kevinkoech357 / transkript

A flask built web app that leverages the power of OpenAI's whisper model to transcribe audio and video files. Has support for various file formats. Generates timestamped .srt files.

Transkript

Transkript is a Flask-based web application designed to transcribe audio and video files, providing downloadable transcripts for users. It utilizes the Whisper Model from OPENAI for accurate transcriptions.

final_transkript.mp4

Table of Contents

• Features
• Getting Started
  • Prerequisites
  • Installation
  • Configuration
  • Running the Application
  • Usage
• Contributing
• License
• Acknowledgments

Features

• Audio Transcription: Transcribe audio files to text.
• Video Transcription: Transcribe video files to text.
• Downloadable Transcripts: Users can download the transcriptions as SRT files.
• User-Friendly Interface: An intuitive and easy-to-use web interface for file upload.

Getting Started

Follow these instructions to set up and run Transkript on your local machine for development or testing purposes.

Prerequisites

• Python (version 3.8+)
• ffmpeg ==> bash sudo apt update && sudo apt install ffmpeg -y

Installation

1. Clone the repository:

   git clone https://github.com/kevinkoech357/transkript.git

2. Navigate to the project directory:

   cd transkript
   mkdir uploads
   
   # Create .env file and add
   UPLOAD_FOLDER='/path/to/upload/folder'

3. Create a virtual environment:

   python3 -m venv venv

4. Activate the virtual environment:

   • On Windows:

   .\venv\Scripts\activate

   • On macOS and Linux:

   source venv/bin/activate

5. Install dependencies:

   pip install -r requirements.txt

Running the Application

1. Run the Flask backend:

   python run.py
   
   or 
   
   gunicorn -w 4 run:app

2. Visit http://localhost:5000 in your web browser.

Usage

1. Upload your audio or video file using the provided form.
2. Wait for the transcription process to complete.
3. Once completed, you can download the transcription as an SRT file.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

• OPENAI for providing the powerful Whisper model.