Sound data is acquired in 3 different locations: an anechoic chamber, studio apartment, and dance studio. The design of the anechoic chamber stops the reflection of sound and electromagnetic waves, which results in an especially low amount of reverb. The anechoic chamber (430 sq. ft) is conducted with 44 speakers and has 137 hours of sound data. The studio apartment (250 sq. ft) cause moderate reverb and has 220 hours of sound data from 66 speakers. The dance studio (750 sq. ft) has a high reverb, with 213 hours of sound data from 68 speakers.
The rooms were included in each other dataset in order to address different levels of reverb. With the addition of audio datasets from various locations, you can build more accurate and reliable AI sound models.
Data sample according to recording space
Recorded in apartment
Recorded in dance studio
Recorded in anechoic chamber
If you would like to hear more samples, please contact us. (email@example.com)
We can provide about 180 samples for the Multiple Location Dataset.
We offered the multiple location data at a much lower price than other datasets since it doesn't include the other meta information such as speaker's gender, age, and etc.
However if you need more information about the speaker's demographic and content of speech, you may purchase other types of audio dataset we have. Please inquire for further information.
Feel free to contact us for more detailed information including the pricing and licensing.
Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0)