Nodes/DeepFuze/Play Audio
ComfyUI Node

Play Audio

Play audio on the machine running ComfyUI — not your browser

By SamKhoze·Created 2 years ago·Updated about a year ago· 461
Play Audio
  • audio

    DeepFuze's voice cloning hands you an AUDIO object, and the fastest way to know whether the clone actually sounds like the reference is to hear it. This node plays any incoming AUDIO through the speakers of the machine that's running ComfyUI. It's a throwaway utility - a quick A/B check after TTS generation or a sanity listen on a Lipsync's audio output - not something you'd ship in a finished workflow.

    How it works

    The node writes the incoming audio to a temp wav, loads it through pydub, and plays it with sounddevice.play on the machine's default audio output. Nothing gets saved permanently, nothing comes out the other side. It's marked as an output node, so it just runs and finishes.

    The one gotcha that bites everyone

    "Play" means the server's speakers. If ComfyUI is running on your desktop, great - you'll hear it. If you're running ComfyUI on a headless box, a cloud GPU instance, or inside WSL, there is no audio device, so the node either plays into the void or errors out depending on the setup. People hit this constantly and assume the node is broken; it's not, you just have no speakers where the code runs. When that's the case, skip this node and use DeepFuze Save to write the clip to disk and listen on your own machine.

    Inputs

    • audio (required) - the clip to play, typically the AUDIO output of the DeepFuze TTS node or the audio output of a Lipsync/FaceSwap run.

    There are no outputs. It's a dead end by design, so it goes at the end of a branch.

    Install

    Same as every node in the pack - one install covers all of DeepFuze:

    cd ComfyUI/custom_nodes
    git clone https://github.com/SamKhoze/ComfyUI-DeepFuze
    

    then pip install -r requirements.txt inside the pack folder and restart ComfyUI. Windows users can instead install from ComfyUI Manager (search "DeepFuze"). The heavier prerequisites from the README still apply to the pack as a whole: Visual Studio C++ Build Tools on Windows, optional CUDA 11.8 + cuDNN for GPU (CPU is fine), and model files placed in ComfyUI/models/deepfuze/ for the face/video nodes - though this little playback node needs none of them.

    Anything else?

    Just the usual pack caveat: DeepFuze is unmaintained and the README states it's not compatible with the latest ComfyUI, so treat it as working-on-the-version-you-installed-it-on. For what this node does, that's an acceptable risk - if it plays, it plays. If it doesn't and you're remote, you now know why.

    CategoryDeepFuze

    Inputs (1)

    NameTypeDefaultDescription
    audioAUDIO

    Outputs (0)

    No outputs