Posts written by angelofarina

The version of Aurora for Audacity does not contain the Impulse Response Select module. So yes, you need to manually cut the results, insulating just the latest IR (linear) or the previous ones (harmonic order distortions).

Strength G; instead, can be correctly calculated in two ways.

1) the general good one, which I always recommend. Emit pink noise from your dodecahedron, having locked the gain to a precise calibrated value.
Measure its power level Lw in octave bands (either using ISO-3744 or ISO 9614 standards).
Then place the dodecahedron on stage in the theater, go with your sound level meter at the listening position, and measure the octave band SPL values while playing the very same calibrated pink noise.
Once you know the value of Lw in each band and the vlue the meaured ZPL in the each band, you can compute Strength G in each band as
G = SPL - Lw + 31 dB

2) the crap, approximate way from the impulse response. This is affected by an error of 2-4 dB, hence it is not recommended. However this can be done as "plan B" when method 1) was unfeasible. In this method, you still first to calibrate your sound source performing a fully anechoic impulse response measurement, with the microphone at a known distance from the source (say 3m), and keeping the gain locked at the very same value which later will be used in the theater. Also the rescaling gain applied during the convolution with the inverse sweep must be locked to a given value, which later must also be employed when processing the measurement in the theater.
You first open the anechoic IR, perform an Acoustic Parameters evaluation. At this point you press on the button labelled "Store G Reference Signal". You will be asked for the source-microphone distance, and the current values of G will be stored as reference.
After this, you open the IR measured in the theater, perform the Acoustical Parameter evaluation, and now the values of G should be "correct". But remember, this procedure is strongly approximate. Procedure 1) is way more reliable and preferable

Which bugs?
Please describe the problem, possibly with screenshots...

Audacity 2.4.1 is not compatible with 64-bit versions of Mac OS.
We need to wait that someone manages to recompile the Aurora plugins for the latest version of Audacity.
In the meanwhile, Aurora for Audition 1.5 works fine through Playonmac...

The error is selecting "cross-talk cancel only". This removes the calculation of inverse filters, performing just the cross-talk matrixing.
This option cannot work with real IRs measured, it can only work for Dirac-like pulses (synthetic IRs created with simulation software, or manually).
Just deselect thst option, and a proper set of cross-talk cancelling filters is generated.

We are working on the version 3.0 of Audacity, we hope to release it soon.
Currently Audacity 2.4.1, but also 3.0.2, is a 32-bit version only program for Windows. it is 64 bits for Mac.
see here:
www.audacityteam.org/download/

I think that having all the mikes with the same gain makes the processing easier. However, if you can set the gain in precise steps (as when using a preamp with digital gain control), nothing forbids to set different gains, take note of them, and later on compensating the SPL values applying these offsets. It simply makes it much easier to make errors, so I warmly recommend to fix a preamp gain which is good everywhere, and then keep it unchanged.

the Ir is too long. You should cut it much shorter, typically 300 to 600 samples, containing just the direct sound.
And the inverse filter should be much longer than the IR, say 4096 samples...

You must set an inverse filter length much longer than the IR to be inverted..
You can do this easily adding silence after the IR before invoking Kirkeby....

Disable the autorange and keep a constant gain in the convolver.
Then you can calibrate the Acoustical Parameters or the STI plugin against a calibrated SLM...

Exactly. The site for downloading Aurora with Audacity is: www.angelofarina.it/Public/Aurora-for-Audacity/
Please read carefully the readme.txt file of the version chosen...

Sorry, I cannot see the attachment. You can send me the wav file and the screenshot at my Email address, [email protected]
Did you properly set your computer for using English-UK international settings?
This is mandatory for Aurora...

First of all: why do you use the very old v. 2.0.0, when the newer version 2.4.1 is available?
Said that, I see that you are Italian.
Unfortunately Italians have the bad habit of using nationalised versions of their operating systems, with crap international settings. The most deleterious one is the number format, using a comma instead of decimal dot.
So simply setup your computer with English-UK settings. This is mandatory for most scientific software...

The current version of Audacity 2.4.1 has all the Aurora modules pre-installed and pre-enabled.
Just download the latest release (dated 11/11/2020).

I loaded on my web site a new version of Aurora for Audacity 2.3.3 with Asio. It is v.3.
This bug should be fixed now...

You know that these spatial parameters are quite unreliable, do you know?
Said that, the Aurora Acoustical Parameters module can compute Jlf, Jlfc and Lj.
You should analyze the B-format impulse response, in Fuma format (not Ambix).
So first convert your impulse response to Fuma B-format, then open it with Audacity and use Acoustical Parameters for computing Jlf.
A correct measurement requires that the X axis of the microphone was pointing exactly to the omnidirectional sound source. If the alignment was wrong, you can fix it using a free rotator plugin, such as IEM Scene Rotator, and using O3A Flare for checking that the alignment is correct.
Here an example of computing "spatial parameters" Jlf, Jlfc and Lj with Aurora for Audacity 2.0.0:

Here instead processing the same Ambisonics FuMa impulse response with the old good Audition 1.5 and Aurora plugins in XFM format:

As you can see comparing the two images, the values of these spatial parameters computed in Audacity 2.0.0 are completely wrong, while those computed by the Audition version of Aurora are perfectly reasonable.
Hence we can conclude that, if you are on a Mac and hence forced to use the old version 2.0.0 of Audacity, you cannot rely on the Aurora Acoustical Parameters plugin for computing correctly these spatial parameters.
I warmly suggest that you find and install the very old Audition 1.5, which runs perfectly fine on a Mac using Playonmac.

Edited by angelo.farina - 9/3/2020, 09:46