Get Audio Stream

Get Audio Stream is the audio half of the pack's "select a stream by index" pair - the sibling of Get Video Stream, in ComfyUI Core Video Nodes (Immac/ComfyUI-CoreVideoMocks). Its proposed job is clean: you've got a batch of audio streams (say, a dialogue track and a music track), you want one of them, and this node picks it by number. In a real muxer that's exactly how you'd reach into a file and pull track N.

Two inputs: streams (a force-input STRING - wire it from Split Video's audio_streams output or Batch Audio Streams), and stream_index, an INT defaulting to 0 with the author's own tooltip: "The zero-based index of the stream to extract." 0 for the first track, 1 for the second, and so on. One output, stream, which in a working pipeline would feed an audio decoder or a transcribe node.

What actually happens

It returns the string "Mock Audio Stream", every time, regardless of index. The index input is real and ranged (0 to 10000), but nothing depends on it. This is the pack's signature move: GitHub describes it as "A set of mock nodes for RFC purposes," the README calls itself "a mock of a possible implementation" of core video nodes, and every output is a hardcoded placeholder. Two stars, no community footprint, dormant since March 2025.

Install

cd ComfyUI/custom_nodes
git clone https://github.com/Immac/ComfyUI-CoreVideoMocks

Restart ComfyUI. No pip packages, no models, no ffmpeg - pure Python on ComfyUI's own folder_paths, also via ComfyUI Manager ("ComfyUI Core Video Nodes").

The honest take

Use it to learn the pattern: batch streams, select by zero-based index, pass the single stream onward. That's a genuinely sensible core-video API shape, and the tooltip on stream_index is the author's own spec for it. But it produces nothing you can hear. Real audio extraction from a file is ffmpeg territory - ffmpeg -i in.webm -map 0:a:1 out.m4a - and VideoHelperSuite (VHS) is where actual video/audio work happens inside ComfyUI. This node is a label on an empty box.