Preview Audio ♾️Mixlab

This is the audio version of PreviewImage. You've got an AUDIO output somewhere in your graph - text-to-speech, a voice clone, a generated sound - and you want to actually hear it without saving a file and hunting for it in a folder. Wire it into Preview Audio and you get a little player right on the node. Press play, listen, done.

It exists because ComfyUI grew audio nodes and you need a way to audition their output inline. Mixlab has a speech-synthesis and speech-recognition corner (the README shows text-to-audio and voice workflows), and this is the "did that sound right?" node for all of it. Nothing more, nothing less.

How it works

It's a terminal output node. You feed it audio; it renders a playback widget in the node UI and lets the front-end play it back. There's no processing, no conversion you configure - it just surfaces the audio for listening. Like all preview nodes, the point is inspection, not transformation.

Inputs and outputs

• audio (AUDIO, required) - the audio to play. That's the whole input surface.

There are no outputs - it's an endpoint. If you also want to keep the audio, put a save node in the chain; Preview Audio is purely for listening in the UI.

How to install it

Ships with the Mixlab pack.

• ComfyUI Manager: search comfyui-mixlab-nodes, install, restart.
• Manual: cd ComfyUI/custom_nodes && git clone https://github.com/shadowcz007/comfyui-mixlab-nodes.git, install requirements (install.bat on Windows or ../../../python_embeded/python.exe -s -m pip install -r requirements.txt), restart.

Common issues

No sound / no player: make sure something actually produces an AUDIO output into it and that the graph ran. As an output node it only renders after execution, so if the upstream audio node errored you get an empty player. Also check the obvious - browser tab isn't muted, system volume is up.

The AUDIO types don't connect: ComfyUI's AUDIO format has evolved, and a very old audio node from another pack might emit a shape this doesn't accept. If the socket won't link, update the pack (and the upstream audio pack) so both speak the current AUDIO type.

It's a live UI feature: playback happens in the browser front-end, so a headless/API run has nobody to play it to. On a hosted or serverless runner, Preview Audio is only meaningful when you're watching in the UI - for a batch/API job, save the audio instead and fetch the file. That's not a bug, it's just what a preview node is.