Speak is an innovative, open-source language learning application designed to help users practice and perfect their speaking skills in a new language. By leveraging the power of AI, Speak simulates realistic conversations, making language practice accessible and engaging.
- Real-time voice transcription using Whisper for accuracy in various languages.
- Conversation simulation and language corrections powered by Hugging Face and Ollama LLMs.
- Personalized feedback through voice cloning with the Bark model.
- Intuitive and accessible user interface built with Streamlit.
- Python
- Hugging Face Transformers
- Ollama for LLM
- Streamlit for frontend
- SQLite for database management
- Whisper for voice transcription
- Bark model for voice cloning
To get "Speak" running locally:
- Install Ollama: https://ollama.com/
- Run:
ollama run qwen:7b-chat - Clone the repository:
git clone https://github.com/5uru/Speak.git - Change directory:
cd Speak - Install required dependencies:
pip install -r requirements.txt - Run the Streamlit application:
streamlit run app.py
- Containerization with Docker for easier deployment and scalability.
- Expansion of language offerings and personalized learning experiences.
- Enhancing the community aspect by enabling user contributions and collaborative learning.