ttocs
Global Moderator 
I always have a wonderful time, wherever I am, whomever I'm with. (Elwood P Dowd)
Posts: 8,115
Member is Online
|
Post by ttocs on Jun 24, 2021 14:46:15 GMT -5
I suspect that Dirac donβt measure the post correction result because they donβt want customers to see target vs actual corrected response. As was pointed out, REW and Dirac donβt measure the same things so there is no way to know if making manual tweaks to the Dirac filters will make objective improvements. We are supposed to just trust the result. Iβm going to continue to use Dirac filters . On my system both before and after Dirac sounds good, Dirac might sound better, but itβs difficult to get all levels matched. Dirac target and REW show discrepancies that are more than +/- 3 dB for me. So we can just enjoy the sound, whatever it is. I just wish there was a better way to know for sure that what we are hearing is faithful to what the music artist and producer intended. I don't believe it to be the case that Dirac doesn't want customers to see the "after". But I do believe that no matter how Dirac would go about being able to measure after the fact would prove to be problematic and deter from the point of their application, which is to measure and provide a correction filter. IOW it's a no win situation for them. It would be great if it had that capability, but it doesn't, and I'm happy with REW. Try this, it's pretty easy. Dirac measures the "before" and shows the result for each mic location. That's the starting point for Dirac. But the last mic position should be the MLP so the mic can be left there and used with REW so it's an apples comparison. During the Dirac run you can screenshot the before at the MLP for as many channels as you like, and also screenshot the predicted curves. Run Levels for each Preset so you know everything is zeroed out. I use REW for this because it's really easy to use its SPL meter. Now run REW sweeps and measure with the USER EQ on one Preset for the "before", and with Dirac filter on the other Preset for the "after". I'd be surprised if they are vastly different, because I've posted my results and they are freakishly similar. Multiple mic positions are not necessary for this. Dirac and REW measure differently. This is a very open ended statement, but this is my take on it. They measure differently, so what? Try not to overthink this difference thing, and don't use multiple mic positions for comparing the two. The only position that matters for a comparison is the MLP. One single mic position. The most important result is if it sounds good to you. Are you happy with what you hear? For me, measuring is a way of tracking things and figuring out what's going on. The tracking part comes into play whenever I change something. It tells me what changed sonically when I change something physically or otherwise. One thing to keep in mind is how our processors mismanage the Bass Management, so keep that in mind. When I adjust the subwoofers to be the same level as speakers, the bass is way too strong. Using REW I can adjust the subwoofers to make them fall in line with where they are supposed to be. I adjust the subwoofer channels down quite a bit to be lined up with the Small speakers above the XO. I'm going to have an REW day coming up soon so I'll post some results. |
|
|
|
Post by jbrunwa on Jun 24, 2021 17:03:57 GMT -5
