Skip to main content
Music Stage (Audio Stage)

Music Stage (Audio Stage) · screenshot coming soon

​
What this node does

Shown as Music Stage (node id: ComfyTV.AudioStage). Write genre, mood, instruments, and BPM in main_prompt as tags, optionally fill lyrics, set duration, click ▶ Run, and get a COMFYTV_AUDIO music snapshot. Built-in ACE-Step v1 Song uses ACE-Step 3.5B: empty lyrics = instrumental; filled lyrics = sung vocals. This is not TTS. For narration or dialogue, use Speech Stage.

​
When to use it

  • Background music, lo-fi, electronic, jazz (leave lyrics empty).
  • Demo songs with lyrics + style tags.
  • Audio track for Video Stage IA2V (wire audio).
  • Source audio for Timeline or Audio Extract chains.

​
How ComfyTV stages work

  • ▶ Run executes only the subgraph in workflows/audio/.
  • Snapshots: audio URL (e.g. FLAC) is stored in the project.
  • ComfyTV maps tags → ACE-Step TextEncodeAceStepAudio.tags, lyrics → .lyrics, duration → EmptyAceStepLatentAudio.seconds.
Native ComfyUI has no “read this script aloud” TTS; ACE-Step / Stable Audio are music generators—ComfyTV splits music and speech into two stages.

​
Types (COMFYTV_AUDIO vs native ComfyUI)

Conversion:
  • Native → ComfyTV: → ComfyTV Audio
  • ComfyTV → native: ← ComfyTV Audio
See bridges.md.

​
Music Stage vs Speech Stage (important)

​
Parameters

​
workflow

Backend from workflows/audio/:
  • ACE-Step v1 Song (ace-step-v1-song.json)—tags + optional lyrics + duration.
Docs: workflows/audio/README.md.
Model: ace_step_v1_3.5b.safetensors in models/checkpoints/ (models.md).

​
main_prompt (style tags)

Keywords for genre, mood, instruments, tempo. Example: lo-fi, jazz piano, rainy night, 90bpm, soft female vocals Mistake: Pasting a narration script here—that belongs on Speech Stage; Music Stage prompts describe music, not a reading script.

​
duration_s

Length in seconds, 1–240, default 30.

​
lyrics

  • Empty: instrumental / no vocals.
  • Filled: song mode—ACE-Step generates singing to the lyrics.

​
custom_params

Sidebar: seed and other workflow params.

​
Outputs

​
Step by step for beginners

  1. Add Project, then ComfyTV → Generate → Music Stage.
  2. workflow → ACE-Step v1 Song.
  3. main_prompt: cinematic orchestral, epic, 120bpm, no vocals.
  4. Leave lyrics empty, duration_s 30.
  5. Place ace_step_v1_3.5b.safetensors in checkpoints.
  6. ▶ Run and preview on the node.
  7. For IA2V: Video Stage → IA2V, Music audio → Video audio, plus a portrait on images.
Vocal song: fill lyrics, tags like pop, upbeat, female vocal → Run.
This page covers one node only. For end-to-end workflows, multi-stage pipelines, type conversion, and design rationale, see the ComfyTV user guides on GitHub:

​
Repository and workflows

​
FAQ

Q: Sounds like speech, not a song?
A: You may want Speech Stage; Music Stage uses tags + lyrics for music. Q: Model not found?
A: Download ACE-Step checkpoint to models/checkpoints/, restart ComfyUI. Q: Cannot wire audio to Video Stage?
A: Both ends must be COMFYTV_AUDIO; bridge native AUDIO first. Q: Need read-aloud dubbing?
A: Use Speech Stage, not this node.
  • Speech Stage—TTS (do not confuse with Music Stage).
  • Video Stage (IA2V)—audio drives visuals.
  • Audio Extract Vocal / Bg Stage—split vocals or backing.
  • Director Timeline Stage—multi-track timeline.
  • Bridge → / ← ComfyTV Audio—native interop.
  • Project—project context.