Skip to content

Dataset · prathoshap

Classical Sanskrit recitation corpus

prathoshap/vagdhenu-data

Around 5.3 hours of Sanskrit shloka recitation audio by a single speaker for TTS training.

What's inside

It contains approximately 1,467 audio clips (24 kHz, mono) and corresponding Sanskrit texts. The recordings were made by the dataset author, adhering to traditional prosody and metric duration of classical shlokas (without Vedic swaras). Two configurations are available in the repository, obtained from two passes of chunking and normalization.

What you can do with it

  • Training and fine-tuning text-to-speech (TTS) models for Sanskrit
  • Modeling rhythmic and metric prosody in speech synthesis
  • Training automatic speech recognition (ASR) systems on classical texts
Tasks
Text → speech
Languages
Sanskrit
Size
1K–10K rows
Downloads this month
952
Likes
34

How to load it

from datasets import load_dataset

ds = load_dataset("prathoshap/vagdhenu-data")