Gentlemen,
I have a question about auralizations and Auditioner. How is RT calculated and what are you hearing in auralizations in Auditioner? Is RT calculated using the RT source only? When you auralize, is the reverberant field you are hearing based on the omnidirectional RT source or is it calculated using the loudspeakers and their directivity as the source(s)?
This is a big issue if you need to demonstrate the benefits of a loudspeaker system using high Q devices, especially if using sizable fixed or steerable line arrays. If Modeler only calcuates the reverberant level based on the RT source, then auralizing high Q devices would be invalid.
Thanks,
Jerrold
Hi Jerrold, before my other colleges will answer in a more scientific way I have attached files for you to listen to a A/B test of two different systems in the same room.
They are binaural recorded using the track I also have attached. The track is made with a 10sec pause between the speeches so it could be played on Auditioner and a Windows player at the same time. Just start the second 10sec after the first.
The A gain is calibrated by first setting the volume on the windows player to a comfortable level, then calibrate the B gain in the Modeler Playback Tab until its equal.
Thomas also posed a nice picture how the track works.
http://proforum.bose.com/eve/f…010169541/m/79810813
The files are divided into 3 replies, first the Modeler file and the dry track, then the two binaural recordings.
Have fun
Best, jorgen
Opps, it turns out that the maximum size is 1MB…
Jerrold and if there is anybody else that would like the recording, please mail me at:
jorgen_allen@bose.com
And I will mail the files instead.
Best, jorgen
Great, thank you very much Stephen!
I have attached Modeler files, Binaural rec and the dry paused audio track in the same zip.
Best,
Morten,
We actually discussed the question of whether high Q sources change the reverberation time in a space at the SAC Equalization workshop in Salt Lake City a couple of years ago. I stated that you can’t change the actual RT60 in a space, just the level of the reverberant field. It can “sound like” the time changed due to the fact that when the reverberant field starts at a lower level, then the tail could dissappear into the noise floor of the room quicker giving you the impression the the time is shorter. All of the experts there agreed with this assesment.
Jerrold
Thomas,
Warning: I am not an acoustician, but I did sleep at a Holiday Inn Express last night.
I agree that EDT and ITG have a greater impact on intelligibility than “reverberation time” in most cases. I would, however, say that excessive reverberation time and level do have a significant impact as well. All the more reason for me to design systems with appropriate degrees of pattern control. It also validates the benefits of well behaved linear array loudspeakers for highly reverberant spaces.
The good news, if I’m understanding your answers to my orignial question, is that Modeler does calculate these early reflections per loudspeaker and based on the Q of each loudspeaker (and I’m assuming at the Q of that loudspeaker at each Nth oct band). If so, then I could assume that the auralizations generated would be representative of reality.
Jerrold
Hi Jerrold,
We use the omnidirectional source to calculate the RT60 of the room.
The reverberant field level in the HEDC - that you listen to - is based on the actual loudspeakers in your design, however, it assumes that the decay rate at the late part of the reverberant field is defined by the RT60 above.
So when you predict and listen to a high-Q system - like an MA12 system - you will see and hear substantally less reverberant energy.
(I do question and would like to understand whether a high-Q system changes the RT60 of the room, or simply changes the amount of reveberant energy. I have not seen any papers on this topic - but I am sure they exist.)
Listen to Jorgen’s A/B comparisons…
Morten wrote: (I do question and would like to understand whether a high-Q system changes the RT60 of the room, or simply changes the amount of reverberant energy. I have not seen any papers on this topic - but I am sure they exist.)
Some of us “old timers” would refer to this question as the high-Q system changes the “apparent” reverberation time. That is a lay listener would comment the room has a lower reverberation time when listening to the high-Q system. Now we should obtain a lower RT60 measurement using the high-Q system as compared to a dodec or 802’s, etc. However the raw acoustics of the room have not changed.
Hope this helps,
Phil Nelson
quote:it turns out that the maximum size is 1MB…
Jorgen, I increased the attachment size allowance to 50Mb. That should allow us to share most models even if the filters are retained.
If you will please update your previous post with the original attachment, everyone will benefit from seeing the example.
Thanks!
Jerrold,
So, the good news is that Modeler mimics what we believe is the correct approach.
Hi everybody,
here are some additional remarks from my side.
I think it can’t be stressed enough that almost all attributes that are important for either voice or music reproduction in a space, whether amplified or not, are less dependant on “the reverberation time” of a room but much more on what happens in the very early part of the decay. The diffuse sound that masks speech components important for intelligibility and that also diminishes clarity in music reproduction comes much earlier in time than the energy that would typically be used for measuring the reverberation time (i.e. decay ranges from -5 to -25 or -35 dB). The early decay time (EDT) is often much better suitable to evaluate this relationship. In large spaces, there’s also a lot of time passing until any reflective surface is hit by sound from the loudspeaker, so for many milliseconds, nothing’s there but a big (Initial Time Delay) gap.
Side note: Having just returned from the International Conference on Acoustics in Rotterdam, I can report about a plenary lecture from Tammo Houtgast from TNO, one of the two inventors of the STI method, where RT-5 to RT-10 numbers were found to give a much better correlation to intelligibility than the mentioned (standardized) RT-20 or RT-30 values.
In Modeler, everything that concerns early diffuse energy is predicted individually for each sample or listener. The build-up, plateau and start of linear decay of the diffuse sound (the yellow envelope function) is typically different for each position and depends not only on the directivity of the source but also on the distance to the next specular and diffuse reflecting surfaces.
Note that, although the late part of the envelope (what we call the RFEF, the Reverberant Field Envelope Function) is indeed decaying linearly according to the predicted (global) reverberation time, it is not like there’s a “tail” attached at some (fixed) point in time. The decision about when sound is starting to decaying in a linear fashion is made on a ray-by-ray basis. This is also the reason why you may encounter situations where late echo arrivals are visible in regions of the HEDC time response that already decay according to the RT.
In general, I am the opinion that the reverberation time as we know it should be attributed to the room itself and not to the combination of source and room. For this reason, it also makes sense to standardize the use of “omni-directional” sources fur the purpose of making RT measurements, even though such sources are seldomely (if ever) used in typical venues.
Best rgds,
Thomas