Dirac Live Impulse Response View/Results

[camott]

Hi,

 

Long time lurker, first time poster.

 

I am trying out Dirac Live on my mac mini (el capitan) and am wondering if someone - Flavio? - can explain what I am seeing in the attached Impulse Response view after running my measurements and optimizing.

 

I get the basics of the impulse response graph - in an ideal world there would an immediate spike with all energy and no reverb/slow decay afterwards (which are due to reflections etc).

 

So that said:

1. why does the before impulse response start at 10ms and not 0ms? I don't understand why time zero is at 10ms. Does this suggest that it's taking 10ms from time of signal to first retrieval and thus correlate with an approximate 10 ft distance from speakers to listening position/mic? (assuming 1 ft per ms)

2. why is the after/corrected response starting at 26ms? is the Dirac processing adding another 16 ms delay before signal is reached at listening position/mic?

3. How does this show that impulse response is "improved"? I don't see less impulse response messiness - just that it has been pushed 16ms later and looks like it has been "shoved" together into a shorter time period. I have read that Dirac tries to remove early reflections/timing issues and keeps the later ones as those are perceived to sound nice - but I don't quite see nor can correlate this in the graph.

 

the spectrum/frequency response curve optimized "after" curve is more obvious and intuitive. There is really no information in the Dirac documentation that explains what one is seeing on the Impulse Response screen.

 

Note that there are also bugs in the apps:

a. the DAP application will not reopen once closed on my mac mini with 10.11.1 - I have to restart the mac in order to open the DAP application. I don't see anything specific in the system/console logs to indicate why it's not reopening.

b. my latest project will cause the DLCT calibration app to crash when i try to go to "filter design" all of a sudden. But other project files do work.

 

As a longtime senior software engineer myself, these kinds of bugs are really frustrating and shouldn't exist in a $500+ piece of software whatsoever. Makes it a really hard sell.

 

Any help would be much appreciated and helpful in making me make a decision about Dirac!

 

Thanks,

Christian

 

[wgscott]

Try moving the kext from /System/Library/Extensions to /Library/Extensions

 

Their installer puts it in the wrong place, and Apple got more anal. I am surprised it launches at all.

 

The AU plugin, FWIW, works flawlessly. I don't have answers I am confident in for the other questions.

[camott]

Try moving the kext from /System/Library/Extensions to /Library/Extensions

 

Their installer puts it in the wrong place, and Apple got more anal. I am surprised it launches at all.

 

The AU plugin, FWIW, works flawlessly. I don't have answers I am confident in for the other questions.

 

Thanks for that. Actually, I did try moving the kext and I am wondering if it caused the issue with my project not loading into DLCT. I with fiddle with the kext some more and see if it resolves the app launching issue.

 

While I have Audirvana and was thinking of trying the AU plugin with it, I am pretty much hooked on Roon - the ui/discovery/tidal integration is in a league of it's own. Roon plays fine with Dirac in the background though if it could support AU plugins and remove one more layer (DAP and kext) that would be even better.

[camott]

I think I confirmed most of my suspicions about the impulse response results after reading this thread http://www.computeraudiophile.com/f23-dsp-room-correction-and-multi-channel-audio/listening-rooms-and-pulse-response-17633/

 

My main remaining question is what does the fact that my impulse response graph has lots of ongoing noise after (to the right of) the main initial spike mean? i.e. the amplitude doesn't really calm down at any time after the initial spike. Does this imply lots of slap echo or hard reflections or something?

 

Cheers

Christian

[wgscott]

You mean the stuff below 0.02? My guess is it is ambient noise (which is why it doesn't decay). That should be far below the threshold of hearing.

[wgscott]

I know this isn't ideal, but you can use the AU plug-in with anything.

 

Use AU plugins for free with any Apple audio software - Blogs - Computer Audiophile

[flak]

1. why does the before impulse response start at 10ms and not 0ms? I don't understand why time zero is at 10ms. Does this suggest that it's taking 10ms from time of signal to first retrieval and thus correlate with an approximate 10 ft distance from speakers to listening position/mic? (assuming 1 ft per ms)

 

Yes, that's correct

 

2. why is the after/corrected response starting at 26ms? is the Dirac processing adding another 16 ms delay before signal is reached at listening position/mic?

 

The 16 ms delay is necessary because what we want to do with the mixed phase filters is to move all the frequencies so that they come at the same time.

 

Obviously we cannot move them all to time zero, as this requires us to know about the future.... instead we shift all of them to the latest frequency, that is we delay frequencies to match the latest one

 

This latency does not have any influence on sound quality, it only means that the music playback starts 16 ms later

 

3. How does this show that impulse response is "improved"? I don't see less impulse response messiness - just that it has been pushed 16ms later and looks like it has been "shoved" together into a shorter time period. I have read that Dirac tries to remove early reflections/timing issues and keeps the later ones as those are perceived to sound nice - but I don't quite see nor can correlate this in the graph.

 

The impulse response in your graph is much improved by the Dirac correction... it is much higher, there is less ringing and there is much less below zero (where we would like to see nothing)

 

 

This image with an "ideal" impulse response should explain this better than my words:

 

 

Flavio

[camott]

Thanks for the response Flavio - as I later discovered, that falls in line with what I read in the other thread that got into this similar discussion.

 

Can you clarify how you see "less ringing" in the impulse response graph? is that the small/frequent bounces after the initial spike?

 

I can certainly hear the benefits of the Dirac DSP processing in my room:

a) I had a problem with diffuse/recessed center imaging and that all gets cleared up when the filter is turned on. More focussed and forward.

b) bass is smoothed out.

I still have to decide if my concerns about bugs/reliability are worth the (high) cost of admission. We'll see.

[camott]

I know this isn't ideal, but you can use the AU plug-in with anything.

 

Use AU plugins for free with any Apple audio software - Blogs - Computer Audiophile

 

Interesting. But I would have to install Soundflower (+ DIRAC AU plugin) as a layer between Roon and output device, so I am essentially swapping out DAP for Soundflower and not sure I am gaining anything there

[camott]

You mean the stuff below 0.02? My guess is it is ambient noise (which is why it doesn't decay). That should be far below the threshold of hearing.

 

Thanks for the suggestion. I am not sure it's ambient noise unfortunately - I think I would also see it before the initial impact as well as after. I am trying to clarify with flavio what this noise is - possibly 'ringing' as he mentioned but not clear to me yet. I have seen other IR graphs and it looks like this noise/ringing/whatever decays on most of them, so concerned I have some serious echo issues in my room!