Dirac and frequency shifting?

[firedog]

Hi-

 

Haven't used Dirac but in the Audiostream review, much improved results were tried using the "sofa" measurement setting (wider sweet spot) as opposed to the "chair" setting. I know this isn't a learned technical response like some of the others, but it is a simple tweak and might help.

[flak]

Hi-

 

Haven't used Dirac but in the Audiostream review, much improved results were tried using the "sofa" measurement setting (wider sweet spot) as opposed to the "chair" setting. I know this isn't a learned technical response like some of the others, but it is a simple tweak and might help.

 

Yes, your remarks and the review itself are interesting, it also should be noted that the DSP gain control was not available at the time of this early review... and in many cases that's an important setting to be applied as observed by many CA forumers.

 

Flavio

[EuroDriver]

Hi tranz,

 

to better understand the results you reported, can you give us details of your playback system, and measurement mic, calibrated or non calibrated ?

 

also what your current conclusions are ?

[tranz]

Hi-

 

Haven't used Dirac but in the Audiostream review, much improved results were tried using the "sofa" measurement setting (wider sweet spot) as opposed to the "chair" setting. I know this isn't a learned technical response like some of the others, but it is a simple tweak and might help.

 

Thanks for the suggestion. My assumption was that the graphics were more a visual way of putting more distance in between measurements as opposed to it creating different filters from using a chair selection. In other words if I choose a 1m box in either chair or sofa, the math used and end result of the filter is exactly the same.

 

Perhaps Flavio can confirm this though.

[tranz]

Hi tranz,

 

to better understand the results you reported, can you give us details of your playback system, and measurement mic, calibrated or non calibrated ?

 

also what your current conclusions are ?

 

Hi EuroDriver,

 

I bought the recommended XTZ mic Pro package and used the calibration file provided on the Amarra website.

 

Sonic Studio Amarra High Resolution Music Players and Professional Audio Mastering Systems

 

The integration with Amarra around the volume is too cumbersome and I constantly have clipping. Perhaps due to that it did more harm than good in my setup.

 

At this point I have abandoned the IRC for Amarra. Perhaps I will revisit when there is better integration or some other tweak to try. This of course is not to say it will not work for non-Amarra and other physical setups as there are just so many variables.

 

Cheers

[dallasjustice]

Thanks for the suggestion. My assumption was that the graphics were more a visual way of putting more distance in between measurements as opposed to it creating different filters from using a chair selection. In other words if I choose a 1m box in either chair or sofa, the math used and end result of the filter is exactly the same.

 

Perhaps Flavio can confirm this though.

The sofa isn't a box. It's different than the chair. I think chair sounds better. I am sure preferences vary depending on the room.

[sdolezalek]

Any steep frequency or phase response corrections around midrange will make sound really strange. I personally prefer to design corrections based on 1/6th octave smoothed response which helps avoiding too aggressive correction filters that would ruin the sound.

 

I was interested in your views on this in part because of the comments you were making in the "Soundstage - Is it all in the timing?" thread. My concern in my own trials with Dirac was that the introduction of their frequency/phase corrections would affect the delicate timing information our brains want. I read Flavio's suggestions about limiting the correction to the lower octaves as being in line with that thinking as the sub 700 Hz sounds are harder to localize and thus perhaps more immune to damage in timing vs. improvement in frequency response.

 

What I haven't tried, but would be interesting is to first combine passive room treatments focused on places where Dirac suggests there are problems and only at the end use Dirac to clean up a few things that are otherwise hard to pick up. In my own system, I too found that I preferred the uncorrected version with less favorable frequency response over the flat frequency response version. Of course my pre-processor is doing its own timing adjustments for the various speaker distances from the listening position (so Dirac doesn't need to), but I suppose, it too may be adding timing issues in its delay functions.

[wgscott]

Can (or did) you try disabling the pre-processor and try it with only Dirac?

[sdolezalek]

Can (or did) you try disabling the pre-processor and try it with only Dirac?

 

I haven't tried that. The processor asks that you either set the speaker distances from the listening position manually or it has a four microphone calibrated proprietary Lexicon calculation that is very much like that in Dirac to establish the distances. I guess I could set all the distances manually at [15] feet and then let Dirac do its own distance calibration.

[REShaman]

The sofa isn't a box. It's different than the chair. I think chair sounds better. I am sure preferences vary depending on the room.

 

I agree. Excellent perspective. Share the same perception but that is with Amarra Symphony 2.6 with iRC. Others have a different experience. What else is new? Main thing,

 

Enjoy the music,

Richard

[EuroDriver]

To the Dirac users who find chair better than sofa, what size measurement box did you use for your mic positions ?

 

also have you limited the frequency window ?

