As did I.
It's just a pity that drawing boundaries and moving them around is so difficult and inaccurate. What's really needed is line object (rather than pixel) based boundaries including options for bezier curves etc. Being able to move and adjust existing lines as well. Still pretty neat though.
I agree. I suspect that app uses a model where the boundaries are both rigid and very smooth therefore providing maximum reflection and diffraction, so the "head" would actually be simulating something more akin to a polished metal ball than a real head.
A real head would have quite a bit of absorption, both from hair and the softness and surface texture of skin, so I think there would be a lot less diffraction around the head at high frequencies than suggested by that app.
This is good though, and beneficial for stereo reproduction because it means the head would provide more shadowing for the occluded ear than suggested by a simplified simulation - giving increased cross-talk reduction for signals from the opposite hemisphere to a given ear.
The point of the test though, was to see what the Auto EQ system did to the direct, on axis response of the speaker at high frequencies (approximately 1-2Khz and up) when it was doing its thing.Simon, I left a strong reflection was on the impulse and I contend that the Evil manager just got lucky at guessing the distance.
By including a large early reflection in your "windowed" measurement, this question has still not been answered, as you still have a floor or wall bounce mixed up in the results which will totally shift the spectral balance.
Haha, cool.
Anyone having a nice 5.x setup to test it? ..... Earl?
Edit: Unfortunately no source files...
Last edited:
It impresses the hell out of me. So far, every digital radio station I've heard has been lacking. Lacking two things.
- Adverts
- Blah, blah, blah
(sorry for the OT)
I played with a summed LR center channel over 15 years ago and I liked it.
It had some problems I couldn't fix, so I went back to stereo. Eventually back to mono for around 10 years, for simplicity and good sounding voices.
Now I'm back to stereo and it's fun and far from perfect. No amount of engineering will make it perfect. Better, yes.
I know that instead of going to the superior 3 channel, 2 channel was chosen as good enough. Consumers would have to triple their mono spending to get the best sound. Doubling consumer spending was more doable.
Even then stereo was attacked as a way corporations to charge more.
It had some problems I couldn't fix, so I went back to stereo. Eventually back to mono for around 10 years, for simplicity and good sounding voices.
Now I'm back to stereo and it's fun and far from perfect. No amount of engineering will make it perfect. Better, yes.
I know that instead of going to the superior 3 channel, 2 channel was chosen as good enough. Consumers would have to triple their mono spending to get the best sound. Doubling consumer spending was more doable.
Even then stereo was attacked as a way corporations to charge more.
Haha, cool.
Anyone having a nice 5.x setup to test it? ..... Earl?
Edit: Unfortunately no source files...
Maybe you could bring half of El Cuerno as a centre speaker to complement keyser's system. I'd be willing to provide the surround channels
I was talking about real multichannel mixes that use level panning between adjacent speakers only. But you could use an EQ to match them (e.g. Acourate).
I think the bigger problem is differing directivity and differing height.
Look again. You can still see the change--and it's not big. 2ms of data is less than I'd like to see, but I did take a look b/c of the impulse. Change is essentially the same just less squiggly. If you want to me to do something to your liking, I'll need a list for next time. I'm satisfied and posted what seemed most useful here at home after looking at everything. I didn't save the data as it was no further use.
Dan
Haha, cool.
Anyone having a nice 5.x setup to test it? ..... Earl?
Edit: Unfortunately no source files...
I read the paper this is based on, it was very good, certainly gets the issue right. But I don't have MATLAB so I can't run his app.
There are several demonstration files available on the web site, with the original 2ch stereo and the upmixed 5.1 version so you can compare them.
There is also Irvan and Aarts method of upmixing, involving 'principal component analysis'. This method supposedly gives better crosstalk cancellation than other methods - larger sweet spot.
AES E-Library Two-to-Five Channel Sound Processing
AES E-Library Two-to-Five Channel Sound Processing
Just get a creative x-fi soundcard. CMSS is a direct implementation of Aveando and Jot method.
Exactly. I've extensively played with static 2-->3 rematrixing, simplest form is : C=L+R, L'=L-0.5R, R'=R-0.5L, and +45, 0, 45 deg angles, known as Trinaural. A very obvious prerequisite for these equations to work is 3 identical speakers wrt to SPL mag and phase, otherwise phantom sorces fall apart spectrally in localization direction and timbre etc
I found the simple "dumb" static matrices better than the few times I tried intelligent signal-analysing methods. Yes, this gives some signal dependant comb-filtering right in the speakers which only partly recombines to flat at the L.P., but the imaging it does for me excels the one of regular stereo. XY-components remain as direct and clear as before, mono center is even more solid (sides are identical signal and 6dB down in this case). AB-content gets even more airy and diffuse and big (which is what most any engineer/producer wants it to be).
Only for one specific signal condition there are less than all three speakers playing (then it's only two), 99% of signals recombine from three emitting sources, your head is placed in a different flavour of sound fields, combining differently than the fields from two sources only, it makes a more convincing illusion for me with only little drawback (HRTF-based "re-phantomization" like Q-Sound does not work with Tri, for obvious reasons).
Tri tends to have more stable imaging, you can turn your head and body hard right to refill your drink and "the stage" in front of you does not collapse that much and does not loose much of focus, one step closer to the ideal of full wave field synthesis. With conventional playback I always have the flooring sensation of getting very aware of the physical speakers during that move.
- Klaus
Last edited:
Thanks, but that seems to be primarily a subjective experience. What does the measurements show?
Seems to be no doubt that the CBT measures excellent vertically. But horizontally is more important in a home situation. Wide dispersion may be a problem and one might have do dampen the sidewalls a lot to get control over early reflections. Personally I don't believe that early sidewalls reflections are beneficial (even if the power response is good) like Floyd Toole seems to think.
The question may be stupid, but again; Will the CBT have time domain issues?
Would be interesting to discuss the CBT vs. a point source speaker like Geddes. The Geddes speakers become narrower at high frequencies. Do we want this or not?
An externally hosted image should be here but it was not working when we last tested it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.