Mono to Stereo Converter (CRT)

Audio in ComfyUI is a tensor with a channel count, and a lot of the audio nodes and video-save steps quietly assume you've got two channels. The Mono to Stereo Converter (CRT) is the fix for the common case: your AI-generated voiceover or a mono clip comes in as one channel, and the next node in the chain wants stereo. In goes mono, out comes a proper left/right pair - zero mixing skill required.

It's part of the CRT-Nodes audio toolkit, the same category that gives you the Parametric EQ, the Tube Compressor, and the audio preview node. The whole suite treats audio as a first-class citizen, which is still rare enough in ComfyUI that having a small converter like this on hand is genuinely useful.

How it works

ComfyUI audio is a dict with a waveform tensor shaped (batch, channels, samples) plus a sample_rate. The node inspects the channel count:

• 1 channel → it duplicates the mono channel into left and right.
• Already stereo or multi-channel → passes through untouched.

That's the entire mechanism, and honestly it's the right one. Upscaling mono to stereo by simple duplication is the honest, artifact-free approach - you're not inventing width, you're just making the file conform. If you want real stereo width from mono, that's a different job (an M/S or widener effect), and it isn't what this node claims to do.

Inputs and output

One input, audio (an AUDIO tensor), one output, also called audio. Wire it in, wire it out. You could put it directly after a Video Loader Crawl's audio output or an audio loader, and feed the result to a preview, an EQ, or a video-save node that expects stereo.

When you'll reach for it

• Voiceover or TTS pipelines where the speech synth emits mono and your render/save step wants stereo.
• Video loads where the audio stream decodes as mono and a downstream mixer or preview assumes two channels.
• Prepping audio for analysis in a mono-averse node.

The one thing to remember is the reverse case: it won't collapse stereo down to mono, and it won't fix a stereo track that's phasey or out of balance. It's a format-fixer, not a mastering tool.

Install

Pack-level: ComfyUI Manager → search CRT-Nodes → install and restart, or clone https://github.com/PGCRT/CRT-Nodes into ComfyUI/custom_nodes. The audio nodes specifically depend on the pack's audio stack (pedalboard, librosa, soundfile and friends from requirements.txt), so if the audio category shows up broken, check that pip install -r requirements.txt actually completed - the pack skips optional nodes on import failure rather than crashing the whole UI.

Troubleshooting is otherwise quiet. If the output is still mono, check the input actually carries one channel - a silent or empty audio tensor passes through as "mono" and won't magically gain content.