Chinese to English Translate
Translate Chinese to English inside the graph — offline, no API key
- translated_text
If you download workflows from Chinese-speaking creators - and this pack's own README is half Traditional Chinese - you'll eventually hit a wall of prompts you can't read. ChineseTranslate exists so you don't have to tab out to Google every time. It's a single-purpose node: Simplified or Traditional Chinese text in, English out, running entirely on your machine. The "no API key, no cost, no privacy leak" part is the whole appeal, because the alternative (an LLM node) either needs a key or needs you to babysit a local model.
How it works
Under the hood it uses Argos Translate, the open-source neural machine translation library (the same engine behind LibreTranslate). The node auto-installs the argostranslate Python package on first use, then downloads the zh→en language pack. Two things to know about that first run:
- It needs internet once, to fetch the language pack - it doesn't contact Argos for every translation after that.
- The node prints a message telling you to restart ComfyUI so the freshly installed package gets picked up. Ignore it and you'll get a confusing import error.
After that first restart, everything is local and fast - a few hundred characters translates in well under a second, no GPU involved.
Inputs and output
The interface is gloriously short: input_text (multiline, paste whatever you need) in, translated_text (STRING) out. Wire the output into a Show Text node or straight into a positive-prompt input. Since both sides of a STRING can be force-connected, you can chain it after an LLM node that answers in Chinese and let the translation flow downstream into your English-only prompt slots.
It handles both simplified and traditional Chinese inputs, which is more useful than it sounds - the pack's other Chinese nodes (like the photo magazine and person-feature pipeline) routinely produce Traditional Chinese from LLMs, and this is the clean way to convert that before it hits a prompt parser.
Installing it
Bundled with ComfyUI-ListHelper:
cd ComfyUI/custom_nodes
git clone https://github.com/dseditor/ComfyUI-ListHelper
Then restart ComfyUI (the pack's base deps are just regex, accelerate, reportlab; Argos installs itself on demand). Find it under ListHelper/Chinese. ComfyUI Manager search: "ComfyUI-ListHelper".
Where people get burned
The restart-after-first-run step trips up almost everyone, so do that dance on purpose: run the node once, watch the console for the "installed, please restart" message, restart, done. Also temper your expectations on quality - Argos is a solid open-source translator, not a frontier LLM. Idioms and colloquial Chinese come out clunky, and the translation is Chinese-specific (zh→en is the only direction this node supports). For polishing a prompt, that's fine. For translating a novel's worth of dialogue for a video, you might still want an LLM - this is the zero-friction option, not the most eloquent one.
Inputs (1)
| Name | Type | Default | Description |
|---|---|---|---|
| input_text | STRING | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| translated_text | STRING | — |