README.md

metadata

title: French Tutor
emoji: 🚀
colorFrom: blue
colorTo: red
sdk: gradio
sdk_version: 4.31.0
app_file: app.py
pinned: false
secrets:
  - MISTRAL_API_KEY
  - GEMINI_API_KEY
  - GROQ_API_KEY

🇫🇷 French Conversation Tutor

This is a Gradio app for practicing French conversation through natural speech interaction with Mr. Mistral!

How it works:

1. The app gives you a scenario and some helpful phrases.
2. You record your response in French using your microphone.
3. The AI tutor (Mr. Mistral) replies, and you have a short conversation (3 exchanges).
4. At the end, you get a detailed analysis of your grammar, pronunciation, and vocabulary.

Tech Stack:

• Primary LLM: Mistral AI (mistral-large-latest)
• Fallback LLM: Google Gemini API (gemini-1.5-flash-latest)
• STT: Groq API (Whisper large-v3-turbo)
• TTS: Groq API (TTS-1 model)
• UI: Gradio

The app will show which LLM is being used in the interface. It prioritizes Mistral AI, but falls back to Google Gemini if Mistral is unavailable.

Features:

• 🎤 Speech-to-text transcription of French audio
• 🔊 Text-to-speech for AI tutor responses
• 💬 Natural conversation flow
• 📊 Detailed analysis after 3 exchanges
• 🎲 Random topic generation for variety
• 🤖 Model indicator showing which AI is responding

Setup:

1. Create a .env file with:

   MISTRAL_API_KEY=your_mistral_api_key
   GEMINI_API_KEY=your_gemini_api_key  # Optional fallback
   GROQ_API_KEY=your_groq_api_key
   

   Important: Never commit your .env file to version control!

2. Install dependencies: pip install -r requirements.txt

3. Test your API keys: python test_api_keys.py

4. Run: python app.py

For HuggingFace Spaces:

• Add MISTRAL_API_KEY, GEMINI_API_KEY (optional), and GROQ_API_KEY in the Space's Settings > Variables and secrets