Background Music

Generate original, royalty-free instrumental music for reels and demo videos from a text brief - fully offline. A wrapper drives ACE-Step 1.5 (the music model); you (the model) shape the brief into a good prompt, generate a couple of candidates, let the user pick, and optionally mix the winner under a narration.

Pairs with the voice-clone-narration skill: generate the voiceover there, the music here, then mix_voiceover.sh ducks the music under the narration for a finished reel audio track.

flowchart LR
  Brief["brief: genre, mood, bpm, duration"] --> Gen["generate_music.py<br/>ACE-Step 1.5 (MLX), instrumental"]
  Gen --> Mp3["candidate mp3s"]
  Nar["narration.mp3 (optional)"] -.-> Mix["mix_voiceover.sh<br/>ffmpeg sidechain ducking"]
  Mp3 -.-> Mix
  Mix -.-> Reel["reel-audio.mp3"]