I suspect that Dirac donβt measure the post correction result because they donβt want customers to see target vs actual corrected response. As was pointed out, REW and Dirac donβt measure the same things so there is no way to know if making manual tweaks to the Dirac filters will make objective improvements. We are supposed to just trust the result. Iβm going to continue to use Dirac filters . On my system both before and after Dirac sounds good, Dirac might sound better, but itβs difficult to get all levels matched. Dirac target and REW show discrepancies that are more than +/- 3 dB for me. So we can just enjoy the sound, whatever it is. I just wish there was a better way to know for sure that what we are hearing is faithful to what the music artist and producer intended. I don't believe it to be the case that Dirac doesn't want customers to see the "after". But I do believe that no matter how Dirac would go about being able to measure after the fact would prove to be problematic and deter from the point of their application, which is to measure and provide a correction filter. IOW it's a no win situation for them. It would be great if it had that capability, but it doesn't, and I'm happy with REW. Try this, it's pretty easy. Dirac measures the "before" and shows the result for each mic location. That's the starting point for Dirac. But the last mic position should be the MLP so the mic can be left there and used with REW so it's an apples comparison. During the Dirac run you can screenshot the before at the MLP for as many channels as you like, and also screenshot the predicted curves. Run Levels for each Preset so you know everything is zeroed out. I use REW for this because it's really easy to use its SPL meter. Now run REW sweeps and measure with the USER EQ on one Preset for the "before", and with Dirac filter on the other Preset for the "after". I'd be surprised if they are vastly different, because I've posted my results and they are freakishly similar. Multiple mic positions are not necessary for this. Dirac and REW measure differently. This is a very open ended statement, but this is my take on it. They measure differently, so what? Try not to overthink this difference thing, and don't use multiple mic positions for comparing the two. The only position that matters for a comparison is the MLP. One single mic position. The most important result is if it sounds good to you. Are you happy with what you hear? For me, measuring is a way of tracking things and figuring out what's going on. The tracking part comes into play whenever I change something. It tells me what changed sonically when I change something physically or otherwise. One thing to keep in mind is how our processors mismanage the Bass Management, so keep that in mind. When I adjust the subwoofers to be the same level as speakers, the bass is way too strong. Using REW I can adjust the subwoofers to make them fall in line with where they are supposed to be. I adjust the subwoofer channels down quite a bit to be lined up with the Small speakers above the XO. I'm going to have an REW day coming up soon so I'll post some results. Thanks, I will try your procedure. I am also seeing subwoofer higher SPL by quite a lot. |
|
ttocs
Global Moderator
I always have a wonderful time, wherever I am, whomever I'm with. (Elwood P Dowd)
Posts: 8,115
Member is Online
|
Post by ttocs on Jun 24, 2021 21:59:01 GMT -5
The levels I have referred to have nothing to do with Dirac.
When the levels are set, with or without Dirac, all channels are within a dB of each other.
Watching any 5.1, 7.1, ATMOS, etc, it is very evident that the bass is way too strong, too much, bloated. But Menu:: Setup:: Speakers:: Preset-1:: Levels shows that the levels are all the same.
Checking with REW shows the subwoofers to be 4-6dB too high below the XO vs above the XO of any channel setup as Small. So I use this to set the subwoofer levels to the correct level.
Again, Dirac is meaningless in this exercise. The same happens without Dirac ever being used. I've posted sweeps showing this a few months ago and reported to Emotiva who says they will work on it someday.
My workaround works pretty well, but not perfectly. I use Center Sub for LFE, and Left Sub for BM. This way I can set each channel correctly.
|
|
|
|
Post by geebo on Jun 25, 2021 6:52:42 GMT -5
It's firmware related. You can revert back to an older FW and the issue is gone. Change nothing and go back the the latest firmware and it's back. As ttocs does, I reduce both subs' levels by 6dB to get things back in line.