 

I am limiting to 500 Hz for left speaker and 700 Hz for the right speaker. So far only sofa measurements taken

[flak]

I think that some of the above posts somehow imply an interesting question... why do we sometimes suggest to limit the Dirac Live intervention to an undetermined and variable region of frequencies?

 

A comment came from Miska who said:

"Any steep frequency or phase response corrections around midrange will make sound really strange. I personally prefer to design corrections based on 1/6th octave smoothed response which helps avoiding too aggressive correction filters that would ruin the sound"

 

Miska's comment is very to the point and accurate in general.

However, it does not apply to Dirac Live as we do work on a smoothed model as he suggests.

Also we have a system that checks for preringings, making it impossible to produce filters with audible preringings.

As he correctly states this makes it impossible to correct for some things and our preringing check has a more cautious compensation as a result.

 

I'm suggesting to limit the intervention towards the high or even mid frequencies when the listener is perfectly happy with the sound of his speakers at those frequencies so that he prefers the "before correction" sound in that region of frequencies... but why would someone for example prefer a measurably less linear response?

 

There is a number of factors... i.e. if we look at the frequency response everybody is looking for a "neutral balance" but a perceived neutral balance can be different from a linear frequency response, several listeners very much like the sound of the LS3/5A which feature a willingly modified frequency response (you may google about the BBC dip) and in general many speaker manufacturers have applied some sort of "voicing" which gives their unique character to their products.

 

Listeners have often spent many years in selecting the speakers of their liking so it is no coincidence that they prefer that specific tonal balance in their room with their recordings... even more so because recordings themselves are an intertwined factor and the monitors which have been used had a tonal balance of their own which influenced the recordings (again you may google i.e. about the NS10M which had a 'characterful' tonal balance and were popular in the studios in the past)

 

In conclusion in real life we have to find our own right correction which works on the average of our music material and sometimes this may mean either tailoring the target curve to be similar to the original average response or limiting the correction to a region of frequencies... in other words we need a tool that is flexible and powerful enough to easily adjust to our specific requirements and I think that Dirac Live is effective at that.

 

Good listenings

Flavio

[tranz]

Thanks Flavio!

 

Can you please let me know whether the choice of chair or sofa actually changes the math used and filter created? Or is it more a visual guide of where to put the mic. I.e. Using a mic placement of 1.6x1.6x1m area in chair mode will render the exact same filter as in sofa mode?

 

Cheers

[SteveS1]

I did both chair and sofa projects and found sofa far more to my liking but both were considerably better than no filter. Chair was .6m radius Sofa was .75m. The first project I did with Chair at .5m radius was the least effective.

[sdolezalek]

I think that some of the above posts somehow imply an interesting question... why do we sometimes suggest to limit the Dirac Live intervention to an undetermined and variable region of frequencies?

 

I'm suggesting to limit the intervention towards the high or even mid frequencies when the listener is perfectly happy with the sound of his speakers at those frequencies so that he prefers the "before correction" sound in that region of frequencies... but why would someone for example prefer a measurably less linear response?

 

In conclusion in real life we have to find our own right correction which works on the average of our music material and sometimes this may mean either tailoring the target curve to be similar to the original average response or limiting the correction to a region of frequencies... in other words we need a tool that is flexible and powerful enough to easily adjust to our specific requirements and I think that Dirac Live is effective at that.

 

Good listenings

Flavio

 

Thanks Flavio. You addressed part, but not all of the question I was raising about Miska's comment. In general, I think we would all prefer a more linear response if we give ourselves the time to get used to it. But my question also implied that I think there are tradeoffs to be made in achieving a smoother and more linear response. Miska suggested that one of the potential effects of filtering can be to create timing delays or phase shifts and that those timing differences effect what we hear, especially as regards the "localization" of specific instruments in space and that these timing differences effect us much more strongly at high rather than low frequencies (he specifically refers to the fact that subwoofers are less critical in spatial placement because their frequencies are much harder for our ears to localize).

 

By, as you put it "limiting the intervention towards the high or even mid frequencies" you seem to agree with Miska's overall view, but I wasn't clear whether that was because you agree that the necessary filtering for linear response must introduce some level of time smearing/shifting? You specifically refer to taking "pre-ringing" into account, but have you run tests that would show that if I limit my correction to say 700Hz and below that there are no timing effects on frequencies above that range?

 

Thank you for all of your responsiveness on the various Dirac threads. It is a very interesting piece of software, but I think all of us are still trying to figure out just how we use it alongside other software/hardware configurations of our choice.

[flak]

@ Tranz,

 

hello, the math is the same and as you suggested as of now it is a visual guide... Dirac Live has been designed this way to accomodate some planned future modifications

 

@ Sdolezalek,

 

hello... filters can, as mentioned before, introduce time shifts (an equal phase shift for all frequencies) or phase shifts.

