vadtoolkit

Created by Kim Jaeseok

version	1.1.0
license	GPLv3
source	https://github.com/jtkim-kaist/VAD
usage	commercial
languages	kor
format	wav
channel	1
sampling rate	16000
bit depth	16, 32
duration	0 days 02:00:09.703062500
files	4, duration distribution: 1801.5 s 1804.0 s
segments	588, duration distribution: 0.0 s 51.1 s
repository	data-public
published	2024-01-02 by audeering

Description

VAD Toolkit: A Database for Voice Activity Detection At each environment, conversational speech by two Korean male speakers was recorded. The ground truth labels are manually annotated. Because the recording was carried out in the real world, unexpected noises are included to the dataset such as the crying of baby, the chirping of insects, mouse click sound, and etc..

Tables

ID	Type	Columns
segments	segmented	noise	file start end noise bus_stop.wav 0 days 00:00:02.173875 0 days 00:00:03.775937 0 bus_stop.wav 0 days 00:00:05.641875 0 days 00:00:07.691937 0 bus_stop.wav 0 days 00:00:08.876875 0 days 00:00:11.498937 0 bus_stop.wav 0 days 00:00:12.468875 0 days 00:00:13.644937 0 bus_stop.wav 0 days 00:00:14.182875 0 days 00:00:16.498937 0

Schemes

ID	Dtype	Labels	Mappings
noise	int	0, 1, 2, 3	✓