|
|
|
|
Post by jbrunwa on Jun 26, 2021 13:00:44 GMT -5
Ok. Here is what I see on my system. I attached what Dirac Live shows and REW measured at main listening position with 1/12 smoothing with (red) and without (blue) Dirac filter. Does it seem like the bass is elevated over what Dirac shows, or is this from averaging over 13 positions? |
|
|
|
Post by davidl81 on Jun 26, 2021 14:12:01 GMT -5
It's firmware related. You can revert back to an older FW and the issue is gone. Change nothing and go back the the latest firmware and it's back. As ttocs does, I reduce both subs' levels by 6dB to get things back in line. Ok. Here is what I see on my system. I attached what Dirac Live shows and REW measured at main listening position with 1/12 smoothing with (red) and without (blue) Dirac filter. Does it seem like the bass is elevated over what Dirac shows, or is this from averaging over 13 positions?Β This is consistent with what I also find. The bass does not seem elevated when doing the Dirac setup, but when listing to movies/music the elevated bass is there. When setting up REW what do you use for a sound source? Iβve never done it before |
|
|
|
Post by geebo on Jun 26, 2021 14:54:25 GMT -5
Ok. Here is what I see on my system. I attached what Dirac Live shows and REW measured at main listening position with 1/12 smoothing with (red) and without (blue) Dirac filter. Does it seem like the bass is elevated over what Dirac shows, or is this from averaging over 13 positions?Β This is consistent with what I also find. The bass does not seem elevated when doing the Dirac setup, but when listing to movies/music the elevated bass is there. When setting up REW what do you use for a sound source? Iβve never done it before REW has it's own built in customizable sweeps. You only have to connect the computer to the USB port on the back of the processor then select the USB input. |
|
|
|
Post by jbrunwa on Jun 26, 2021 15:07:29 GMT -5
This is consistent with what I also find. The bass does not seem elevated when doing the Dirac setup, but when listing to movies/music the elevated bass is there. When setting up REW what do you use for a sound source? Iβve never done it before REW has it's own built in customizable sweeps. You only have to connect the computer to the USB port on the back of the processor then select the USB input. You can also use HDMI if your laptop has it. I ran REW on a MacBook with audio output over HDMI and the calibrated umik USB for audio input. |
|
|
|
Post by davidl81 on Jun 26, 2021 15:42:19 GMT -5
This is consistent with what I also find. The bass does not seem elevated when doing the Dirac setup, but when listing to movies/music the elevated bass is there. When setting up REW what do you use for a sound source? Iβve never done it before REW has it's own built in customizable sweeps. You only have to connect the computer to the USB port on the back of the processor then select the USB input. Makes sense. Seems like whatever input the G3 processors are processing they amplify the low end. Somehow the Dirac sweeps are not getting the same boost when running the setup. So Dirac sees no need to reduce that bass since it never actually βhearsβ that bass. |
|
ttocs
Global Moderator
I always have a wonderful time, wherever I am, whomever I'm with. (Elwood P Dowd)
Posts: 8,115
Member is Online
|
Post by ttocs on Jun 26, 2021 16:24:45 GMT -5
REW has it's own built in customizable sweeps. You only have to connect the computer to the USB port on the back of the processor then select the USB input. Makes sense. Seems like whatever input the G3 processors are processing they amplify the low end. Somehow the Dirac sweeps are not getting the same boost when running the setup. So Dirac sees no need to reduce that bass since it never actually βhearsβ that bass. More to the point is that Dirac will never "see" how the subwoofer channels interact with the speaker channels while a sound field is applied. This is where REW can be used very effectively. With REW I can set the sound field to Surround, DD Surround, All Stereo, etc, and each of these sound fields behaves differently with regards to the level of bass vs speaker. Bass is split into Bass Management and LFE. With some sound fields the BM is 3dB higher than speaker level, and LFE can be 6dB higher than the +10dB that it's supposed to be for a total of +16dB. With other sound fields the LFE can be low, and BM can be at +11dB. It all comes down to speaker setup, if any channels are setup as Large, if Center Sub is active and how it is set, and if Left Sub is used either with or without Right Sub. Sound complicated? This is why Emo hasn't touched it yet. I've got lots of data on this stuff, all of which I sent to Emo, which is why I've been able to work out "most" of the kinks. |
|
|
|
Post by jbrunwa on Jun 26, 2021 18:00:40 GMT -5
