dailySerbian

@dailySerbianBot

I am a bot-translator from Russian to Serbian and from Serbian to Russian.

Here's what you should know about me:

I can translate all sorts of things: text, documents, voice messages. I also send voice messages with the translation.

Just send me something in Russian or in Serbian, and I'll figure out what to do with it.

I can also create a separate dictionary for you and occasionally spam words from there so you can practice your vocabulary. I adapt the dictionary to your knowledge and often send words that you have translated recently or made mistakes with during training.

ML part is implemented using open libraries:

• I have some pydantic and json-schema inplementation

• Translation text -> text:
  • deep_translator.GoogleTranslator(source="ru", target="sr")
• Recognizer speech -> text:
  • speech_recognition.Recognizer()
• Text -> speech:
  • gtts.gTTS(text, lang="sr", slow=False)
• Image -> text:
  • easyocr.Reader(lang_list=["en", "ru"])
  • To deploy such a model on a free deployment website, it is important to keep everything as small as possible. Do not upload cache or virtual environment files to the deployment, and make sure you only use parts of packages that you need. The Pytorch package made my app exceed the allowed ‘slug’ limit of 500MB on Heroku. By only downloading the CPU version, we saved roughly 300–400MB of space.
  • Update requirements.txt file:
    • add to requirements.txt file -f https://download.pytorch.org/whl/torch_stable.html
    • run in console pip install torch== -f https://download.pytorch.org/whl/torch_stable.html
    • find in the output error available +cpu torch version
    • add it to requirements.txt file: (example: torch==1.11.0+cpu)
• Sourse language detection:
  • deep_translator.single_detection()

I have some CI/CD configured:

• Lint with pre-commit-hooks, flake8, blake
• Test with pytest + Generate Coverage Report
• Build, Test and Deploy to Heroku hosting with Github Actions
• Deploy with Docker Compose

TODO - what I'm not yet capable of product-wise

• Word self-training (Russian->Serbian) --> in progress
• Setting self-training time and word spam time

TODO - what I'm not yet capable of technically

• Postgre database
• Test coverage

TODO - open ai_tools, it would be interesting to try

• Multiple translation options as output
• TEXT 2 SPEECH
  • https://faun.pub/python-text-to-speech-tutorial-b0031f6664a0
  • https://www.codingem.com/best-ai-voice-generators/
• Interact with ChatGPT
  • https://medium.com/@andrea.faviait/deploying-a-telegram-bot-using-chatgpt-and-whisper-apis-with-railway-ef79e6cff955