theonlyamos/ghana-languages-speech-to-speech-pipeline

A comprehensive pipeline for building multilingual Speech-to-Speech (S2S) AI systems for Ghanaian languages: **Akan (Twi/Fante)**, **Ewe**, **Ga**, and **Dagbani**. # Ghana Speech-to-Speech Pipeline A comprehensive pipeline for building multilingual Speech-to-Speech (S2S) AI systems for Ghanaian languages: **Akan (Twi/Fante)**, **Ewe**, **Ga**, and **Dagbani**. ## Architecture ``` +------------------+ +------------------+ +------------------+ +------------------+ | THE EAR'S | | THE EAR | | THE BRAIN | | THE MOUTH | | TUNER (LID) | --> | (ASR) | --> | (Translation) | --> | (TTS) | | MMS-LID | | Meta MMS | | NLLB-200 | | XTTS v2 | +------------------+ +------------------+ +------------------+ +------------------+ ``` ## Features - **Automatic Language Detection**: Auto-detect which Ghanaian language is being spoken - **Multi-language ASR**: Transcribe Akan, Ewe, Ga, Dagbani, and English - **Cross-language Translation**: Translate between any supported language pair - **High-quality TTS**: Natural speech synthesis with voice cloning support - **Full S2S Pipeline**: End-to-end speech translation - **Web Interface**: Gradio-based demo UI - **REST API**: FastAPI endpoints for production deployment ## Installation ### 1. Clone the repository ```bash git clone github.com cd ghana-languages-speech-to-speech-pipeline ``` ### 2. Create virtual environment ```bash python -m venv venv source venv/bin/activate # Linux/Mac # or venv\Scripts\activate # Windows ``` ### 3. Install PyTorch with CUDA ```bash pip install torch torchvision torchaudio --index-url download.pytorch.org ``` ### 4. Install dependencies ```bash pip install -r requirements.txt ``` ### 5. Install espeak-ng (for TTS) ```bash # Ubuntu/Debian sudo apt-get install espeak-ng # Windows: Download from github.com ``` ## Quick Start ### Using the Jupyter Notebook ```bash jupyter notebook ghana_s2s_pipeline …