Project README

Overview

This project involves processing and extracting data from various Sanskrit texts and audios and compile it as SwaraSangraha, A Sanskrit Chanting-style Speech Dataset.

SwaraSangraha (स्वरसंग्रह)

1. Amarakosha (अमरकोषः)
2. Ashtadhyayi (अष्टाध्यायी)
3. Meghaduta (मेघदूतम्)
4. Valmiki Ramayana (वाल्मीकि रामायण)
5. TarkaSangraga (तर्कसंग्रह)
6. Patanjali Yoga Sutrani (पातञ्जलयोगसूत्राणि)

Data Collection & Processing

It includes modules for:

• Web Scraping of Sanskrit texts
• Computing Total Duration of Audio Files
• Demucs-based Speech Separation

Directory Structure

📁 code
  📁 processing
    📄 demucs.py         # Demucs-based audio separation
    📄 duration.py       # Computes duration of audio files
  📁 scraping
    📄 amarakosha.py     # Scrapes Amarakosha text & audio
    📄 ashtadhyayi.py    # Scrapes Ashtadhyayi text & audio
    📄 meghaduta.py      # Scrapes Meghaduta text & audio
    📄 ramayana.py       # Scrapes Ramayana text & audio
    📄 tarkasangraha.py  # Scrapes Tarkasangraha text & audio
    📄 yogasutra.py      # Scrapes Yogasutra text & audio
  📁 test               # Directory for test files

📁 demucs               # Output directory for processed audio
📁 demucs_temp          # Temporary files during Demucs processing
📁 SwaraSangraha        # Collection of scraped Sanskrit audio/text
📁 separated_audio      # Storage for separated audio components

Installation

Dependencies

Ensure you have the required dependencies installed:

pip install numpy pandas librosa mutagen tqdm beautifulsoup4 requests pydub

Running the Scripts

1. Scrape Sanskrit Text and Audio

python code/scraping/amarakosha.py
python code/scraping/ashtadhyayi.py
python code/scraping/meghaduta.py
python code/scraping/ramayana.py
python code/scraping/tarkasangraha.py
python code/scraping/yogasutra.py

2. Compute Total Duration of Audio Files

python code/processing/duration.py

3. Run Demucs for Speech Separation

python code/processing/demucs.py

Notes

• Ensure you have access to the internet while running the scraping scripts.
• Errors and warnings will be logged in error_log.txt.

---