The LanguageDecoder translates a given text from the source language to a desired target language word by word (decoding). For this purpose, the Google translator is used in the backend (Or alternatively an experimental LLM translator). To improve the translation, a dictionary can be created to automatically correct common translation errors. Afterwards, the decoded text can be exported to a PDF file.
The LanguageDecoder can also be used to create language learning lessons based on the
Birkenbihl approach.
To use the LanguageDecoder you can either clone the repo with git:
git clone https://github.com/PumucklRandom/language-decoder.git
cd language-decoder
pip install -r requirements.txt
python ./__main__.py
or just download the latest desktop application from the releases.
For Windows use LanguageDecoder.zip, extract the files and execute LanguageDecoder.exe.
For Linux use LanguageDecoder-lx.zip, extract the files and execute LanguageDecoder.desktop.
With the build.py file you can create your on local desktop application.
The LanguageDecoder requires a common modern browser such as Edge, Chrome or Firefox and a free
port on localhost port 8080 (127.0.0.1:8080). If this port is occupied, the IP address and the port
can be adjusted as desired in the config file. (language-decoder/backend/config/config.yml)
For the desktop application: (LanguageDecoder/_internal/backend/config/config.yml)
Windows Security may block desktop application. (Workaround: Run application as admin)
Nicegui for the grafical user interface
OpenRouter for the access of Large Language Models
Deep Translator for the language translation
FPDF2 for the pdf creation
TextStudio for the application icon