• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

Audyssey XT32 vs Dirac Live

anotherhobby

Senior Member
Forum Donor
Joined
Dec 17, 2021
Messages
423
Likes
825
Well, at least it's supposedly better than EQing each individually and independently. But yeah, seems like there are much better ways to do it, but they require additional hardware. It might be possible to get reasonable results with a Y splitter and some DSP built-in to the sub for delay/phase and PEQ?

Anyway, looks like the Audyssey solution is a "better than nothing, but worse than any reasonable external solution" kind of thing.
Audyssey did not do very well for me with more than 1 sub for seating positions other than the primary. I was able to get better results for all seating locations by using MSO and a cheap $100 miniDSP. I posted about it here.
 

GalZohar

Senior Member
Joined
Aug 8, 2021
Messages
428
Likes
203
Audyssey did not do very well for me with more than 1 sub for seating positions other than the primary. I was able to get better results for all seating locations by using MSO and a cheap $100 miniDSP. I posted about it here.
Yeah seems like that's what most people are getting - Audyssey better than nothing, but worse than other solutions that handle multiple subwoofers and not just duplicate the signal.
 

peng

Major Contributor
Forum Donor
Joined
May 12, 2019
Messages
3,556
Likes
3,146
Audyssey did not do very well for me with more than 1 sub for seating positions other than the primary. I was able to get better results for all seating locations by using MSO and a cheap $100 miniDSP. I posted about it here.

Very interesting.. I don't have the issue at all, in fact I removed my minidsp to eliminate the extra wires because it hardly make any difference especially after I managed to get even better results with the App and Ratbuddyssey. So are you using XT32/Sub EQ HT? And how many subs?

May be mine is doing so well without using the minidsp/MSO because I only have two main subs for the front.
 

anotherhobby

Senior Member
Forum Donor
Joined
Dec 17, 2021
Messages
423
Likes
825
Very interesting.. I don't have the issue at all, in fact I removed my minidsp to eliminate the extra wires because it hardly make any difference especially after I managed to get even better results with the App and Ratbuddyssey. So are you using XT32/Sub EQ HT? And how many subs?

May be mine is doing so well without using the minidsp/MSO because I only have two main subs for the front.
I have XT32 with two JL Audio e112 subs, and they are located up front next to my mains. I am only using 1 of the sub outputs from my X4300H to the miniDSP.

Every room is different so you need to experiment and run sweeps to see works, although I think it'd be harder (you'd need to be lucky) to get better results for more seats with Audyssey on it's own with multiple subs vs. using MSO first in that situation.

I don't think MSO makes much of a difference for the primary seat, it's the rest of the seats that MSO helps. Also, getting good results from MSO was also not point and shoot for me. I had to try several different configurations and limiting various settings to get the result I wanted. Running it with default settings (following the tutorial) was not good at all, but I also have a very challenging space.
 

peng

Major Contributor
Forum Donor
Joined
May 12, 2019
Messages
3,556
Likes
3,146
I have XT32 with two JL Audio e112 subs, and they are located up front next to my mains. I am only using 1 of the sub outputs from my X4300H to the miniDSP.

Every room is different so you need to experiment and run sweeps to see works, although I think it'd be harder (you'd need to be lucky) to get better results for more seats with Audyssey on it's own with multiple subs vs. using MSO first in that situation.

I don't think MSO makes much of a difference for the primary seat, it's the rest of the seats that MSO helps. Also, getting good results from MSO was also not point and shoot for me. I had to try several different configurations and limiting various settings to get the result I wanted. Running it with default settings (following the tutorial) was not good at all, but I also have a very challenging space.

Understood, but in my case I have measured up to 10 positions, about 8 inches between one position to the next in both horizontal to vertical directions. Using 1/12 smoothing, I got +/- 1 to 1.5 dB peak to peak (averaged).

MSO is free so I may try it for fun on one of my two channel system that I don't have Audyssey to run.

Without EQ, it was horrible, so my room isn't great.
 

Dj7675

Major Contributor
Forum Donor
Joined
Jan 12, 2019
Messages
1,844
Likes
2,259
Understood, but in my case I have measured up to 10 positions, about 8 inches between one position to the next in both horizontal to vertical directions. Using 1/12 smoothing, I got +/- 1 to 1.5 dB peak to peak (averaged).

MSO is free so I may try it for fun on one of my two channel system that I don't have Audyssey to run.

