[music-dsp] A question about Dirac pulses

Angelo Farina farina at pcfarina.eng.unipr.it
Sun Jan 11 07:56:01 EST 2004


> -----Original Message-----
> From: music-dsp-admin at shoko.calarts.edu 
> One of our members is very experienced at this sort of 
> measurement.  Care to comment, Angelo?
> Bob

I just repost here what I have just posted on the sursound list. I also
suggest that Glen downloads and tests my plugins for performing IR
measurements inside Cooldit (now Adobe Audition) 
Here is the link: HTTP://www.aurora-plugins.com
As always, any suggestion for improving them is welcome...


> -----Original Message-----
> From: sursound-bounces at music.vt.edu
> [mailto:sursound-bounces at music.vt.edu] On Behalf Of Jonathan S. Abrams
> Sent: 10 January 2004 23:51
> To: sursound at music.vt.edu
> Subject: [Sursound] WAVES Convolution
> Someone on a different newsgroup pointed the following out:
> The IR-1 Library (from WAVES)
> Waves worked with leading acoustician Prof. Angelo Farina to develop 
> new and innovative techniques (presented at the AES 24th International
> Conference) to create a library that establish new benchmarks for 
> clarity and accuracy in sampling reverbs.
> Can Angelo satisfy my curiosity and comment on this?

First of all, look here:
Regarding my involvement in the project, of course I cannot say too much
until the Waves plugin is officially released (it is just matter of days,
now). I suggest, as a starting point, to download the paper written by me
and Regev Ayalon (the researcher from Waves who performed most of the
measurements in various acoustical spaces all around the world).
All my papers can be downloaded, in PDF format, from
It is also interesting to have a look (and a listen to) the presentation
given at the AES-24 Conference.
You can access to my presentations at
The presentation also contains samples of the impulse responses measured in
some of the concert halls which had been measured at July 2003. Now this
number is much greater....
Regarding the Waves convolution software, I can just say that it will be the
first available convolver plugin supporting natively surround convolution
(the only other available software of this kind are FirReverb by Bengt Inge
Dalenback, and the BruteFIR project by Anders Torger).
I had the possibility to test a very preliminary version of the WAVES
convolver, and it worked very well, outperforming by approximately 30% all
the previously known convolvers....
I suppose that the release version will be even more efficient...
About the surround capabilities, initially the software will be tailored
mainly to discrete horizontal-only systems (5.1 and derivates), but, as
clearly explained in the paper, the impulse responses were collected with a
very open approach, which will allow in the future to support many other
formats: binaural, 1st-order Ambisonics 3D, higher-order Ambisonics 2D, and
hybrid formats (Ambiophonics, B+, mixed orders Ambisonics, Wave Field
Synthesis, etc.).
It is fair here to give thanks to the originator of this project, Ralph
Glasgal, who convinced the management of Waves about the effectiveness and
feasibility of convolution reverb, particularly for surround applications,
and placed me in contact with them.
I must also thank here the other researchers who helped this project: prof.
Ando from the Kobe University in Japan, Densil Cabrera from the University
of Sydney, Australia, Thomas Chen in USA, and many others in the various
sites which I and Regev did visit.
I can add that, thanks to the cooperation with Waves, the Italian
governement granted me a substantial research fund for the years 2004 and
2005, with the goal to measure, analyze and archive the spatial transfer
functions of dozens of Italian historical theatres. The results of this
research will be published, accompained with the measured data already
stored in the format suitable for being employed in the Waves software. This
will allow to everyone to listen to the virtual acoustical sound field of
these theatre, recreating their spatial effect.
I have also to say what is, actually, the greater problem which we did find:
the sound source! In fact, it resulted impossible, till now, to build a
proper loudspeaker which simultaneously has smooth, almost uniform
directivity and frequency response curves, and maintains a good efficiency. 
We built a special dodechaedron loudspeaker for being used as omni source
(which makes sense for simulating sections of an orchestra, particularly
when inside the orchestra pit of opera houses), and a Genelec A30D
loudspeaker (which insted is a good model for a singer or a soloist over the
stage). Both these loudspeakers, indeed, revelead to have too little power
in some situations, as when measuring outdoors, or in theatres equipped with
noisy HVAC systems, which was not possible to switch off during the
Although the measurement method employed (Aurora log-sweep method) gives a
sunstantial rejection of the background noise, in these problematic spaces
it was not possible to reach the scheduled S/N ratio of more than 100 dB,
which indeed was successfully achieved in all the other spaces. In the worst
case (Siracusa Greek Theater, which is inside the town of Siracusa, and in a
position which collects a lot of natural and traffic noise) the S/N ratio
was around 85 dB, which is still quite good, but definitely not the 104-108
dB achieved in some other theatres (there was one room in Israel where Regev
Ayalon managed to get a S/N ratio close to 120 dB).
In fact, one of the steps involved in the new research project funded by the
Italian governement will be the construction of a new reference sound
source. This will have digitally-controllable directivity pattern (so the
same loudspeaker will be employed both for measuring omnidirectional and
directive impulse responses), and higher efficiency due to partial
horn-loading of the drivers.


Angelo Farina

More information about the music-dsp mailing list