|
Hello Pierre,
thanks for your praise and feedback!
I'll try to answer to each of your three questions as best as I can:
1-You are right about the individual nature of binaural audio rendering. This is why within the 3D Tune-In Toolkit we allow for a certain level of customisation
2-I would argue that our auditory system does not "suppress" early reflections, but to a certain extent (depending on the delay and on the spectral and temporal characteristics of the stimulus) does not localise them. There are interesting studies looking at
the need for spatial accuracy when simulating reverberation - we have recently carried out some work on this topic as well (see
https://asa.scitation.org/doi/full/10.1121/10.0003437). Nonetheless I completely agree that it'd be great to be able to add a "spatial hearing model" which can inform the real-time
approximation of the accuracy of our reverberation rendering according to known psychoacoustics mechanisms...I'm though not sure such model exists yet!
3-Our priority for this work has been to create real-time simulations/applications rather than tools to be used offline by modelers to further explore the auditory mechanisms. For this reason, we have decided to implement the 3DTI Toolkit in C++ in order to
make it very efficient, creating then different releases which could all work in real-time. I agree that it'd be interesting to produce a Matlab or Python version of the Toolkit, but i need to confess that this is not a priority right now.
Thanks again for your feedback.
Best
Lorenzo
-- Dr Lorenzo Picinali Reader in Audio Experience Design Dyson School of Design Engineering Imperial College London Dyson Building Imperial College Road South Kensington, SW7 2DB, London T: 0044 (0)20 7594 8158 E: l.picinali@xxxxxxxxxxxxxx http://www.imperial.ac.uk/people/l.picinali www.imperial.ac.uk/design-engineering-school From: PIerre DIVENYI <pdivenyi@xxxxxxxxxxxxxxxxxx>
Sent: 23 June 2021 16:50 To: Picinali, Lorenzo <l.picinali@xxxxxxxxxxxxxx> Cc: AUDITORY@xxxxxxxxxxxxxxx <AUDITORY@xxxxxxxxxxxxxxx> Subject: Re: 3D Tune-In Toolkit - new releases of the VST plugins
Lorenzo, Your toolkit appears to be an extremely commendable and interesting effort to combine many things we know about auditory processing into a single package accessible by anyone fluent in C++. Still, I have three questions.
First, headphone-based auditory spatialization is known to be highly individual. Without individualizing these functions the spatial location of any auditory object will be uncertain, with a degree of uncertainty being a function of an azimuth-controlled
difference between the true individual external ear functions and those imposed by your model.
Second, while I applaud your wish to include room acoustics and reverberation in your model, we know that the auditory system suppresses early echoes. My recommendation would be to include such a suppression module consistent with known results starting
with Blauert’s work.
Third, most auditory modelers nowadays prefer working with higher-level packages, such as Matlab or Python, instead of the precise but clumsy C++. Have you thought of implementing your model on either of these platforms?
Best wishes to you and your team,
Pierre
Sent from my autocorrecting iPad
On Jun 22, 2021, at 21:49, Picinali, Lorenzo <l.picinali@xxxxxxxxxxxxxx> wrote:
|