SaveAudioBL

SaveAudioBL is the ComfyUI-CUP node that writes generated audio to disk as FLAC - the "BL" meaning it's the variant the Blender bridge routes audio through. It takes a ComfyUI AUDIO dict and saves each waveform in it to output/audio/ with a random _temp_ prefix, one .flac file per batch item. That's the whole job.

In practice this is the node that appears at the end of an audio-generation or AI-voice workflow running inside the AIGODLIKE Blender pipeline - think a character line generated by an audio model, then fed back into Blender's timeline as a sound strip. ComfyUI's audio support grew up around video-generation packs (audio comes out of models like Wan and LTX-Video as a tensor plus sample rate), and this node is the minimal save-for-Blender endpoint for that data.

How it works

• audio (AUDIO) - the only input: the standard audio dict carrying a waveform tensor and its sample rate.

It's an output node, so no outputs to wire. Under the hood it uses torchaudio to encode each waveform as FLAC, and here's the one thing worth knowing: torchaudio is imported lazily, inside the save call, and it's not listed in the pack's own dependencies. ComfyUI core normally provides torchaudio, but if your install is stripped down or a venv reshuffle left it out, this node is where you'll discover it - the error will be a plain ImportError: No module named 'torchaudio' at save time. If you hit that, pip install torchaudio (matching your torch build) fixes it.

Install

Part of ComfyUI-CUP. ComfyUI Manager → search ComfyUI-CUP, or:

cd ComfyUI/custom_nodes
git clone https://github.com/AIGODLIKE/ComfyUI-CUP.git

then restart. No model files to download.

The honest assessment

SaveAudioBL is a thin utility with a narrow purpose, and it's honest about it. If you're generating audio in a normal web workflow, you probably already use a save node from a video or audio pack you're more familiar with - this one is essentially the pack's internal audio sink, renamed so the Blender addon can find it. It works, it produces valid FLAC files, and if you're inside the AIGODLIKE ecosystem it's the one that keeps your sound in sync with your renders. Outside that ecosystem, it's just a FLAC writer with a confusing name.