Makes sense. Seems like whatever input the G3 processors are processing they amplify the low end. Somehow the Dirac sweeps are not getting the same boost when running the setup. So Dirac sees no need to reduce that bass since it never actually βhearsβ that bass. More to the point is that Dirac will never "see" how the subwoofer channels interact with the speaker channels while a sound field is applied. This is where REW can be used very effectively. With REW I can set the sound field to Surround, DD Surround, All Stereo, etc, and each of these sound fields behaves differently with regards to the level of bass vs speaker. Bass is split into Bass Management and LFE. With some sound fields the BM is 3dB higher than speaker level, and LFE can be 6dB higher than the +10dB that it's supposed to be for a total of +16dB. With other sound fields the LFE can be low, and BM can be at +11dB. It all comes down to speaker setup, if any channels are setup as Large, if Center Sub is active and how it is set, and if Left Sub is used either with or without Right Sub. Sound complicated? This is why Emo hasn't touched it yet. I've got lots of data on this stuff, all of which I sent to Emo, which is why I've been able to work out "most" of the kinks. I am confused about the Dirac Live implementation in the Emotiva processors. Maybe I am doing something to cause the elevated bass. When running Dirac Live, it is my understanding that the processor speaker size, crossover frequency and slope settings are ignored. Without the Bass Control option, does Dirac Live sum speaker and sub response when generating filters against the target curve, generate optimum crossovers or slopes between main and subs. Or are the main speakers effectively large size other than any manually added control points in Dirac that would create a high-pass or low-pass filter? In the Emotiva processor, when EQ is set to a Dirac filter, does the processor use the speaker size, crossover frequency and slope settings? Or are these settings only active when EQ is set to User? |
|
|
|
Post by geebo on Jun 26, 2021 18:08:58 GMT -5
REW has it's own built in customizable sweeps. You only have to connect the computer to the USB port on the back of the processor then select the USB input. Makes sense. Seems like whatever input the G3 processors are processing they amplify the low end. Somehow the Dirac sweeps are not getting the same boost when running the setup. So Dirac sees no need to reduce that bass since it never actually βhearsβ that bass. And it doesn't get reflected in the Emotiva Setup - Speaker Preset - Levels. |
|
ttocs
Global Moderator
I always have a wonderful time, wherever I am, whomever I'm with. (Elwood P Dowd)
Posts: 8,115
Member is Online
|
Post by ttocs on Jun 26, 2021 18:15:43 GMT -5
More to the point is that Dirac will never "see" how the subwoofer channels interact with the speaker channels while a sound field is applied. This is where REW can be used very effectively. With REW I can set the sound field to Surround, DD Surround, All Stereo, etc, and each of these sound fields behaves differently with regards to the level of bass vs speaker. Bass is split into Bass Management and LFE. With some sound fields the BM is 3dB higher than speaker level, and LFE can be 6dB higher than the +10dB that it's supposed to be for a total of +16dB. With other sound fields the LFE can be low, and BM can be at +11dB. It all comes down to speaker setup, if any channels are setup as Large, if Center Sub is active and how it is set, and if Left Sub is used either with or without Right Sub. Sound complicated? This is why Emo hasn't touched it yet. I've got lots of data on this stuff, all of which I sent to Emo, which is why I've been able to work out "most" of the kinks.
1. When running Dirac Live, it is my understanding that the processor speaker size, crossover frequency and slope settings are ignored. 2a. Without the Bass Control option, does Dirac Live sum speaker and sub response when generating filters against the target curve, generate optimum crossovers or slopes between main and subs?. 2b. Or are the main speakers effectively large size other than any manually added control points in Dirac that would create a high-pass or low-pass filter? 3. In the Emotiva processor, when EQ is set to a Dirac filter, does the processor use the speaker size, crossover frequency and slope settings? Or are these settings only active when EQ is set to User?1. Correct. 2a. Nope, not a chance. 2b. Dirac measures all channels full range and makes filters for the range within the Curtains, which are user adjustable. Nothing more. This is why I drag the upper Curtain for the Subwoofer channels down to within the range of my subs. I don't want "loose" frequencies out there that might do something (I don't know what, I just don't like the idea) if the Curtain is not restricting the upper frequencies. 3. Yes. Whatever is "visible" in the Menu when any EQ setting is the active one is what is adjustable. If you don't see it, then you can't adjust it, no matter if a Dirac filter is the active EQ or User is the active EQ. |
|
|
|
Post by geebo on Jun 26, 2021 18:26:39 GMT -5
