AceMusic Caption/Style Input

AceMusic Caption/Style Input is the text box where the song's sound gets described - the companion to AceMusic Lyrics Input, and the single biggest quality lever you have in this pack. It takes one multiline string and passes it through unchanged as a caption output. Nothing clever happens inside; the cleverness is what you type.

ACE-Step reads these as natural-language style tags, and the README gives you a workable vocabulary to mix:

• Genre: pop, rock, electronic, jazz, classical, hip-hop, r&b, country, folk, metal, indie, j-pop, k-pop
• Vocal: female vocal, male vocal, duet, choir, instrumental
• Mood: energetic, melancholic, uplifting, calm, aggressive, romantic, dreamy, dark
• Tempo: slow, medium, fast
• Instruments: piano, guitar, drums, synth, strings, bass, violin, saxophone

A good caption looks like: j-pop, female vocal, energetic, bright synthesizer, catchy melody. A lazy one looks like song. Guess which one produces something you'd actually listen to. The model does what it does with any prompt - diffusion models are forgiving of short prompts - but the detail-to-quality relationship here is real, and it's the cheapest improvement available to you.

Wire the caption output into AceMusic Generator (Text2Music), or into the editing nodes (Cover, Retake, Extend, Edit) where a caption steers how the original audio is transformed. It's just a STRING, so anything that produces text can feed it - including AceMusic Create Sample, which generates a caption from a natural-language query. In the direct Generator workflow the caption socket is force-input, which is exactly why this node exists: you need a string source anyway, and keeping the style description in its own visible box beats hunting through widget menus.

Installing it

Via ComfyUI Manager, search "ComfyUI-AceMusic", or:

cd ComfyUI/custom_nodes
git clone https://github.com/hiroki-abe-58/ComfyUI-AceMusic.git
cd ComfyUI-AceMusic
pip install -r requirements.txt
pip install git+https://github.com/ace-step/ACE-Step.git

It's part of the pack, so the install is the pack install - including the separate ACE-Step engine, without which nothing actually generates.

Common issues

• Everything comes out sounding generic - your caption is too thin. Pack in genre + vocal + mood + an instrument or two.
• Caption describes a style, vocals still don't match - the vocals follow vocal_language in AceMusic Settings; make sure the language matches your lyrics and your described vocal type.
• Caption not reaching the generator - the socket is force-input, so an unwired generator simply won't run. Connect it.

There's no configuration to get wrong here - just a text box and a wire. Spend the extra thirty seconds on the sentence, and the whole workflow gets better for free.