Without EQ, it was horrible, so my room isn't great.
I have seen your measurements which clearly show a very flat response over the entire listening area ( I can't recall how many listening spots you have). Have you taken an MMM measurement (or averaged response) of just the head area of each listeners spot? I don't recall if you have posted that or not but it would be interesting.
I need to see if I can dig up my Audyssey results, but I had a situation where the overall response when I measured the entire area looked flat when averaged, but individual response at each seat were not. I think response measured at each listening position is much more interesting than averaged over a larger area. It seems very tricky to get good response at each seat.
Appologies if you have already shown that!
 

ernestcarl

Major Contributor
Joined
Sep 4, 2019
Messages
2,396
Likes
1,745
Location
Canada
response measured at each listening position is much more interesting than averaged over a larger area. It seems very tricky to get good response at each seat.

Tricky indeed... directivity control and positioning needs to be "good" for things to work well.

Professional calibrators will set multiple mics (at least four) throughout a room that has multiple listening positions.

One can try to emulate the same in one's own space....


Not at all moving (bass managed L+R output using steady microphone) RTA capture at four disparate listening positions and angles (relative to mains).
1647713259479.png

*not perfect, but quite good, IMO

Yellow trace is about 2.5 meters across the front of the couch MLP, right at the Desk area -- very much way off-axis my Sceptre S8 coaxial monitors which are directly to the sides.

The all-important mid frequency range is maintained relatively flat for the most part -- this holds true when vertically standing as well given the coaxially wave-guided horn design.

Right monitor sits precariously in the middle of the hallway to the right:
1647714074260.jpeg 1647713331058.png 1647713290183.jpeg 1647713304719.jpeg
 

peng

Major Contributor
Forum Donor
Joined
May 12, 2019
Messages
3,556
Likes
3,146
I have seen your measurements which clearly show a very flat response over the entire listening area ( I can't recall how many listening spots you have). Have you taken an MMM measurement (or averaged response) of just the head area of each listeners spot? I don't recall if you have posted that or not but it would be interesting.
I need to see if I can dig up my Audyssey results, but I had a situation where the overall response when I measured the entire area looked flat when averaged, but individual response at each seat were not. I think response measured at each listening position is much more interesting than averaged over a larger area. It seems very tricky to get good response at each seat.
Appologies if you have already shown that!

I did post the FRs for all 9 positions and like you said but it would be time consuming to locate the post, quicker to capture and post again. As you said, the individual positions (except the mmp) don't look flat at all, REW's average feature definitely make them look much flatter.

The important thing for me to watch was whether by flattening the MMP FR using the App/Ratbuddyssey would make non mmp positions worse. From what I can see so far, by tweaking to make mmp performed FR flatter, it also made other positions flatter. To me, that make logical sense.

1647716480798.jpeg


1647716760127.jpeg


1647717611731.jpeg
 

Dj7675

Major Contributor
Forum Donor
Joined
Jan 12, 2019
Messages
1,844
Likes
2,259
@peng, I need to take some time to understand your measurements more, but below are mine. I have 4 subs, corner loaded. Two 15 inch ported PSA TV1512 in the front corners and 2 S2410 in the back (dual opposing 12 inch sealed). I am using the StormAudio with DLBC I am using a +6dB curve from the very useful site site that @markus maintains (here) which is extremely helpful in getting things configured properly. I wanted to show the sub curve, the projected response, the averaged measurement of the entire listening area, and an average measurement of each listeners head space to get show the response at each listeners space. All the measurements are taken with MMM Method (not single point averaged). I don't know if using the MMM is best for measuring subs or if doing multiple sweeps and averaging them is better, but I think it should be fine for this purpose.
I am using REW for the measurements and measuring the LFE channel. Dirac applies a LPF at 120hz which is why the predicted LFE curve looks like it does. Not sure what smoothing is best, but I chose 1/12. I think I can get better with a little more attention to setup, but I think it shows...
-A good match of the room curve to the full listening position as well as a good match to the curve to each listening position which to me is the most important measurement. It seems the trickiest part is not the average over the entire listening area but the response at each position. I have 2 rows of seating and this is just the first row of the main 2 seats. Later I will try measurements of both rows and see if DLBC can maintain good response at both rows.

1.First this is the sub curve in green and the projected response
1647920687545.png


2. Uncorrected and Corrected Response of the full listening area. Disregard level differences.
ALL-LP Uncorrected.jpg

Corrected-All LP.jpg

2. Listening positing 1
LP1-Uncorrected.jpg

LP1 Corrected.jpg

3. Listening position 2
LP2-Uncorrected.jpg

lp2-corrected.jpg
 

peng

Major Contributor
Forum Donor
Joined
May 12, 2019
Messages
3,556
Likes
3,146
@peng, I need to take some time to understand your measurements more,

As you asked, the first graph was for the individual mic positions, not averaged. I measured 9 positions, the graph shows only 3 of the 9 positions that measured worse, in terms of flatness.

