Umutoni-Rita/kinyarwanda-voice-assistant

This project is a Kinyarwanda voice assistant built for the Intelligent Robotics course assignment. # Kinyarwanda Voice Assistant A Kinyarwanda voice assistant built for the *Intelligent Robotics* course assignment. It uses the `benax-rw/KinyaWhisper` model for Automatic Speech Recognition (ASR), basic NLP for question-answer matching, and gTTS for Text-to-Speech (TTS). The `benax-rw/my_kinyarwanda_dataset` dataset is used for testing and validation. --- ## Features - **ASR:** Transcribes Kinyarwanda speech using `benax-rw/KinyaWhisper`. - **NLP:** Matches questions to answers using fuzzy string matching. - **TTS:** Converts answers to spoken Kinyarwanda using gTTS. - **GUI:** User-friendly Streamlit interface. - **Dataset Integration:** Tests ASR with `benax-rw/my_kinyarwanda_dataset`. --- ## Requirements - Python 3.8+ - Required libraries: `streamlit`, `openai-whisper`, `gtts`, `sounddevice`, `scipy`, `numpy`, `fuzzywuzzy`, `python-Levenshtein`, `datasets` Install all dependencies: ```bash pip install -r requirements.txt ``` --- ## Setup 1. **Clone the repository:** ```bash git clone github.com cd kinyarwanda-voice-assistant ``` 2. **Install dependencies:** ```bash pip install -r requirements.txt ``` 3. **Run the application:** ```bash streamlit run kinyarwanda_voice_assistant.py ``` --- ## Usage - Open the Streamlit app in your browser. - **Record Audio:** Click **"Record Audio"** to record a 5-second Kinyarwanda question. - **Test Dataset:** Select a sample from `benax-rw/my_kinyarwanda_dataset` to test ASR. - The assistant will: - Transcribe the audio - Match the transcription to the closest known question - Speak the corresponding answer - All audio files and transcriptions are saved in the `audio_files/` folder. --- ## Project Structure ```bash kinyarwanda-voice-assistant/ ├── kinyarwanda_voice_assistant.py # Main application code ├── requirements.txt # Required libraries ├── audio_files/ # Sample audio, dataset samples, transcriptions │ └── tr …