PersianVox_v1_1k Dataset 📚🎤

Dataset Overview

PersianVox_v1_1k is a smaller version (1.8 hours of audio) of the much larger PersianVox_v1 dataset, which contains a total of ~450 hours of Persian speech data. This dataset is designed to support research in speech recognition, natural language processing, and other related fields. PersianVox_v1_1k is a preview of the full PersianVox_v1 dataset, providing a small sample of the larger collection.

Audio Samples 🎧

These are samples from the model trained on the full PersianVox_v1 during early stages of training:

Test 01:

Test 02:

Test 03:

Test 04:

Test 05:

Test 06:

Test 07:

Test 08:

Source

The audio data has been gathered from a variety of YouTube videos, ensuring diversity in topics, speech styles, and accents.

Dataset Details

Size: 1.8 hours of audio
Language: Persian (Farsi)
Content Source: YouTube Videos
Licensing: Licensed for research purposes only
Purpose: Preview of the larger PersianVox_v1 dataset

Direct Links

Alternatively, you can download or listen to the samples here:

Important Notes

This dataset is a preview and is made available for research purposes only.
Access to the full PersianVox_v1 dataset requires a purchase for research use.
The full dataset contains 380 hours of processed audio, and the processing and gathering work is valuable for researchers.
If you are interested in purchasing access to the full PersianVox_v1 dataset for research purposes, please contact me at:
✉️ saeedzou2012@gmail.com

Usage Guidelines

This dataset is intended solely for non-commercial, academic research purposes.
Please respect the licensing terms and use the dataset accordingly.

Acknowledgment

This work was conducted under the supervision and support of FAIM Lab, Sharif University of Technology.

Contact Information

For inquiries or to request access to the full PersianVox_v1 dataset for research (purchasing required), reach out to me at:
📧 saeedzou2012@gmail.com

Thank you for your interest! 🙌🎧