Video to Caption Tool

A web-based tool to extract captions from videos with timestamps. Supports Hindi, English, and auto-language detection.

Features

• Language support: Hindi, English, and Hinglish
• Multiple output formats: SRT, VTT, JSON with timestamps
• Drag and drop: Easy video upload interface
• Flexible model selection: From tiny (fast) to large-v3 (most accurate)
• Web UI: Easy-to-use browser interface
• Local processing: No API costs, runs entirely on your machine

Screenshots

Upload a video → Configure settings → Generate captions → Download

Prerequisites

• Python 3.9 or higher
• FFmpeg installed on your system

Installing FFmpeg

macOS (Homebrew):

brew install ffmpeg

Ubuntu/Debian:

sudo apt update && sudo apt install ffmpeg

Windows: Download from https://ffmpeg.org/download.html and add to PATH.

Installation

1. Clone the repository:

git clone https://github.com/PrasanBora/video-caption-tool.git
cd video-caption-tool

2. Install dependencies:

pip install -r requirements.txt

3. Run the app:

streamlit run app.py

4. Open http://localhost:8501 in your browser

Usage

1. Upload a video file (MP4, MKV, AVI, MOV, WebM)
2. Configure settings in the sidebar:
   • Choose a Whisper model (Medium recommended)
   • Select language or use auto-detect
   • Pick output format (SRT, VTT, or JSON)
3. Click "Generate Captions"
4. Preview the generated captions
5. Download the caption file

Model Options

Model	RAM Required	Speed	Accuracy
Tiny	~1 GB	Fastest	Basic
Base	~1 GB	Fast	Good
Small	~2 GB	Medium	Better
Medium	~5 GB	Slower	High (Recommended)
Large	~10 GB	Slow	Very High
Large-v3	~10 GB	Slowest	Best

Output Formats

• SRT: Standard subtitle format, works with most video players
• VTT: WebVTT format, good for web applications
• JSON: Raw data format with timestamps for custom processing

Project Structure

video-caption-tool/
├── app.py                    # Main Streamlit web app
├── config.py                 # Configuration & settings
├── requirements.txt          # Python dependencies
├── core/
│   ├── video_handler.py      # Video validation & metadata
│   ├── audio_extractor.py    # FFmpeg audio extraction
│   ├── transcriber.py        # Whisper transcription
│   └── caption_formatter.py  # SRT/VTT/JSON output
├── models/
│   ├── transcript.py         # Transcript data models
│   └── caption.py            # Caption segment models
└── utils/
    └── time_utils.py         # Timestamp utilities

Tech Stack

• OpenAI Whisper: Speech-to-text transcription
• FFmpeg: Audio extraction from videos
• Streamlit: Web UI framework
• Pydantic: Data validation

Tips

• For Hindi/English videos, auto-detect works well
• Use Medium model for best balance of speed and accuracy
• GPU (CUDA or Apple Silicon) significantly speeds up transcription
• SRT format is most widely supported by video players

License

MIT License - feel free to use and modify.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.