ace-step

To generate music from text tags, provide genre, mood, and instrument tags along with optional structured lyrics including section markers. The Skill routes this input to the RunComfy ACE Step endpoint to produce a stereo music track of your specified duration.

Yes, you can regenerate a specific time range inside an audio track using audio inpainting. By specifying start and end timestamps in your request, the Skill targets and regenerates only that selected segment without altering the rest of the recording.

The best way to extend an existing music track is through audio outpainting. The Skill takes your existing audio and generates new content before or after it, allowing you to seamlessly lengthen the track with optional structured lyrics.

Yes, multilingual vocal generation supports broader language coverage by utilizing the ACE Step 1.5 endpoint. This endpoint provides expanded language support and cleaner handling of structured lyrics for your text-to-audio workflows.

You must register YAML metadata containing the name and description before routing calls to the RunComfy CLI endpoints. This structured metadata ensures the appropriate JSON input schemas for tags, lyrics, durations, timestamps, and seeds are correctly processed.

You should use audio outpainting instead of generating a new track when you want to preserve and build upon an existing recording. Outpainting seamlessly extends your current audio before or after the existing timeline, whereas fresh generation creates an entirely new composition.