Transcription Web App

The project consists of a FastAPI server (server.app) and a frontend component with Python scripts (front-end) for various audio-related operations. It also utilizes AWS S3 for storing audio files and includes some external services such as VAD (Voice Activity Detection) and ASR (Automatic Speech Recognition).

Getting Started

Speech Models

Voice Activity Detection

1. Download release from https://github.com/snakers4/silero-vad/releases
2. Unzip it to backend/app/ml_models/vad
3. Copy files from files/ to vad/
4. Update utils.py from the VAD repo, if required.

ASR speech-to-text

Any ASR model can be instantiated by implementing the class ASR from backend/app/src/asr.py

Running locally

1. The project can be run locally using Localstack to simulate the creation of the AWS resources.

docker-compose up

2. Access http://localhost:8501/ to see the website.

License

This project is open-source and available under the MIT License. You are free to use, modify, and distribute it as per the terms of the license.