Dataset Viewer
Auto-converted to Parquet Duplicate
identifier
stringlengths
9
15
gender
stringclasses
2 values
name
stringlengths
9
15
age
int64
16
80
audio
audioduration (s)
21
330
language
stringclasses
1 value
language_code
stringclasses
1 value
url
stringlengths
43
43
start_time
stringlengths
4
7
end_time
stringlengths
4
7
donald_trump
M
Donald Trump
79
english
e
https://www.youtube.com/watch?v=wQPTUa8vxRU
14:52
15:14
joe_biden
M
Joe Biden
80
english
e
https://www.youtube.com/watch?v=u1UC89H4Swc
2:11
2:32
barack_obama
M
Barack Obama
59
english
e
https://www.youtube.com/watch?v=mAFv55o47ok
4:10
4:54
andrew_tate
M
Andrew Tate
36
english
e
https://www.youtube.com/watch?v=VGWGcESPltM
5:09
5:50
jensen_huang
M
Jensen Huang
62
english
e
https://www.youtube.com/watch?v=3hptKYix4X8
54:14
55:59
elon_musk
M
Elon Musk
53
english
e
https://www.youtube.com/watch?v=7qZl_5xHoBw
27:45
29:51
mark_zuckerberg
M
Mark Zuckerberg
40
english
e
https://www.youtube.com/watch?v=7k1ehaE0bdU
2:21:30
2:27:00
j_k_rowling
F
J.K. Rowling
59
english
e
https://www.youtube.com/watch?v=3uLa3WKXwLg
0:14
1:38
oprah_winfrey
F
Oprah Winfrey
70
english
e
https://www.youtube.com/watch?v=1sq_GJiGckM
6:38
8:06
kamala_harris
F
Kamala Harris
60
english
e
https://www.youtube.com/watch?v=TJys7OVH24E
8:44
9:23
bill_gates
M
Bill Gates
70
english
e
https://www.youtube.com/watch?v=pBRSZBtirAk
10:14
10:48
hillary_clinton
F
Hillary Clinton
78
english
e
https://www.youtube.com/watch?v=4xCxWOnK33Q
8:03
8:50
greta_thunberg
F
Greta Thunberg
16
english
e
https://www.youtube.com/watch?v=BQ4rBLCpEeM
0:43
1:27

Voice Cloning Dataset

This repository provides a curated dataset of voice samples designed for Text-to-Speech (TTS) voice cloning applications. The primary goal is to offer high-quality audio clips suitable for training and fine-tuning voice cloning models.

Dataset Structure

This repository contains:

  • audio/: A folder containing the raw .wav audio files.
  • metadata.csv: A CSV file with metadata for each audio file.

Metadata

The metadata.csv file provides essential information about each voice sample.

For voices-celebrities, the columns are:

  • identifier: A unique ID for the speaker.
  • gender: The gender of the speaker.
  • name: The name of the speaker.
  • age: The age of the speaker.
  • file_name: The path to the audio file.
  • language: The language spoken.
  • language_code: A code for the language.
  • url: The source YouTube URL.
  • start_time: The start time of the clip in the YouTube video.
  • end_time: The end time of the clip in the YouTube video.

Usage

To use this dataset, you can clone the repository and then process the metadata.csv files to match audio files with their corresponding information.

The voices-celebrities directory includes a build_dataset.py script which demonstrates how to download the audio clips from YouTube using the provided metadata. You will need yt-dlp (installation wiki) and ffmpeg installed (website).

pip install pandas
cd voices-celebrities
python build_dataset.py

Contributing

Contributions to this dataset are welcome! You can contribute by:

  • Adding new voice samples to existing categories.
  • Creating new categories.
  • Improving the metadata.
  • Enhancing the data processing scripts.

When contributing, please ensure the audio quality is high and the metadata is accurate.

Legal Disclaimer and Removal Requests

The voice samples in this dataset are collected from publicly available sources for research purposes. We respect the rights and privacy of individuals.

If your voice is included in this dataset and you would like it to be removed, please follow these steps:

  1. Create a GitHub Issue: Open an issue in this repository with the title "Data Removal Request".
  2. Provide Details: In the issue, please provide your name and identify the specific audio file(s) you wish to have removed.
  3. Processing: We will process your request, remove the specified data from the dataset, and confirm the removal in the issue.

We are committed to maintaining this dataset ethically and will address removal requests promptly.

Downloads last month
55