpose makes later choices about duration, structure and density much clearer than asking for a copy of an existing song.

Use five concrete controls

  1. Genre and palette: describe broad musical language, such as lo-fi instrumental or acoustic cinematic underscore.
  2. Mood: specify calm, tense, hopeful or reflective rather than naming a performer.
  3. Tempo: include an approximate BPM when timing matters.
  4. Instruments: name sounds that define the arrangement.
  5. Structure: state how the cue should open, develop and end.
Example: Create 60 seconds of warm lo-fi instrumental background audio at about 78 BPM, with dusty electric piano, muted bass and soft drums; short intro, relaxed loop and clean outro; original composition only, without recognizable existing melody or copied lyrics.

Why CleanStems avoids imitation prompts

A named artist, recognizable melody or cloned voice introduces rights and platform-policy questions before a track is even edited. CleanStems therefore focuses on generic musical descriptions and audio the user may legally process.

After generation

Review the generation platform's current rights terms and disclosure features. If you are permitted to process the output, you can separate stems, note tempo, estimate a key or build a practice loop.

Build a Prompt Locally

Open the CleanStems AI Music Prompt Generator to assemble an original prompt in your browser.