I can see that the legend's wording does look confusing. To clarify, the 3 red ones were with Audyssey On, and the black ones were Audyssey Off and the positions shown correspond to the following 3 positions (hence 6 curves):

- Position 4: 10 inches to the left of MMP
- Position 5: 10 inches to the left and 10 inches above MMP.
- Position 6: 10 inches to the left and 8 inches below MMP.

If I were to post all 9 positions in one graph, it would be hard to read as there would be a totally of 18 curves.

The 9 positions were:

1648044542920.png


If I included all 9 in the graph it would look too busy, the other 6 positions look flatter anyway. You can see that even for the 3 worst positions, by flattening the FR for the MMP, they all improved as well. I just tried the same using Dirac DL3 PC standalone version on a two channel system and found the same effects, that customizing the MMP curve did not make the other mic positions worse.

but below are mine. I have 4 subs, corner loaded. Two 15 inch ported PSA TV1512 in the front corners and 2 S2410 in the back (dual opposing 12 inch sealed). I am using the StormAudio with DLBC I am using a +6dB curve from the very useful site site that @markus maintains (here) which is extremely helpful in getting things configured properly. I wanted to show the sub curve, the projected response, the averaged measurement of the entire listening area, and an average measurement of each listeners head space to get show the response at each listeners space. All the measurements are taken with MMM Method (not single point averaged). I don't know if using the MMM is best for measuring subs or if doing multiple sweeps and averaging them is better, but I think it should be fine for this purpose.
I am using REW for the measurements and measuring the LFE channel. Dirac applies a LPF at 120hz which is why the predicted LFE curve looks like it does. Not sure what smoothing is best, but I chose 1/12. I think I can get better with a little more attention to setup, but I think it shows...
-A good match of the room curve to the full listening position as well as a good match to the curve to each listening position which to me is the most important measurement. It seems the trickiest part is not the average over the entire listening area but the response at each position. I have 2 rows of seating and this is just the first row of the main 2 seats. Later I will try measurements of both rows and see if DLBC can maintain good response at both rows.

1.First this is the sub curve in green and the projected response
View attachment 194271

2. Uncorrected and Corrected Response of the full listening area. Disregard level differences.
View attachment 194275
View attachment 194276
2. Listening positing 1
View attachment 194277
View attachment 194278
3. Listening position 2
View attachment 194279
View attachment 194280

Look great for various positions, but what smoothing?
 

Dj7675

Major Contributor
Forum Donor
Joined
Jan 12, 2019
Messages
1,844
Likes
2,259
As you asked, the first graph was for the individual mic positions, not averaged. I measured 9 positions, the graph shows only 3 of the 9 positions that measured worse, in terms of flatness.

I can see that the legend's wording does look confusing. To clarify, the 3 red ones were with Audyssey On, and the black ones were Audyssey Off and the positions shown correspond to the following 3 positions (hence 6 curves):

- Position 4: 10 inches to the left of MMP
- Position 5: 10 inches to the left and 10 inches above MMP.
- Position 6: 10 inches to the left and 8 inches below MMP.

If I were to post all 9 positions in one graph, it would be hard to read as there would be a totally of 18 curves.

The 9 positions were:

View attachment 194609

If I included all 9 in the graph it would look too busy, the other 6 positions look flatter anyway. You can see that even for the 3 worst positions, by flattening the FR for the MMP, they all improved as well. I just tried the same using Dirac DL3 PC standalone version on a two channel system and found the same effects, that customizing the MMP curve did not make the other mic positions worse.



Look great for various positions, but what smoothing?
Thanks for the explanation of your measurements, that helps. Have you taken a measurement of the listeners position? I think it is interesting to see what the room correction is doing at each of the the listeners position. To measure this, I did a MMM measurement of the approximate head space where each listener is. Have you tried anything like that yet?
I used 1/12 smoothing on my previous measurements but can certainly change it.
My next measurement will be sub+L/R/C to see how well DLBC blends the sub with the LCR. With Audyssey I had some issues there that I had to try to deal with after Audyssey.
 

peng

Major Contributor
Forum Donor
Joined
May 12, 2019
Messages
3,556
Likes
3,146
Thanks for the explanation of your measurements, that helps. Have you taken a measurement of the listeners position? I think it is interesting to see what the room correction is doing at each of the the listeners position. To measure this, I did a MMM measurement of the approximate head space where each listener is. Have you tried anything like that yet?
I used 1/12 smoothing on my previous measurements but can certainly change it.
My next measurement will be sub+L/R/C to see how well DLBC blends the sub with the LCR. With Audyssey I had some issues there that I had to try to deal with after Audyssey.

