SaveSubtitle

The endpoint of every subtitle workflow in this pack. FasterWhisperTranscription turns audio into segments, FasterWhisperToSubtitle turns those into caption text, and SaveSubtitle is what actually writes the thing to disk - the one node in the chain that produces a file you can hand to a video editor, an uploader, or an ffmpeg command. If you build the whole pipeline and forget this node, you get a pretty string in a box and nothing on your hard drive.

How it works. This is an output node, which in ComfyUI means it always runs and it's the thing the queue reports on. It takes the SUBTITLE object (text plus its format - SRT or VTT - packed together by FasterWhisperToSubtitle), and writes it as a UTF-8 text file into ComfyUI's standard output directory. Filenames are auto-incremented: with the default prefix you get subtitle_0000.srt, then subtitle_0001.srt, and so on. Nothing is ever overwritten, so you can run the workflow twenty times and keep all twenty files. The node then returns the full path it wrote to, which is genuinely useful if you're chaining to something downstream that needs to know where the file landed.

Two inputs, one output:

• subtitle (SUBTITLE, required) - wire from FasterWhisperToSubtitle.
• prefix (STRING, optional, default "subtitle") - the base name before the auto-number. Set it to something meaningful like my_video_captions and you get my_video_captions_0000.srt instead of hunting through a folder full of subtitle_XXXX.

Output is output_path (STRING) - the absolute path of the file just written.

Where the file goes. This trips up people more than anything else about the node: there's no folder picker, and you can't choose where it saves. It writes to ComfyUI's output directory (ComfyUI/output for a normal install, inside the portable folder if you run the portable build). That's the same place your generated images land, and it's deliberate - it keeps everything in one known place. If you need the caption moved or renamed, grab it from there or rename prefix so the next run names it better.

The workflow it completes. The pack's included example wires it as: InputFilePath → FasterWhisperTranscription → FasterWhisperToSubtitle → SaveSubtitle. That's the whole thing - the four nodes are a reference implementation of faster-whisper for ComfyUI, and SaveSubtitle is the one that makes the output real.

Install & troubleshooting. Same as the pack: ComfyUI Manager search "ComfyUI-faster-whisper", or cd ComfyUI/custom_nodes && git clone https://github.com/jhj0517/ComfyUI-faster-whisper then pip install -r requirements.txt. Nothing extra to download - this node only touches files ComfyUI already manages. If you run the workflow and can't find your subtitle, you almost certainly looked in the wrong place: it's in the output folder, named subtitle_0000.srt (or prefix_0000), not next to your video and not next to your audio file. And if the node errors with "Output format not supported," check that the SUBTITLE coming in is actually from FasterWhisperToSubtitle - the format has to be .srt or .vtt, and a hand-typed string won't carry the right shape.