- frames
- audio_file
- fps
- frame_count
- height
- width
Despite the name, video2audio does more than pull audio out of a video - it also hands you the video's frames. It's the older name for what Mana Nodes now calls Split Video, and the "2audio" label undersells it: yes, it extracts the audio track, but it outputs the image frames, the fps, and the dimensions too. Think of it as "explode this video into its parts."
Why you'd reach for it
It's the entry point to the captioning workflow. ComfyUI can't chew on an .mp4 directly, so before you can overlay captions or restyle a clip you have to decompose it - frames to work on, audio to transcribe. This node does both in one pass, which is why it sits at the top of the "auto-subtitle a video" graph: audio goes to a transcription node, frames go on as the caption background.
The inputs that matter
video- the source clip, chosen from a dropdown of files ComfyUI can see (your input folder).frame_limit- default 16, capped high. The max frames to extract. Leave this low unless you mean it - pulling a whole long clip into one batch is a memory bomb.frame_start- default 0. The frame to start from, so you can grab a middle slice.filename_prefix- output path for the extracted audio, relative toComfyUI/output(defaultaudio_files\audio).
Outputs are the full set: frames (the IMAGE batch), audio_file (path to the saved audio, string - this is the "2audio" part), fps, frame_count, height, and width. That fps output matters: feed it into your transcription/caption node so word timing lines up with the real frame rate.
Installing it
It's in the Mana Nodes pack. ComfyUI Manager → search Mana Nodes → install → restart. Or clone:
cd ComfyUI/custom_nodes
git clone https://github.com/ForeignGods/ComfyUI-Mana-Nodes.git
Install requirements and restart. Decoding and audio extraction go through FFmpeg, so it needs to be present on your machine; a "can't open the video / no audio came out" error usually means FFmpeg is missing or the clip has no audio track.
Common issues
Naming, first: a current install may show Split Video instead of video2audio - same node, newer label. If a downloaded workflow references video2audio and won't load, rewire it to Split Video.
Then the practical ones. Don't crank frame_limit on a long clip; a single giant frame batch will exhaust memory and kill the run. Use frame_start to page through a long video instead. Pass the fps output downstream rather than hardcoding a frame rate - the caption pipeline times everything by frame number, so a mismatch here makes subtitles drift. And if the video dropdown is empty, put your file in ComfyUI's input folder and refresh so the node can list it.
Inputs (4)
| Name | Type | Default | Description |
|---|---|---|---|
| video | COMBO | 0 options: | |
| frame_limit | INT | 161–10240 | — |
| frame_start | INT | 00–4294967295 | — |
| filename_prefix | STRING | audio_files\audio | — |
Outputs (6)
| Name | Type | Description |
|---|---|---|
| frames | IMAGE | — |
| audio_file | STRING | — |
| fps | INT | — |
| frame_count | INT | — |
| height | INT | — |
| width | INT | — |