However, placing a speaker in a room is a filter... the speaker position, the listening position and the room "decide" what this filter will look like

Moving the listening position, even slightly, will change this filter.

To handle this it is possible to add another filter to even out the frequency response of the "speaker + room filter", in the process possibly introducing even more phase shifts.

 

However, it is possible to add a third filter, an all pass filter, that is, a filter with flat frequency response that only introduces phase shifts to cancel the phase shifts introduced by the "speaker+room"

The net result is then several phase shifts in series, but (in theory) a linear phase and a time shift (due to the phase shifts).

At this point a frequency response AND a phase compensation has been performed, or to see it in another way, the impulse response has been corrected.

This is also the reason why you cannot correct for everything using a traditional minimum phase EQ (you cannot do anything about the phase)

 

As a result Dirac Live will improve the measured outcome (frequency and impulse response) so that the "output" will be more similar to the "input" but if a listener is happy with a "filter" (I call it that way to make me understood) that he has selected after many trials it may be perfectly reasonable to leave it as it is by limiting the intervention (I explained the reasons in my previous post)

 

Flavio

[sdolezalek]

However, it is possible to add a third filter, an all pass filter, that is, a filter with flat frequency response that only introduces phase shifts to cancel the phase shifts introduced by the "speaker+room" The net result is then several phase shifts in series, but (in theory) a linear phase and a time shift (due to the phase shifts). At this point a frequency response AND a phase compensation has been performed, or to see it in another way, the impulse response has been corrected.

 

Flavio

 

Thank you, very helpful. I'll have to spend more time looking at the impulse response graphs on a before and after basis and see if I can correlate what i see with what I hear.

[Boris75]

@ Sdolezalek,

 

hello... filters can, as mentioned before, introduce time shifts (an equal phase shift for all frequencies) or phase shifts.

However, placing a speaker in a room is a filter... the speaker position, the listening position and the room "decide" what this filter will look like

Moving the listening position, even slightly, will change this filter.

To handle this it is possible to add another filter to even out the frequency response of the "speaker + room filter", in the process possibly introducing even more phase shifts.

 

However, it is possible to add a third filter, an all pass filter, that is, a filter with flat frequency response that only introduces phase shifts to cancel the phase shifts introduced by the "speaker+room"

The net result is then several phase shifts in series, but (in theory) a linear phase and a time shift (due to the phase shifts).

At this point a frequency response AND a phase compensation has been performed, or to see it in another way, the impulse response has been corrected.

This is also the reason why you cannot correct for everything using a traditional minimum phase EQ (you cannot do anything about the phase)

 

As a result Dirac Live will improve the measured outcome (frequency and impulse response) so that the "output" will be more similar to the "input" but if a listener is happy with a "filter" (I call it that way to make me understood) that he has selected after many trials it may be perfectly reasonable to leave it as it is by limiting the intervention (I explained the reasons in my previous post)

 

Flavio

 

This is extremely clear, many thanks.

[flak]

Thanks Sdolezalek and Boris,

 

a testimonial of the usefulness of trials and testing together with the flexibility of the DRC solution can be seen here:

http://www.computeraudiophile.com/f23-dsp-room-correction-and-multi-channel-audio/new-dirac-series-minidsp-19908/index2.html#post318869

 

Flavio

[dallasjustice]

I agree. Excellent perspective. Share the same perception but that is with Amarra Symphony 2.6 with iRC. Others have a different experience. What else is new? Main thing,

 

Enjoy the music,

Richard

I got a new sofa. So I re-measured. Now the sofa sounds better than chair. I have no explanation. The new sofa is more solid than the old one. Who knows?

[REShaman]

I got a new sofa. So I re-measured. Now the sofa sounds better than chair. I have no explanation. The new sofa is more solid than the old one. Who knows?

 

I am inspired. Shall re-measure as if I am sitting on a sofa instead of my Ekornes "Jazz" recliner. Always open to possibilities.

 

Enjoying the music,

Richard

 

[dallasjustice]

I narrowed it down to stressless and cinematech. I bought a cinematech sofa. Here's a shot of it:

 

All three seats recline and have retractable headrests. Alcantera fabric.

 

Michael.

[REShaman]

I narrowed it down to stressless and cinematech. I bought a cinematech sofa. Here's a shot of it:

[ATTACH=CONFIG]12339[/ATTACH]

 

All three seats recline and have retractable headrests. Alcantera fabric.

 

Michael.

 

Looks lovely. Alcantera fabric also lovely. Enjoy your sofa and the music!

Best,

Richard

[retro]

Well..I use the chair option..

 

[wgscott]

My listening spot is the right-hand end of a long sofa, so i had been using the chair setting. Should I just experiment, or is there a suggested option or preference?