Nodes/ComfyUI-VideoHelperSuite/Load Audio (Upload)πŸŽ₯πŸ…₯πŸ…—πŸ…’
ComfyUI Node Runs on cloud

Load Audio (Upload)πŸŽ₯πŸ…₯πŸ…—πŸ…’

Drop in an audio file with start and duration

By KosinkadinkΒ·Created 3 years agoΒ·Updated 3 months agoΒ· 1,727
Load Audio (Upload)πŸŽ₯πŸ…₯πŸ…—πŸ…’
    • audio
    • duration
    β—„audioβ–Ύβ–Ί
    β—„start_time0.00β–Ί
    β—„duration0.00β–Ί

    You need a soundtrack on your generated video, or an audio track to drive a lip-sync or audio-reactive workflow, and you want to upload it straight from the browser. That's this node. It loads an audio file from ComfyUI's input folder - with an upload button so you can add one on the spot - and outputs it in Comfy's standard AUDIO format, ready to plug into a Video Combine or any audio-consuming node. It's the audio-loading utility from Kosinkadink's VideoHelperSuite.

    Here's the honest context: ComfyUI has a built-in Load Audio node now, and it does the basic job. This VHS version predates that and stuck around because it offers two things the native one doesn't - a start offset and a duration limit. If you just need "load this whole file," the built-in is fine. If you need "load 8 seconds starting at the 30-second mark," you want this.

    How it works

    It reads the file and decodes it to the AUDIO type ComfyUI passes between audio nodes. The optional offset and length are applied at load time, so you're trimming before the audio ever enters your graph rather than dragging the whole file around.

    The inputs and outputs that matter

    The audio widget is a dropdown of files in your input folder, with "choose audio to upload" next to it. The two optional knobs are the reason to use this node:

    • start_time - an offset from the beginning of the file, in seconds. The audio starts playing from here.
    • duration - a hard cap on how much audio to load, in seconds. 0 disables it (load to the end).

    Outputs are audio (the loaded track, wire it into Video Combine's optional audio input) and duration (the length as a float, handy for driving downstream timing).

    How to install it

    ComfyUI Manager: search ComfyUI-VideoHelperSuite, install, restart. Or:

    cd ComfyUI/custom_nodes
    git clone https://github.com/Kosinkadink/ComfyUI-VideoHelperSuite
    

    then restart. No model downloads. Audio decoding leans on ffmpeg, which comes in through the imageio-ffmpeg dependency; a system ffmpeg on PATH helps with less common formats.

    Common issues & troubleshooting

    The file won't load. Some formats need a real ffmpeg. If an mp3/m4a/flac refuses, install system ffmpeg, put it on PATH, and restart.

    My audio and video are out of sync in the final render. That's usually not this node - it's the frame rate on Video Combine. If you trimmed or resampled your video frames, wire the loaded_fps output of a Video Info node into Video Combine's frame_rate so the durations line up.

    I don't see my uploaded file. Refresh the node or reload the page; ComfyUI rescans the input folder on refresh.

    Do I even need this? If you're not using start_time or duration, the built-in Load Audio is one less thing to think about. Reach for the VHS version specifically when you need to start partway in or clip the length.

    CategoryVideo Helper Suite πŸŽ₯πŸ…₯πŸ…—πŸ…’/audio

    Inputs (3)

    NameTypeDefaultDescription
    audioCOMBO0 options:
    start_timeoptFLOAT0.000–10000000β€”
    durationoptFLOAT0.000–10000000β€”

    Outputs (2)

    NameTypeDescription
    audioAUDIOβ€”
    durationFLOATβ€”