Speaker Isolator API

This tool allows you to isolate speakers from a YouTube video using AssemblyAI's Speaker Diarization API.

Prerequisites

• Python 3.x
• AssemblyAI API key

Installation

1. Clone the repository:

   git clone https://github.com/AayushGupta16/speaker-isolator-api.git

2. Navigate to the project directory:

   cd speaker-isolator-api

3. Create a virtual environment:

   python3 -m venv venv

4. Activate the virtual environment:

   • For Windows:

     venv\Scripts\activate

   • For macOS and Linux:

     source venv/bin/activate

5. Install the required dependencies:

   python3 -m pip install -r requirements.txt

Configuration

1. Create a .env file in the project root directory.

2. Add your AssemblyAI API key to the .env file:

   ASSEMBLY_AI_API_KEY="YOUR_API_KEY"
   

   Replace "YOUR_API_KEY" with your actual AssemblyAI API key, including the double quotes.

Usage

1. Run the Flask application:

   python3 main.py

2. Make a POST request to the /process_video endpoint with the following JSON payload:

   {
       "youtube_url": "YOUR_YOUTUBE_VIDEO_URL"
   }

   You can use the following curl command to make the request:

   curl -X POST -H "Content-Type: application/json" -d '{"youtube_url": "YOUR_YOUTUBE_VIDEO_URL"}' -o speaker_segments.zip http://localhost:8000/process_video

   Replace YOUR_YOUTUBE_VIDEO_URL with the actual URL of the YouTube video you want to process.

3. The API will process the YouTube video, isolate speaker segments, and return a ZIP file named speaker_segments.zip containing the output audio files.

Error Handling

The tool includes error handling for the following scenarios:

• Invalid request payload
• Invalid YouTube URL
• Missing API key
• Errors during YouTube video download
• Errors during audio upload to AssemblyAI
• Errors during transcription process
• Errors during speaker segment creation

In case of an error, an appropriate HTTP status code and error description will be returned.

Logging

The tool uses Python's logging module for logging. Log messages are output to the console with timestamps and log levels.

License

This project is distributed under the MIT License. See the LICENSE file for more information.