Why Build DeepLX#
There are many translation software options available on the market today. Traditional translation software includes Netease Youdao Translation, ByteDance Volcano Translation, Google Translation, Microsoft Bing Translation, and more. Translation based on the LLM large model can call Gemini, OpenAI, and other APIs for translation. So why choose DeepL?
-
Fast and accurate translation (compared to traditional translators)
When I use Youdao and other translation tools, the most common problem I encounter is low translation accuracy (especially when translating from non-English to Chinese), and there are usually omissions in the translation (such as when encountering more colloquial expressions).
As the first translator to use AI for translation model training, DeepL is naturally more accurate compared to general translators.
-
Fast translation speed, and not limited by concurrency (compared to LLM large model translation)
While translation based on Gemini, OpenAI, and other technologies is accurate, it is limited by the number of API calls and cannot complete translations in high-concurrency environments (such as using immersive translation plugins to automatically translate Twitter). In comparison, DeepL can maintain a certain level of translation accuracy while having translation speed comparable to traditional translators.
However, the official DeepL itself is slow and has limitations on usage, such as limited quotas and no API access, which can discourage many users. DeepLX is open source on GitHub and does not limit the number of requests (although DeepL may limit IP addresses). By default, it listens on port 1188 locally and provides multiple installation methods.
How to Build DeepLX#
Thanks to the original author OwO-Network for their hard work.
One-Click Installation#
bash <(curl -Ls https://raw.githubusercontent.com/OwO-Network/DeepLX/main/install.sh)
Or
bash <(curl -Ls https://owo.nz/deeplx)
Open Port 1188#
This step should be completed in the backend of your VPS service provider. The method of opening ports may vary for different service providers, so please search for the specific method.
Testing#
Visit ip:1188 to perform a test. If it returns code: 200 ok, it means the setup was successful.
Using in Translation Plugins (Using Immersive Translation as an Example)#
-
In the settings of the immersive translation plugin, click on Developer Settings and enable Beta Testing Features.
-
Then go back to the basic settings and select DeepLX (Beta) as the translation service. Fill in the URL with the URL of your own deployed service, which is
[ip]:1188/translate.
