Corpora for simulating environments

Submitted by davidg on Mon, 2007-02-26 02:34.

The Corpora section of the web site has the title "Speech/Text Corpora". Currently, it also contains some collections of noises, which can be added to speech to simulate speech in noisy conditions. These noise collections are not speech or text. I also have some links to collections of room impulse responses that can be used to simulate room acoustics. I haven't submitted these links yet but they are not speech or text either.

Perhaps we should split Corpora into "Speech/Text Corpora" and "Environmental Corpora", with the latter containing things like noise collections, impulse response collections, or (hypothetically) network packet loss and jitter logs to be used in Voice-over-IP simulation?

( categories: )
Submitted by Christophe Van Bael on Mon, 2007-02-26 07:59.

Good suggestion, David. Thanks.

I decided to group all corpora under Corpora. You can now specify the type of corpus material in the corpora: speech, text, speech/text or other.

Christophe.

Submitted by davidg on Mon, 2007-02-26 21:25.

Great, thanks! Today I submitted several impulse response collections, with the label "other". It's great that we can bring more attention to these collections. I noticed a few things while I was doing this: (1) When I view entries in the corpora listings, the URL displays twice, but the href link is only correct for the second one. I suppose you've noticed this already. (2) In the entry at _labs_varechoic_chamber...
the URL only displays as .zip instead of the full URL which is .zip....
I guess the .zip suffix confused some code which was expecting only a .htm or .html suffix? (3) There are two noise collections already in the listings, "Additive Noise Sources" and "RSG-10 (NOISEX) noise collection". Perhaps someone with administrator privileges could add the label "other" to those.

Submitted by Christophe Van Bael on Tue, 2007-02-27 07:53.

Hi David,

@1: we'll script our way around it asap.
@2: likewise
@3: the list will soon be updated by one of our volunteering students.

Cheers,

Christophe.

Submitted by agus on Thu, 2007-03-01 00:52.

Hi David and Christophe,
I just fixed the problems @1 and @2.
~Agus