Skip to main content
Load Audio

Load Audio · screenshot coming soon

​
What this node does

Load Audio is an instant Input stage. It reads audio files from ComfyUI’s input/ folder (or uploads written there), registers a project snapshot, and outputs COMFYTV_AUDIO. Use it for existing BGM, voice WAV, demuxed tracks, etc.—not for AI generation. Selecting a file immediately sets output without invoking Speech or Music Stage workflows. vs Load Audio from Asset: this node = raw disk files; asset loader = project library from runs and imports.

​
When to use it

  • Import external narration, music (FLAC/WAV/MP3).
  • Feed reference audio into Speech Stage (voice cloning).
  • Wire into Video Stage → audio for image+audio-to-video.
  • Optional track on Director Timeline.

​
How it works (why ComfyTV is designed this way)

  • Stage vs Run: No ▶ Run—select or upload to output.
  • Snapshots: Downstream Runs read the current file URL; they won’t re-read this node automatically.
  • No workflow: Not tied to workflows/audio/ or workflows/speech/ generators.

​
Types (COMFYTV_* vs native ComfyUI)

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

​
UI and parameters

​
audio

  • What: Dropdown of audio in input/ plus upload.
  • What to set: Pick or upload (WAV, MP3, FLAC, etc.).
  • Effect: Defines source audio for downstream; must be a format ComfyUI recognizes.
  • Common mistake: Looking here for Music/Speech Stage outputs—use Load Audio from Asset or wire from the generator.

​
project_id / parent_output_id (internal)

  • Hidden; maintained by Project and wiring.

​
Outputs

​
Step by step for beginners

  1. Add Node → ComfyTV → Input → Load Audio.
  2. Wire a Project node (recommended).
  3. Put files in input/ or upload here.
  4. Select in dropdown—no Run needed.
  5. Connect audio downstream, e.g. Video Stage → audio.
  6. ▶ Run the downstream generator.
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: Load vs generate (Speech/Music Stage)?
A: Load Audio = existing files, no GPU inference. Speech/Music = AI creation with workflow + Run. Loaders import; generators create. Q: Load Audio vs Load Audio from Asset?
A: This = input/ disk. Asset = project library after generation/import. Q: Type mismatch?
A: Downstream must expect COMFYTV_AUDIO; native AUDIO needs → ComfyTV Audio Bridge. Q: No Run button?
A: Normal—same as Load Image/Video.
  • Load Audio from Asset — library picker.
  • Speech Stage / Music Stage — AI audio generation.
  • AudioVideoDemuxAudioStage — extract audio from video.
  • Director Timeline — arrange audio on a timeline.
  • → ComfyTV Audio (Bridge).