Smart Audio Transcription and Editing Tool

Introduction

This project combines audio transcription and smart editing into a convenient end-to-end user tool. The smart editing feature parses the transcription, adds appropriate paragraph breaks and corrects grammatical / sentence errors. Users are able to manually edit the text to their liking at any stage of processing. It uses the following models:

• Transcription: OpenAI's Whisper model
• Post-transcription smart editing: Microsoft's Phi-3 Mini 128K Instruct OpenRouter.AI API

Installation

1. Clone the repository.
2. Install python3 and pip.
3. Inside the project root folder, run python -m venv venv to create a virtual environment.
4. Activate the virtual environment: run source ./venv/bin/activate.
5. Inside the project root folder, run pip install -r requirements.txt to install the necessary packages.
6. To download the Whisper model, run pip install --upgrade --no-deps --force-reinstall git+https://github.com/openai/whisper.git. Follow the instructions on https://github.com/openai/whisper for more information.

Running the web app

1. Run python -m flask --app tool run --port 8000
2. Open the web app on localhost:8000 in your Chrome browser.

Technologies

• Flask
• Python
• HTML
• JavaScript
• CSS