AudioSpeedChange🐶

Its display name is literally "AudioSpeedChange 🐶" - LLM Party's author sprinkles dog emoji on a handful of nodes, no idea why, it's just part of the pack's personality. What it actually does is exactly what it says: take an audio file, change its playback speed, write out the result.

It fits into the audio corner of this pack, which the README frames as part of a "one-stop LLM + TTS + comfyui workflow" pitch for streaming/media work. The realistic use case: your agent generates narration through a TTS node, and before that audio goes into a final export or gets muxed with generated video, you need to nudge its pacing - speed it up slightly to fit a time budget, slow it down for clarity. That's this node's job.

Inputs and outputs that matter

• input_audio_path / output_folder_path - this node works off file paths on disk, not a piped-in AUDIO socket for the source. Point it at an existing audio file and an existing output folder.
• speed_factor - default 1 (no change), floored at 0, adjustable in 0.1 steps. Above 1 speeds up, below 1 slows down.
• suffix - default _processed, appended to the output filename so you don't clobber the original.
• is_enable - the pack-wide skip switch.

Two outputs: audio, an in-graph AUDIO object so you can keep chaining without leaving ComfyUI, and audio_path, the string path of the file it wrote - handy if a downstream step (or a node outside this pack) needs the path rather than the object.

Installing it

Search comfyui_LLM_party in ComfyUI Manager and install, or clone it by hand:

cd ComfyUI/custom_nodes
git clone https://github.com/heshengtao/comfyui_LLM_party

Run pip install -r requirements.txt inside your ComfyUI Python environment afterward - this is a large, all-in-one pack (LLM APIs, local model loading, RAG, TTS, and dozens of tool nodes), so its dependency list is bigger than a typical single-purpose audio pack. Restart ComfyUI when it's done.

Common issues

Audio processing nodes like this one typically lean on ffmpeg under the hood for anything beyond raw waveform math - if ffmpeg isn't installed and on your system PATH, this is one of the first nodes in the pack that'll surface it as a cryptic error rather than a clear "install ffmpeg" message. Worth checking that first if the node errors immediately on run.

The other trap is paths. output_folder_path generally needs to already exist - most file-writing nodes in this ecosystem won't create missing directories for you - and input_audio_path has to be visible from wherever the ComfyUI process is actually executing. If you're running on a remote or cloud GPU rather than your own desktop, a path that's valid on your local machine (like a Windows desktop path) means nothing to the executor; the file needs to already be on that machine's filesystem, not just yours. And since this pack pulls in a heavy set of requirements at install time, a partial or failed pip install is a common reason nodes from it silently don't appear in the node list at all - re-run the install and actually read the pip output if that happens.