AI Content Studio Next-Gen (v2.0) transforms video generation from a rigid script-to-video pipeline into an autonomous creative production system. Equipped with a supervisory multi-agent director, multimodal viral clipping, computer-vision style cloning, broadcast loudness mastering, and a nonlinear transcript ripple-editor, it handles your entire creative production workflow from research to export.

If you love this project and want to support its ongoing development, please consider helping out on Patreon or follow the journey on Instagram!

• Supervisory Agent Team: Orchestrates research, scriptwriting, quality assurance, and persistent memory across specialized agents (StrategyAgent, TrendDiscoveryAgent, ResearchAgent, CreativeDirectorAgent, QualityAssuranceAgent, LearningAgent).
• Self-Healing QA Retry Loop: Automatically inspects draft scripts against factual accuracy, hook retention, and narrative pacing. If the QA score is below 80/100, the director autonomously critiques and rewrites weak scenes before generating expensive media.
• Interactive Storyboard Scene Cards: Review and edit scene-by-scene script narrations, visual prompts, framing badges, and music/SFX cues.
• Scene Locking & Selective Regeneration: Lock approved scenes to keep them intact, and selectively regenerate individual scenes with custom user instructions (e.g., "Make this shot a drone wide establishing shot with more suspense").

Convert any long-form YouTube video, podcast, or stream into high-retention vertical shorts (TikTok, Reels, YouTube Shorts).

• Multimodal Hotspot Fusion: Combines Whisper transcriptions with audio RMS volume peaks (detecting applause, laughter, volume spikes, punchlines) and OpenCV HSV histogram cuts to identify true audio-visual moments of viral intensity.
• Active Speaker Tracking with Cut Hysteresis: Face detection with intelligent shot hysteresis (minimum 1.5s interval) prevents camera jitter on minor head turns.
• B-Roll Auto-Injection: Automatically fetches contextual stock B-roll to cover talking-head pauses and amplify viewer retention.
• Dynamic Captions: Renders high-accuracy karaoke subtitles directly into the video with custom fonts, colors, and word-grouping.

• Clone Editing Styles via Computer Vision: Upload reference footage from creators like MrBeast, Alex Hormozi, or Vox. The engine calculates average shot durations, cut frequencies, and caption animations into reusable style profiles without infringing on copyrighted footage.
• Live Brand Kit Studio: Real-time 9:16 short preview with custom dual-accent color palettes, typography selection, watermark placement/opacity, and stinger intros.

• Multi-Track Timeline: Full multi-layered canvas supporting Video clips, B-Roll overlays, Audio/SFX tracks, and Subtitle events.
• Text-Based Ripple Editing: Delete words or sentences in the transcript, and the underlying video and audio automatically ripple-cut to match.
• AI Timeline Assistant: Natural language timeline adjustments (e.g., "Speed up pacing", "Mute section").

• Broadcast Loudness Normalization: Integrated FFmpeg EBU R128 loudnorm=I=-14:TP=-1.5:LRA=11 audio mastering across all video renders, matching YouTube and TikTok loudness standards.
• Urdu & Roman Urdu Phonetic Engine: Comprehensive Urdu Unicode character normalization and Roman Urdu phonetic dictionary (e.g., kese -> kaisay, kyun -> kyoon, mujhe -> mujhey, dunya -> duniya, acha -> ach-cha) for natural, glitch-free neural TTS pronunciation.

• Durable Task Queue: SQLite-backed job engine with stage-by-stage checkpoints (research, storyboard, qa_review) and automatic recovery of interrupted jobs across restarts.
• Multi-Model Router: Unified provider gateway supporting Google Gemini, OpenAI-compatible endpoints, and local Ollama with automatic failover and error isolation.
• Zero Mock Exploits: True OAuth token management, SQLite credential storage, and manual publishing packages.

• Automated Campaigns: Schedule 24/7 background bots to scan trending topics and generate content on automated schedules.
• Faceless Explainers: Script, narrate, and compile multi-minute educational and storytelling videos from movie files or YouTube URLs.

• Standalone Auto-Captioner: Burn animated captions into pre-existing MP4 videos.
• Aspect Ratio Converter: Seamlessly crop and scale videos between 9:16, 16:9, and 1:1.
• Studio TTS Generator: Instant voiceover generator with character voice presets.

