Logo Lanfrica

JessesCUkaegbu/Liberia-ASR-Microservice-Dialect-Language-Model-API

Domaine:

natural language processing

Type de record:

software
Créateur:
Jes
HĂ´te:
🎙️ ASR Microservice — Speech-to-Text API with Whisper A lightweight Automatic Speech Recognition (ASR) microservice built with Flask and Faster-Whisper. Users can upload audio files (.wav or .mp3) through a web interface or API endpoint and receive fast, accurate transcriptions in real time. # 🎙️ ASR Microservice — Speech-to-Text API with Whisper A lightweight Automatic Speech Recognition (ASR) microservice built with Flask and Faster-Whisper. Users can upload audio files (.wav or .mp3) through a web interface or API endpoint and receive fast, accurate transcriptions. --- ## 🚀 Features - Upload audio files via browser or REST API - Fast transcription using OpenAI Whisper models - CPU-friendly (runs without GPU) - Simple web UI for file upload - Easy public access using ngrok - JSON API responses --- ## 🏗️ Architecture Flask API Server → Whisper Model → Transcription JSON --- ## 📦 Tech Stack - Python 3.10+ - Flask - Faster-Whisper - Pyngrok (for public access) - FFmpeg / PyAV (audio decoding) --- ## 🔧 Installation ### 1. Create a virtual environment ### 2. pip install faster-whisper flask pyngrok soundfile ### 3. Generate the requirements.txt file Running the Server python app.py Expose Public API with ngrok from pyngrok import ngrok ngrok.set_auth_token("YOUR_NGROK_AUTH_TOKEN") public_url = ngrok.connect(5000) print("Public API URL:", public_url) ```bash python -m venv asr_env source asr_env/bin/activate