🎬 Paheli–Boojho AI Video Generator

An AI-powered Streamlit app that generates short two-character dialogue reels — complete with TTS audio, speaker diarization, character expression overlays, AI image inserts, and auto-captions.

---

✨ Features

• Script generation via Gemini 2.5 Flash (customisable tone, humor level, topic)
• Multi-speaker TTS using Gemini's preview TTS model
• Speaker diarization via AssemblyAI (maps audio segments back to each character)
• Expression detection — Gemini picks the right character face per dialogue line
• AI image overlays — keyword-matched images generated and timed to the script
• Auto-captioning via PyCaps
• ffmpeg-based video renderer with background video support

---

🚀 Getting Started

1. Clone the repository

git clone https://github.com/your-username/your-repo-name.git
cd your-repo-name

2. Install dependencies

pip install -r requirements.txt

Note: pyannote.audio requires a Hugging Face token and model access. See pyannote docs for setup.
ffmpeg must also be installed on your system: https://ffmpeg.org/download.html

3. Configure API keys

Copy the example env file and fill in your keys:

cp .env.example .env

Then edit .env:

GEMINI_API_KEY=your_gemini_key_here
ASSEMBLY_AI_API_KEY=your_assemblyai_key_here
ZAPCAP_API_KEY=your_zapcap_key_here
ZAPCAP_TEMPLATE_ID=your_template_id_here

You can also enter keys directly in the 🔑 API Keys tab inside the app — manual entries override the .env values.

4. Run the app

streamlit run main.py

---

🔑 Where to get API keys

Service	Link
Gemini	https://aistudio.google.com/api-keys
AssemblyAI	https://www.assemblyai.com/dashboard/api-keys
Zapcap	https://platform.zapcap.ai/dashboard/api-key
Zapcap Templates	https://platform.zapcap.ai/dashboard/templates

---

📁 Project Structure

├── main.py                  # Main Streamlit app
├── character_data.py        # Character image path mappings
├── assets/
│   ├── backgrounds/         # Background video files
│   ├── characters/          # Character PNG overlays
│   └── temp/                # Temp storage for uploaded backgrounds
├── pycaps_templates_own/    # PyCaps caption templates
├── .env.example             # Template for environment variables
├── requirements.txt
└── README.md

---

⚠️ Notes

• Generated audio, images, and video files are listed in .gitignore and will not be committed.
• Never commit your .env file.
• The assets/ folder (character images, backgrounds) is not included in this repo — add your own or contact the author.