this is a public repo the actual one i wrote is in private, it's all messy, that's why i had to create a new one all the features, i implemented in that will be implemented in these aswell, by will implement slowly, like once in a week or 2 weeks

all the contributions and suggestions are welcome,

happy coding :>

Smart Subtitle Generator

A Python tool that automatically adds subtitles to videos by transcribing speech to text and embedding the resulting subtitles directly into the video.

Features

• Extracts audio from input video
• Transcribes speech to text using Vosk speech recognition
• Generates subtitles with precise timing
• Burns subtitles into the video with clean formatting
• Preserves the original audio in the output
• Zero-delay subtitle display - words appear exactly when they're spoken
• Semi-transparent background for better subtitle readability
• Properly formats text with capitalization and punctuation
• Multi-line subtitle support with word wrapping

Requirements

• Python 3.6+
• Required Python packages:
  • ffmpeg-python
  • opencv-python (cv2)
  • vosk
  • numpy
  • tqdm
• FFmpeg installed on your system and available in PATH

Installation

1. Clone or download this repository:

   git clone https://github.com/humbledneuron/SubTalker.git
   cd SubTalker
   

2. Install the required packages:

   pip install ffmpeg-python opencv-python vosk numpy tqdm
   

3. Install FFmpeg:

   • Windows: Download from ffmpeg.org and add to your PATH
   • macOS: brew install ffmpeg
   • Linux: sudo apt install ffmpeg or equivalent for your distribution

4. Download a Vosk model:

   • Small models work well and are faster: https://alphacephei.com/vosk/models
   • Extract the model to a folder named "model" in the same directory as the script
   • Alternatively, the script will download a small English model if none is provided

Usage

Basic usage:

python subtitle_generator.py input_video.mp4

This will create a file called input_video_subtitled.mp4 with embedded subtitles.

Advanced Options

python subtitle_generator.py input_video.mp4 -o output_video.mp4 -m /path/to/vosk/model --keep-temp

Parameters:

• input: Path to the input video file
• -o, --output: Path to the output video file (default: input_subtitled.mp4)
• -m, --model: Path to Vosk model directory
• --keep-temp: Keep temporary files (useful for debugging)

How It Works

1. Audio Extraction: The program extracts the audio track from the video into a temporary WAV file.

2. Speech Recognition: Using Vosk, the program transcribes the speech in the audio to text with timestamps for each word.

3. Subtitle Generation: Words are grouped into subtitle segments with proper formatting and timing.

4. Video Processing: The program creates a new video with the subtitles rendered on each frame at the correct time.

5. Audio Merging: The original audio is merged back with the subtitled video to produce the final output.

Customization

You can modify the code to customize:

• Subtitle appearance (font, size, color, background opacity)
• Maximum characters per subtitle
• Subtitle positioning
• Text formatting rules

Troubleshooting

• FFmpeg errors: Ensure FFmpeg is properly installed and accessible in your PATH
• Video without audio: Check if the input video has an audio track
• Poor transcription quality: Try using a different/larger Vosk model
• Memory issues with large videos: Process the video in smaller segments

License

This software is provided under the MIT License. Feel free to modify and distribute as needed.

Acknowledgments

• Vosk for the speech recognition engine
• OpenCV for video processing
• FFmpeg for audio extraction and video merging