Datasets:

abhirl
/

hindi-tts-dataset

Tasks:

Modalities:

Formats:

Languages:

Size:

Tags:

Libraries:

License:

Dataset card Data Studio Files Files and versions

You need to agree to share your contact information to access this dataset

This repository is publicly accessible, but you have to accept the conditions to access its files and content.

Log in or Sign Up to review the conditions and access this dataset content.

Hindi Wikipedia TTS Dataset

A collection of 1,825 audio clips (~12.6 hours) generated from Hindi Wikipedia using Sarvam AI's bulbul:v3 TTS model.

📊 Quick Stats

Samples: 1,825 WAV files (22,050 Hz)
Speakers: 4 (shubh, ritu, pooja, amit)
Source: Hindi Wikipedia paragraphs (30-100 words)

📂 Features

audio: The audio waveform
text: Hindi transcript
speaker: Voice name
file_name: Original filename

⚖️ License & Attribution

License: CC-BY-NC-SA 4.0
Audio: Generated via Sarvam AI API
Text: Sourced from Wikipedia

🚀 Quick Start

from datasets import load_dataset
ds = load_dataset("abhirl/hindi-tts-dataset")
print(ds["train"][0])

Downloads last month: 23

Total file size:

Models trained or fine-tuned on abhirl/hindi-tts-dataset

Solo448/SpeechT5-Unified-TTS-PEFT

Text-to-Speech • Updated Apr 15 • 18

Solo448/SpeechT5-Unified-STT-PEFT

Text-to-Speech • Updated Apr 15 • 1