I have tried that many times in the long past but stopped doing it since about two years ago, especially since Audyssey's App+Ratbuddyssey already show the measurements though that would be with Audyssey Off only. DL3 does the same kind of things. I only have 3 main seats in my HT room as the other 2 are not usable for music/movie at all.

Below is an example of the 3 main mic positions/seats with Audyssey Off. You can tell the main seat (black) happened to look not too bad but not great, luckily. The one to the right looked horrific (blue).

1648048826771.png





How about Trinnov, don't they show the measurements for each mic position/seats also, perhaps just projected? Or they do, but you still want to verify with REW? For verification purpose, it is good to do it but I think just once would be enough.
 

luft262

Senior Member
Forum Donor
Joined
Mar 25, 2021
Messages
399
Likes
187
Location
Phoenix
Based on the opinions and comments I've read in some threads here recently there seem to be people with very strong negative opinions about the sound that Dirac Live produces. I have an Onkyo TX-RZ50 on order that I can still cancel if I wanted to go with a Denon X3700 instead. Did I make the wrong choice by choosing the receiver with Dirac?
Dirac is great, but XT32 is also very good. The X3700H has a proven DAC and AMP so I would go with that, but it's up to you and your preferences.
 

chych7

Active Member
Joined
Aug 28, 2020
Messages
206
Likes
329
My own study of XT32 vs. Dirac Live (PC trial version) for stereo listening concluded that I prefer the XT32 sound for full range EQ, and do not like what Dirac Live does above 500 Hz. I use a miniDSP/MSO for subwoofer integration, so that wasn't part of the study.

Relevant posts:



And given that the X3700 has much better amplification than the RZ50, I'd go with the Denon (+miniDSP and MSO to really integrate subwoofers well).
 

hmt

Senior Member
Joined
Apr 11, 2020
Messages
334
Likes
420
To me it sounds like you did not use the same target curve for dirac and XT32. Not possible to infer anything useful then.
 

chych7

Active Member
Joined
Aug 28, 2020
Messages
206
Likes
329
To me it sounds like you did not use the same target curve for dirac and XT32. Not possible to infer anything useful then.
I most certainly attempted to use the same target curve, read the post in more detail. Main difference is an upshift in Audyssey from 4 kHz, which I suspect is coming from the microphone, but the targets were initially the same (-1 dB/dec Harman-like target). Dirac ended up being a few dB lower in volume; when comparing subjectively, this didn't matter because it was a comparison of room correction on vs. off, not directly between Audyssey and Dirac. At that point, I did try and match the FR to each other for the upper freqs. There's only so much you can do, to do a fair comparison.

Main conclusion I came up with is that Dirac is way overrated. Many people cut it off at ~500 Hz because it sounds better, Amir mentioned he does that too. This applies to the curve I put for Dirac in my testing, as well as the default Dirac curve. IMO Audyssey XT32 sounds better on than off full range, in my system, after setting a Harman-like target curve (instead of the default cinema X-curve which makes it sound too bright/recessed mids).

Anyhow try out the Dirac Live free trial for yourself and see what you get.
 
Last edited:

HarmonicTHD

Major Contributor
Forum Donor
Joined
Mar 18, 2022
Messages
1,694
Likes
2,069
I most certainly attempted to use the same target curve, read the post in more detail. Main difference is an upshift in Audyssey from 4 kHz, which I suspect is coming from the microphone, but the targets were initially the same (-1 dB/dec Harman-like target). Dirac ended up being a few dB lower in volume; when comparing subjectively, this didn't matter because it was a comparison of room correction on vs. off, not directly between Audyssey and Dirac. At that point, I did try and match the FR to each other for the upper freqs. There's only so much you can do, to do a fair comparison.

Main conclusion I came up with is that Dirac is way overrated. Many people cut it off at ~500 Hz because it sounds better, Amir mentioned he does that too. This applies to the curve I put for Dirac in my testing, as well as the default Dirac curve. IMO Audyssey XT32 sounds better on than off full range, in my system, after setting a Harman-like target curve (instead of the default cinema X-curve which makes it sound too bright/recessed mids).