🔥 NEW FULL DETAILED VIDEO TUTORIAL COMING SOON! 🔥
A complete step-by-step video for this Next-Gen v2.0 version is currently in production. Make sure to Subscribe to our YouTube Channel (@naqashai) so you don't miss it when it drops!

Old Version References:

• If you are looking for the original codebase, you can find the old version safely stored in the Git history.
• 👉 Watch the Old Version Video Tutorial

Built with a modern, high-performance tech stack designed for speed and scalability:

• Framework: Next.js (React) with Turbopack
• Styling: Vanilla CSS design system + Glassmorphism UI
• Icons: Lucide React
• Real-time Engine: WebSockets for live agent logs, progress updates, and timeline syncing

• Server: FastAPI (Python 3.10+) & Uvicorn
• Database / Queue: SQLite + SQLAlchemy ORM with stage checkpoints
• Video Processing: FFmpeg (multimodal slicing, cropping, loudnorm mastering, caption burning)
• Downloading: yt-dlp (bypasses YouTube blocks and downloads source media)
• Transcription: OpenAI Whisper (offline, word-level audio-to-text transcription)
• Subtitles: pysubs2 (generates and styles .ass subtitle files)
• AI Brain: Multi-Model Gateway (Google Gemini, OpenAI-compatible APIs, local Ollama)
• Computer Vision: OpenCV (cv2) for active speaker tracking and histogram-based scene cut analysis

Before running the application, ensure you have the following installed on your system:

1. Python 3.10+ (Make sure to check "Add Python to PATH" during installation)
2. Node.js (v18+) & npm
3. FFmpeg (Must be installed and added to your system's PATH variable)
   • Windows via winget: winget install Gyan.FFmpeg
   • macOS via homebrew: brew install ffmpeg
   • Ubuntu/Debian: sudo apt update && sudo apt install -y ffmpeg
4. A Google Gemini API Key (Set via the UI Settings tab)

1. Clone the repository:

   git clone https://github.com/naqashafzal/AI-Content-Studio.git
   cd AI-Content-Studio

2. Setup the Python Backend:

   # Create a virtual environment
   python -m venv .venv
   
   # Activate it (Windows)
   .venv\Scripts\activate
   # Activate it (Mac/Linux)
   source .venv/bin/activate
   
   # Install backend dependencies
   pip install -r requirements.txt

3. Setup the Next.js Frontend:

   cd web
   npm install
   cd ..

Simply run or double-click start.bat:

start.bat

The automated launcher will verify Python 3.10+, Node.js 18+, and FFmpeg, initialize storage folders, activate .venv, check dependencies, launch both server consoles, and automatically open your browser to http://localhost:3000!

You can run both servers manually in separate terminal windows:

1. Start the Backend API (from the root directory):

# Ensure your virtual environment is active
python server/main.py
# The FastAPI server will run on http://localhost:8008

2. Start the Frontend UI (from the /web directory):

cd web
npm run dev
# The Next.js app will run on http://localhost:3000

Open your browser and navigate to http://localhost:3000 to start creating!

The repository includes a comprehensive automated test suite covering job queues, media processing, AI orchestrator loops, secret masking, and viral topic discovery:

# Run the full test suite
pytest tests/
# 19 passed

Building and maintaining this open-source project takes a lot of time and effort! If this tool has helped you grow your channels, automate your workflows, or make money, please consider supporting me:

• 💖 Support on Patreon: patreon.com/c/naqashafzal
• 📸 Follow on Instagram: @naqashafzal
• 📺 Subscribe on YouTube: @naqashai

Your support helps me keep updating the tool, adding new AI features, and keeping the core open-source for everyone!

• Clipper Audio is out of sync: Ensure you are using the latest backend code. We force -c:a aac during FFmpeg extraction to prevent audio PTS drift.
• 0.00s Clips generated: This means the AI couldn't parse the timestamps. Ensure pipeline_shorts.py formats the transcript chunks with [start - end] brackets.
• FFmpeg errors / "Subtitles not found": Ensure FFmpeg is installed globally on your machine and that ffmpeg is a recognized command in your terminal.
• Secret Keys Showing as sk-***: This is an intentional security hardening feature. When updating other configuration fields, your saved keys are securely retained without being overwritten by masked values.

This project is licensed under the MIT License.