Re: Listening Tests (=?iso-8859-1?Q?S=F8ren_Bech?= )

Subject: Re: Listening Tests
From:    =?iso-8859-1?Q?S=F8ren_Bech?=  <SBE(at)BANG-OLUFSEN.DK>
Date:    Tue, 25 Feb 2003 22:07:32 +0100

Hi again

The goal in watermarking and low bit-rate codecs is the same: the coded
version should be indistinguishable from the original. If you check the ITU
cite you will find that ITU at the moment allows for free downloads of
three standard so try and get the BS.1116 - It will answer all (I guess :
-)) of your questions on procedure, experimental set-up, analysis of the
results etc. You should be aware that this type of test where you have
quite small perceptual differences are very difficult to conduct if you
want reliable results - so my best advise is to check the standard - it has
been used for all the published results in e.g. JAES - you could check that
Journal too.

Best from Soren


                                                                                                                                       
                      Lawrence Miranda                                                                                                 
                      <lawrence.miranda        To:       AUDITORY(at)LISTS.MCGILL.CA                                                      
                      (at)UP.EDU.PH>              cc:                                                                                     
                      Sent by: AUDITORY        Subject:  Re: Listening Tests                                                           
                      Research in                                                                                                      
                      Auditory                                                                                                         
                      Perception                                                                                                       
                      <AUDITORY(at)LISTS.M                                                                                                
                      CGILL.CA>                                                                                                        
                                                                                                                                       
                                                                                                                                       
                      26-02-2003 13:36                                                                                                 
                      Please respond to                                                                                                
                      Lawrence Miranda                                                                                                 
                                                                                                                                       
                                                                                                                                       




The goal in audio watermarking is that the listener should not distinguish
any difference between the original audio from the watermarked one. Now to
test this, I mean, should a same-different test do? Are there other tests
more suitable and generously documented and has its free gui
implementation?
I don't have ITU access. Thanks.

Lawrence Miranda


----- Original Message -----
From: "Søren Bech" <SBE(at)bang-olufsen.dk>
To: "Lawrence Miranda" <lawrence.miranda(at)UP.EDU.PH>;
<AUDITORY(at)LISTS.MCGILL.CA>
Sent: Tuesday, February 25, 2003 11:10 AM
Subject: Re: Listening Tests


>
> A good source to begin with is the ITU-R recommendation BS.1116 - This
> standard was developed for listening tests on low bit-rate codecs with
> small impairments. I don't know what type and magnitude of impairments
you
> are planning to examine but if they are larger there is another ITU-R
> standard (not accepted yet) for that: BS.[DOC.6/106(rev.1)-E](1534)
>
> You can obtain these docs at the ITU web page
>
(
http://www.itu.int/publications/main_publ/itur.html#ITU-R%20Recommendations
%20on%20paper)
>
> Hope this helps
>
> Best from
> Soren
>
> Dear List,
> What are the proper/most accurate listening tests to test the audibility
of
> a watermaked audio signal? I would greatly appreciate any links to
> resources/papers/tutorials/matlab codes for gui/source codes  to conduct
> the
> listening tests.