This project provides an open-source, self-hosted solution for generating subtitles from video content using AI. It's designed for users who need control over their data and want to integrate subtitle generation into their workflows, offering multilingual support and ease of use.

How It Works

The tool leverages the OpenAI Whisper model for speech-to-text transcription. Users can provide a video file path or URL, and the script processes the audio to generate subtitles. It supports various Whisper model sizes and quantization levels for a trade-off between accuracy and performance, and offers extensive command-line options for fine-tuning transcription parameters like threads, language detection, and output formats.

Quick Start & Requirements

• Install FFmpeg: sudo apt install ffmpeg (Linux example).
• Run script: python subtitle.py <filepath | video_url> [--model <modelname>]
• Models: Supports various Whisper models (e.g., tiny.en, base, medium, large-v2).
• See README for detailed command-line options and binary usage examples.

Highlighted Details

• AI-powered subtitle generation using OpenAI Whisper.
• Self-hosted for enhanced privacy and control.
• Multilingual support.
• Extensive customization via command-line arguments for transcription.
• Supports multiple output formats including .txt, .vtt, .srt, .lrc, and .csv.

Maintenance & Community

• Project initiated by Ved Gupta.
• Credits OpenAI Whisper and ggerganov/whisper.cpp.
• Support via email: vedgupta@protonmail.com.

Licensing & Compatibility

• MIT License.
• Permissive license suitable for commercial use and integration into closed-source projects.

Limitations & Caveats

The project is described as being made "for fun," and while useful, it may not have the robust testing or long-term support of a commercially backed project. Specific hardware requirements for running larger models are not detailed.