Load ASS subtitles

ASSSubtitleReader is the odd one out in this pack: it doesn't touch ffmpeg at all. It's a pure-Python parser that reads an .ass subtitle file and hands the text to your workflow as strings. Why would you want that? Because subtitles aren't just output - they're data. You can hand the dialogue to an LLM node for translation or rewording, feed it into a search, pull the line count for pacing checks, or just dump the raw file into a prompt. For anime subs especially, the encoding list alone is worth the install.

It parses the ASS format properly - sections, Dialogue: lines, styles - and gives you three outputs:

• all_text - the entire file content, verbatim, as one string.
• dialog_only - just the spoken dialogue, one line per subtitle, with the timing and position cruft stripped. This is the one you feed to an LLM.
• styles_info - the raw Style: lines from the [V4+ Styles] section, so you can see what fonts/colors the file uses.

The inputs that matter

• ass_file_path - path to the .ass file. Same story as the rest of the pack: plain text, no file picker, so use an absolute path or one relative to where ComfyUI runs.
• encoding - default utf-8, but here's the genuinely nice part: utf-8-sig, latin-1, cp1252, shift-jis, euc-jp, gbk, big5. Old or Japanese subtitle files in the wrong encoding are a classic mojibake disaster, and this node gives you the fix without a round trip through a text editor.
• filter_style - only keep Dialogue: lines whose style name matches. Handy when a file mixes "Default" dialogue with OP/ED credit styles you don't want in your transcript.
• strip_formatting - set to True and it strips ASS override codes like {\\pos(..)} and {\\b1} from the dialogue, leaving clean text. Do this before feeding an LLM; the formatting codes will just confuse it.
• max_lines - cap how many dialogue lines come out (0 = everything).

One thing to know: this is an ASS parser, not a universal subtitle reader. Feed it a .srt file and dialog_only will come back empty because there's no Dialogue: syntax to match - all_text still returns the raw content, but the parsing is ASS-shaped. If you're dealing with SRT, convert it first.

Using it with the rest of the pack

The natural pipeline: ASSSubtitleReader → LLM (translate/rewrite the dialogue) → ASSSubtitleSave → SubtitleEmbedding. You can rebuild a whole subtitled video from scratch, or just use the reader as an inspection tool. It's also genuinely handy as a way to extract the subtitle track you pulled with SubtitleExtraction and turn it into workflow-usable text.

Installing

It ships in the ComfyUI-ASSSSA pack. Install via ComfyUI Manager (search "ComfyUI-ASSSSA") or:

cd ComfyUI/custom_nodes
git clone https://github.com/HighDoping/ComfyUI_ASSSSA

Restart ComfyUI and you're done. No pip packages, no models - and unlike the ffmpeg-backed nodes, this one works even if your FFmpeg install is missing, since it never calls out to it.