AutoSub is a command-line interface (CLI) tool designed to automatically generate subtitle files (SRT, VTT, TXT) for any video. It targets users who download video content and require subtitles, offering a convenient solution for creating them using open-source speech-to-text engines.

How It Works

AutoSub leverages Mozilla DeepSpeech or Coqui STT for speech-to-text inference. It first uses FFmpeg to extract audio from the video, ensuring a 16kHz sampling rate compatible with DeepSpeech. Then, it employs pyAudioAnalysis to segment the audio based on silence, creating smaller, manageable files. Inference is performed on each audio segment, and the resulting text is compiled into subtitle files.

Quick Start & Requirements

• Install via pip: pip install . (after cloning the repo and optionally activating a virtual environment).
• Prerequisites: Python 3, FFmpeg. For GPU support, install requirements-gpu.txt and ensure appropriate CUDA version.
• Model files are downloaded automatically if not present.
• See: DeepSpeech Examples, Coqui STT

Highlighted Details

• Supports both DeepSpeech and Coqui STT engines.
• Audio segmentation via silence detection for efficient processing.
• Customizable subtitle split duration and output formats (SRT, VTT, TXT).
• Docker support for CPU and GPU builds.

Maintenance & Community

• Project is maintained by abhirooptalasila.
• References include DeepSpeech and pyAudioAnalysis repositories.

Licensing & Compatibility

• The README does not explicitly state a license. Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

• Requires manual download of model and scorer files for languages other than English.
• Performance on a dual-core i5 with 8GB RAM for a 70-minute video was approximately 40 minutes, suggesting potentially long processing times for longer videos on less powerful hardware.