More to the point is that Dirac will never "see" how the subwoofer channels interact with the speaker channels while a sound field is applied. This is where REW can be used very effectively. With REW I can set the sound field to Surround, DD Surround, All Stereo, etc, and each of these sound fields behaves differently with regards to the level of bass vs speaker. Bass is split into Bass Management and LFE. With some sound fields the BM is 3dB higher than speaker level, and LFE can be 6dB higher than the +10dB that it's supposed to be for a total of +16dB. With other sound fields the LFE can be low, and BM can be at +11dB. It all comes down to speaker setup, if any channels are setup as Large, if Center Sub is active and how it is set, and if Left Sub is used either with or without Right Sub. Sound complicated? This is why Emo hasn't touched it yet. I've got lots of data on this stuff, all of which I sent to Emo, which is why I've been able to work out "most" of the kinks. I am confused about the Dirac Live implementation in the Emotiva processors. Maybe I am doing something to cause the elevated bass. When running Dirac Live, it is my understanding that the processor speaker size, crossover frequency and slope settings are ignored. Without the Bass Control option, does Dirac Live sum speaker and sub response when generating filters against the target curve, generate optimum crossovers or slopes between main and subs. Or are the main speakers effectively large size other than any manually added control points in Dirac that would create a high-pass or low-pass filter? In the Emotiva processor, when EQ is set to a Dirac filter, does the processor use the speaker size, crossover frequency and slope settings? Or are these settings only active when EQ is set to User? Dirac only adjusts the speakers frequency response (and phase/delay) and nothing else. It attempts to match the chosen target curve for each speaker individually whether you assign a separate target curve to each speaker or a target curve to a group of speakers. The crossovers and speakers sizes are set by the user in the processor setup. |
|
|
|
Post by jbrunwa on Jun 26, 2021 19:11:25 GMT -5
I am confused about the Dirac Live implementation in the Emotiva processors. Maybe I am doing something to cause the elevated bass.
1. When running Dirac Live, it is my understanding that the processor speaker size, crossover frequency and slope settings are ignored. 2a. Without the Bass Control option, does Dirac Live sum speaker and sub response when generating filters against the target curve, generate optimum crossovers or slopes between main and subs?. 2b. Or are the main speakers effectively large size other than any manually added control points in Dirac that would create a high-pass or low-pass filter? 3. In the Emotiva processor, when EQ is set to a Dirac filter, does the processor use the speaker size, crossover frequency and slope settings? Or are these settings only active when EQ is set to User? 1. Correct. 2a. Nope, not a chance. 2b. Dirac measures all channels full range and makes filters for the range within the Curtains, which are user adjustable. Nothing more. This is why I drag the upper Curtain for the Subwoofer channels down to within the range of my subs. I don't want "loose" frequencies out there that might do something (I don't know what, I just don't like the idea) if the Curtain is not restricting the upper frequencies. 3. Yes. Whatever is "visible" in the Menu when any EQ setting is the active one is what is adjustable. If you don't see it, then you can't adjust it, no matter if a Dirac filter is the active EQ or User is the active EQ. Thanks for this and thanks to all who helped me. So I improved the response by adding a control point in Dirac to lower the sub around 30 Hz, exported the filter and then in the XMC-2 settings set the fronts to small, 80 Hz crossover, 24 dB slope, and lowered the sub level. I want to see if I can lower the peak around 50 Hz but I'm pretty happy with the result, so thanks everyone. Although Iβm happy to get this sorted out, Iβm not happy with Dirac Live. In my opinion itβs not a complete product without bass management, an extra feature for more money and only available on a few processors. I will keep looking for something better. Attachments:
|
|
|
|
Post by marcl on Jun 27, 2021 13:34:06 GMT -5
As ttocs says, making the MLP measurement last and leaving the microphone in place so that the REW measurement is as apples:apples as possible is a very valid test. Even though Dirac and REW may use different measurement windows, etc. you can still look at the Dirac MLP "before" and compare it to the REW MLP "after". Note also that Dirac uses the MLP measurement to set distances/delays and to do the impulse correction. You can verify both in REW. This probably won't happen for most systems, but I have had Dirac misinterpret a measurement and flip polarity on one speaker causing bass to be greatly reduced. You can see this in the REW phase and impulse plots.
