AL future development suggestion

[Iansr]

I’ve been reading / watching a lot about cross talk cancellation software and the benefits it delivers in stereo imaging and ambience recreation. This is all done using filters in the digital domain of course, so how cool would it be if AL had a cross talk cancellation option that combined with its standard filters so as to also deliver these additional benefits.

I appreciate this would probably involve quite a lot development work, but the end result would be really something and would (a) put clear water between you and every other DSP solution and (b) considerably strengthen and widen AL’s appeal.

 

[Omid]

I’ve been reading / watching a lot about cross talk cancellation software and the benefits it delivers in stereo imaging and ambience recreation. This is all done using filters in the digital domain of course, so how cool would it be if AL had a cross talk cancellation option that combined with its standard filters so as to also deliver these additional benefits.

I appreciate this would probably involve quite a lot development work, but the end result would be really something and would (a) put clear water between you and every other DSP solution and (b) considerably strengthen and widen AL’s appeal.

That's an interesting concept. I haven't read much about it, so I might be missing something. At first it seemed like a good idea, but thinking about it a bit more raised some issues for me:

Most of the music we listen to is not a live recording, but a studio engineered mix. We' re trying to hear the same thing the studio engineer heard at his/her station. So the 'crosstalk' is part of the aural experience the engineer heard and expects you to have. If the recording was taken in a live setting with a dummy head and binaural mics, then I could see why one would want the right speaker to only reproduce what the right ear heard with no crosstalk (and I think that has been done with mixed results in the past). Also, this correction implements a crosstalk cancellation that would only work if your head is precisely centered. Move right or left and you'd get a comb filtered result instead. Another concern is that the audio signal is more or less mono in low frequencies, so the low frequency signal might come out abnormally low.

I checked on Roon to see if there is a way to implement what you're looking for and found the opposite. It's called the "crossfeed" filter. It adds crosstalk to headphones so they sound more like speakers...

I found a study with some useful info on this (https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3561850/).

Sorry re-reading my post, it sounds negative. I have not tried or heard this, so I might be totally off base. If it works well for you, please ignore all the above comments.

 

[juicehifi]

I have taken a look at various versions of cross talk cancellation on several occations, but it seems to me like very few are interested in this. I will keep it on the watch-list for now. If I see the interest blossom I may take it on.

 

[Dali_20]

I take advantage of the thread.
Regarding the reduction of crosstalk, Ubacch usage is increasing, but the results are poorly documented and in any case always of a subjective type.
And I suspect that any such system cannot provide particularly good results on a large listening area, it is technically impossible, therefore it is difficult to justify that cost.
The search for the best performance/compromise on a large listening area rather than on a sweet spot should be the focus for today DRC developments.
Dirac in fact is attempting at this with its Bass Control and Active Room Treatment (ART) technologies.
But having Dirac Bass Control in my system I can say that it is far from being in the state of the art... it manifests extreme sensitivity to measurements, to the point that two filters created with the same measuring scheme sound completely different.
Not only that, with the same measurements the software is able to generate at least two different types of filters, based only on the filter creation iteration (if I press create one filter is generated... and then if I press again a little different filter is created... and so on... inexplicable).
In addition, the filters generated with Bass Control dramatically increase the group delay at low frequencies, even going from 20ms of the raw system to 120ms of the corrected system.
Dirac admits this malfunction but is in hiding in finding a solution.
As for ART, it seems more promising but the result depends a lot on how many speakers the system has and how much they extend at the bottom.
For the classic 2.x stereo there is to be seen what benefits it will provide.

So going back to Audiolense, I believe and I take the liberty of suggesting an algorithm to improve the bass on a wider area as the next step of development. Basically the low frequencies is where there is the most variation depending on the position in the room, and from the research they account for a good 30% / 40% in the perceived audio quality.
Improving the phase coherence between the two speakers could be a subsequent or parallel development step (as Acourate DRC does).
Reducing crosstalk seems much less a priority than these things, and surely could not justify the development effort compared to the potential consumer growth.

Just my 2 cents...