Anyhow try out the Dirac Live free trial for yourself and see what you get.
Hi.
Next week I will do a new XT32 (Denon X3700 pre-out into Neurochrome Mod686 monoblocks) sweep as we have rearranged our listening room.
So far I have used the Audyssey app and also Ratbudyessey also the App did a good job which I confirmed with REW.

Question. Could you describe the preference curve in more detail please so I can try it next week?

So far I used the so-called Reference curve. Is that the X-Curve or does it aim for flat response? I don’t remember atm.

Thx.
 

peng

Major Contributor
Forum Donor
Joined
May 12, 2019
Messages
3,556
Likes
3,146
I most certainly attempted to use the same target curve, read the post in more detail. Main difference is an upshift in Audyssey from 4 kHz, which I suspect is coming from the microphone, but the targets were initially the same (-1 dB/dec Harman-like target). Dirac ended up being a few dB lower in volume; when comparing subjectively, this didn't matter because it was a comparison of room correction on vs. off, not directly between Audyssey and Dirac. At that point, I did try and match the FR to each other for the upper freqs. There's only so much you can do, to do a fair comparison.

Main conclusion I came up with is that Dirac is way overrated. Many people cut it off at ~500 Hz because it sounds better, Amir mentioned he does that too. This applies to the curve I put for Dirac in my testing, as well as the default Dirac curve. IMO Audyssey XT32 sounds better on than off full range, in my system, after setting a Harman-like target curve (instead of the default cinema X-curve which makes it sound too bright/recessed mids).

Anyhow try out the Dirac Live free trial for yourself and see what you get.

Thank you for sharing your results. I have been testing DL3 PC version too but have not compare it with Audyssey as I have been using it for my 3 two channel systems. When I beta tested DL about two years ago I found Audyssey did slightly better in the below 200 Hz range and DL's impulse response looked slightly cleaner. The current version DL3 definitely looked better than the Beta one, no surprise there I guess.

Before I attempt to do a listening test comparison of DL3 vs Audyssey XT32 in my HT setup I need to figure out a way to do it to avoid the long delay time between switching the two. I am think of using a laptop with a dac and use analog input. When using DL, I would then set Audyssey to L/R bypass, and then when using XT32, I would have to select Reference and turn the DL3 filter off. To all that, it will take a while, and I highly doubt I would be able to hear a difference with any more than a few seconds. So I am curious to know when you did your listening tests between the two REQ system, how did you do the switching and what would be your estimated time delay changing from one to the other?

My Beta test was done long time ago but I managed to find the following graphs (plotted a lot more though), I was impressed enough with DL even with less than stellar results at the time, realizing it was a beta version and given the limited time to learn how to use it, in between bug fixes.

DL was beta version only:

1648131404072.jpeg



1648131427026.jpeg
1648131446758.jpeg
 

chych7

Active Member
Joined
Aug 28, 2020
Messages
206
Likes
329
So I am curious to know when you did your listening tests between the two REQ system, how did you do the switching and what would be your estimated time delay changing from one to the other?

Yes I had exactly this issue, it would take too long to switch between the two REQs. So instead my comparison was REQ on vs. off, instead of Dirac vs. Audyssey directly. From there I could do a relative comparison to no correction, where I found that I prefer Dirac to be off and Audyssey to be on, for the mid-high freqs. The bass freqs improved for both cases vs. off.
 

chych7

Active Member
Joined
Aug 28, 2020
Messages
206
Likes
329
Hi.
Next week I will do a new XT32 (Denon X3700 pre-out into Neurochrome Mod686 monoblocks) sweep as we have rearranged our listening room.
So far I have used the Audyssey app and also Ratbudyessey also the App did a good job which I confirmed with REW.

Question. Could you describe the preference curve in more detail please so I can try it next week?

So far I used the so-called Reference curve. Is that the X-Curve or does it aim for flat response? I don’t remember atm.

Thx.

No the default Audyssey curves will try and make the response flat at the listening location, with some high freq roll-off for the reference mode. In my system I don't see that high freq roll off and it sounds bright (my Audyssey mic may not be correctly calibrated for high freqs, I'm not entirely sure why I don't see it), and the mids/upper bass get recessed. The target curve I use is just a -1 dB/dec target, like in all of the plots Amir shows for speaker reviews, red arrow in this example. So for example I would put in something like:

100 Hz: +2.5 dB
1 kHz: 0 dB
10 kHz: -2.5 dB

And use REW/UMIK1 to verify the results and adjust it as needed. Actually I now just use MultEQ-X and just put in a -1 dB/dec tilt, makes it easier (but I also keep the additional high freq roll-off, to compensate whatever my mic seems to be doing incorrectly).
 
Top Bottom