This project provides an enhanced fork of MoneyPrinterTurbo, focusing on automated YouTube video generation with advanced subtitle and Text-to-Speech (TTS) capabilities. It targets content creators seeking to produce engaging, science-backed videos efficiently, offering a free, locally-run TTS solution with voice cloning and precise subtitle synchronization as key benefits.

How It Works

This fork integrates a sophisticated subtitle system featuring word-by-word highlighting that synchronizes precisely with spoken audio, enhancing viewer engagement. Video selection is improved through semantic search and text similarity analysis, moving beyond random clip generation. A core differentiator is the inclusion of Chatterbox TTS, a free, open-source, locally-run TTS engine capable of voice cloning from short audio samples, eliminating API costs and rate limits. The system leverages a detailed LLM prompt to generate video titles, scripts, visual search keywords, and descriptions, streamlining the content creation pipeline.

Quick Start & Requirements

Installation involves cloning the repository, setting up a Conda environment using environment.yml, and activating it. Chatterbox TTS requires a separate clone and installation (pip install -e .). CUDA 12 and cuDNN 8.x libraries are recommended for GPU acceleration, with setup scripts (setup_cuda_env.sh) provided. The web interface is launched via ./webui.sh, accessible at http://localhost:8501. Speech speed can be adjusted using the CHATTERBOX_CFG_WEIGHT environment variable.

Highlighted Details

• Word-by-word subtitle highlighting with real-time synchronization.
• Open-source Chatterbox TTS with voice cloning (10-60s reference audio) and no API costs.
• Semantic search and text similarity for improved video clip relevance.
• Optional thumbnail analysis for source selection.
• Automatic TTS speed control via environment variables.

Maintenance & Community

The project encourages contributions via GitHub stars and issue reporting. It acknowledges the original project's authors and contributors, maintaining compatibility with the base MoneyPrinterTurbo repository. Specific community channels (e.g., Discord, Slack) or active maintainer details are not detailed in this README.

Licensing & Compatibility

The README does not explicitly state a license for this fork. It notes full credit to the original author and maintains compatibility with the original MoneyPrinterTurbo, suggesting its licensing may be inherited. Compatibility for commercial use or closed-source linking is not specified.

Limitations & Caveats

Troubleshooting indicates potential issues with garbled TTS audio (mitigated by preprocessing), CUDA errors (with CPU fallback), and voice cloning requiring clear, single-speaker audio. Compatibility issues with newer MoviePy versions regarding the align parameter may require code adjustments. A dependency on cuDNN 8.x is noted.