[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: request for dataset



Hi Shabih,

If the only labeling you need is between the categorical values you list above, then you could combine several datasets that have only one of those types into a meta-dataset that may suit your needs.

Here is a list of many speech datasets (with transcriptions): https://wiki.inria.fr/rosp/Datasets#Speech_datasets

You could add other types of sounds by downloading them from archive.org. They have many hours of live music, tv recordings (mostly videos from which you could extract the audio), political speeches, poem readings, etc. that may potentially be combined.

Hope that helps.

Best,
Ross



--
Ross Maddox, Ph.D.
Postdoctoral Fellow
Institute for Learning & Brain Sciences
University of Washington
phone: 206-685-4662
http://faculty.washington.edu/rkmaddox/

On Wed, Oct 15, 2014 at 8:04 AM, Syed Shabih Hasan <hasanshabih@xxxxxxxxx> wrote:
Dear All

I am working on creating a classifier that can identify live speech, music, media sounds (tv, radio etc). Can someone, please, point me to publicly available datasets of audio that are also annotated with the proper labels?

Best Regards
Shabih

— 
Syed Shabih Hasan
Graduate Student in CS
University of Iowa