Convert Audio Channels

ConvertAudioChannels fixes the second-most-annoying thing about audio in ComfyUI: the moment you try to combine a stereo clip with a mono clip and everything refuses to cooperate. It changes the channel count of an AUDIO track - mono to stereo, stereo to mono - and leaves everything else alone. It ships in lebrosoft/ComfyUI-VideoChatWrapper, a small pack whose audio utilities are honestly more useful than its namesake video-chat feature, and this node is the quiet workhorse of that toolkit.

How it works

The mechanism is a faithful copy of the convert_audio_channels helper from Facebook's AudioCraft library, which the pack's source credits. The logic:

• Stereo → mono: averages the two channels together (downmixing).
• Mono → stereo: duplicates the single channel onto both.
• Stereo → 4+ channels: keeps the first N channels and drops the rest - though this node only exposes 1 or 2, so you won't hit that branch from the UI.

Note what it doesn't do: it doesn't resample, doesn't change loudness, doesn't touch duration. It's purely a channel-count conversion, and it's instant.

The inputs

• audio - the AUDIO track to convert.
• to_channels - an integer, 1 or 2. Default 1. That's the entire UI.

The output is audio, an AUDIO track with the requested channel count.

Why you'll actually use it

Almost always as prep for something else. The other audio nodes in this pack are strict about channel layout - ConcatAudio and JoinAudio throw an error the moment you mix mono and stereo inputs - so ConvertAudioChannels goes in front of them like a diplomatic escort. It's also what you want before feeding audio into a speech model or a dataset pipeline that expects mono, or before dropping a mono narration into a stereo mix.

The practical tip: decide your channel standard early. If everything downstream wants stereo, convert once at the start of your graph rather than chasing a mono clip through three nodes and discovering the mismatch at the join.

Installing it

Same story as every node in this pack. Via ComfyUI Manager, search for ComfyUI-VideoChatWrapper and install, or:

cd ComfyUI/custom_nodes
git clone https://github.com/lebrosoft/ComfyUI-VideoChatWrapper

Then restart ComfyUI. The audio half of the pack needs torchaudio - already bundled with ComfyUI's portable distribution, but worth pip install torchaudio if you use a hand-rolled environment. The requirements.txt only lists qwen_vl_utils for the video-chat nodes, so audio users usually install nothing extra.

Realistic expectations: this pack is nearly invisible in the community - essentially zero discussion, zero usage stats. For a utility this small that's fine; the code is short and it does exactly what it says. There's not much to go wrong, and when there is, it's the usual ComfyUI custom-node story: a dependency mismatch in your Python env, not a flaw in the node.