Auto-Create-Video  by hoquanghai

Article-to-video automation pipeline

Created 3 weeks ago

New!

253 stars

Top 99.3% on SourcePulse

GitHubView on GitHub
Project Summary

Summary

This project addresses the time-consuming process of creating short-form news videos from articles. It offers an automated pipeline for generating TikTok-ready 9:16 videos in minutes, targeting content creators and social media managers seeking studio-quality output with zero editing.

How It Works

The pipeline integrates AI-driven script generation via Claude Code with a deterministic Node.js/TypeScript rendering engine. Articles are parsed, and Claude Code crafts scripts, selects from 12 smart templates, and orchestrates scene content. Text-to-Speech (TTS) is handled by LucyLab (Vietnamese cloning) or ElevenLabs (multilingual). The HyperFrames engine, leveraging Puppeteer and GSAP, renders animated HTML compositions. FFmpeg encodes the final MP4. Novelty lies in precise voice-text synchronization using voiceChunks for beat-accurate animations and pre-render quality gates (linting, WCAG validation, overflow inspection).

Quick Start & Requirements

  • Install: Clone the repository and run npm install.
  • Prerequisites: Node.js (≥ 22), FFmpeg (in PATH), Chrome/Chromium (auto-downloaded), Claude Code CLI, and an API key for either LucyLab or ElevenLabs.
  • Setup Time: Minimal; video generation takes ~3-5 minutes.
  • Links: Watch Demo, Full Setup Guide.

Highlighted Details

  • AI Scripting: Claude Code skill (/create-news-video <url>) automates scriptwriting and template selection.
  • TTS Options: LucyLab for natural Vietnamese voice cloning with free SRT subtitles, or ElevenLabs for 30+ languages.
  • Visuals: 12 smart templates, HeyGen-quality shell, GSAP animations, 6 theme palettes, and auto-generated 9:16 thumbnails via Gemini 2.5 Flash.
  • Audio: Auto SFX mixing with a 3-tier picker, anti-repetition, and anti-overlap guards.
  • Synchronization: voiceChunks enable beat-accurate visual element timing synchronized with TTS.
  • Quality Assurance: Pre-render checks include linting, WCAG contrast validation, and text overflow inspection.
  • Output: 1080x1920 @ 30fps MP4, with CapCut-friendly exports (script.txt, voice.mp3).

Maintenance & Community

The project is maintained by Ho Quang Hai. No specific community channels (Discord, Slack) or sponsorship details are listed in the README.

Licensing & Compatibility

  • License: MIT.
  • Compatibility: Permissive for commercial use and integration into closed-source projects.

Limitations & Caveats

The Claude Code skill's script generation is optimized for Vietnamese; other languages may require prompt adjustments. Video length is heuristically capped between 45-180 seconds. TTS API usage incurs minor costs. FFmpeg must be installed and accessible in the system's PATH.

Health Check
Last Commit

3 weeks ago

Responsiveness

Inactive

Pull Requests (30d)
4
Issues (30d)
2
Star History
254 stars in the last 24 days

Explore Similar Projects

Starred by Chip Huyen Chip Huyen(Author of "AI Engineering", "Designing Machine Learning Systems") and Jiaming Song Jiaming Song(Chief Scientist at Luma AI).

MoneyPrinterTurbo by harry0703

0.6%
58k
AI tool for one-click short video generation from text prompts
Created 2 years ago
Updated 2 days ago
Feedback? Help us improve.