Re: Gaussian vs uniform noise audibility (Bob Masta )


Subject: Re: Gaussian vs uniform noise audibility
From:    Bob Masta  <audio(at)DAQARTA.COM>
Date:    Wed, 9 Jun 2004 08:44:49 -0400

Dear list: Back at the end of January I posted a note here regarding my observation that there was no audible difference between Gaussian and uniform noise. At that time my DaqGen software had problems generating narrow distributions, so my findings only applied to distributions whose tails went out to about 4.35 deviations (user Standard Deviation setting of 1.00 or more); narrower distributions whose tails extended farther were audibly different from uniform noise, due to this problem. The lastest version of DaqGen, v1.20 (free download), has corrected this difficulty, and allows the tails to extend out to 8.7 deviations (user Standard Deviation of 0.50 or less). At a sample rate of 48000 Hertz, that would result in peak tail samples (+/-32767 on a 16-bit system) only about once per 45000 years. (!) I apologize for the delay in rectifying this problem. In addition to developing the new wide-range algorithm, I have added extensive signal averaging options to DaqGen, including histograms that allow visualization of the distribution. As always, I would be glad to answer any questions. Bob Masta audioATdaqartaDOTcom


This message came from the mail archive
http://www.auditory.org/postings/2004/
maintained by:
DAn Ellis <dpwe@ee.columbia.edu>
Electrical Engineering Dept., Columbia University