Also as some have said, using REW lets you measure what Dirac does not ... the response of a small speaker with bass management. I often have had to tweak a crossover point so that the transition is smooth.
As for Dirac measuring the "after" ... who knows why they don't. Has anyone used an Antimode 8033 for sub EQ? It does at least six sweeps and iteratively corrects the response. Maybe Dirac would benefit from at least taking a second pass at the MLP and do one iteration on the filters.
|
|
|
|
Post by jbrunwa on Jun 27, 2021 16:29:58 GMT -5
As ttocs says, making the MLP measurement last and leaving the microphone in place so that the REW measurement is as apples:apples as possible is a very valid test. Even though Dirac and REW may use different measurement windows, etc. you can still look at the Dirac MLP "before" and compare it to the REW MLP "after". Note also that Dirac uses the MLP measurement to set distances/delays and to do the impulse correction. You can verify both in REW. This probably won't happen for most systems, but I have had Dirac misinterpret a measurement and flip polarity on one speaker causing bass to be greatly reduced. You can see this in the REW phase and impulse plots. Also as some have said, using REW lets you measure what Dirac does not ... the response of a small speaker with bass management. I often have had to tweak a crossover point so that the transition is smooth. As for Dirac measuring the "after" ... who knows why they don't. Has anyone used an Antimode 8033 for sub EQ? It does at least six sweeps and iteratively corrects the response. Maybe Dirac would benefit from at least taking a second pass at the MLP and do one iteration on the filters. Well, itβs still possible, maybe even likely I am doing something wrong. Dirac Live does not bring down the peaks on my subs even though REW shows their overall levels are very elevated. I have made some improvement by adding control points but Iβm not sure I am adept enough to get it right. I do have a Behringer DCX2496 left over from my old system. I almost feel more comfortable defining PEQ than trying to get a graphical curve to do what I need. I also suppose that I could try that freeware Multi-Sub Optimizer to insert some filters between the XMC-2 and subs to flatten the response. Itβs a shame that such quality sound equipment doesnβt come with quality tools to make bass management easier. |
|
|
|
Post by adam631 on Jun 27, 2021 21:23:28 GMT -5
With no more green zone in the latest Dirac Live, how loud do i need to make the levels?
|
|
|
|
Post by marcl on Jun 28, 2021 7:00:37 GMT -5
With no more green zone in the latest Dirac Live, how loud do i need to make the levels? Speaker efficiency varies so much that it's hard to say. And the level you hear with the noise and tones during volume calibration is not the level you hear during actual measurements. Here's what works in my 7.1.4 system where the side surrounds are least efficient and top rears are most efficient: - Leave mic at 100%
- Play subwoofer tones and raise gain control until sub sweeps peak at -20db. Subwoofer slider stays all the way up
- Pull sliders for top rears all the way down; leave sliders for side surrounds all the way up; verify levels for both are around -27db
- Using sliders for all other speakers, adjust their levels to -27db
In my system ... this results in a subwoofer measurement whose peak is at the same level as bass peaks for my large front speakers. The level of the sweep tones during actual measurement is no louder than 75db and there are no clipping warnings or rattles/distortion from speakers.
It's wise to be conservative with levels the first time to be sure they are not too loud. Dirac works over a wide range and it will warn you if levels are too high or low.
|
|
|
|
Post by AudioHTIT on Jun 28, 2021 9:55:17 GMT -5
Can anyone confirm that the latest Dirac Live (3.0.14) runs without issue on an M1 based Mac running macOS Big Sur 11.4? I can confirm this for an Intel based Mac mini 2018.
|
|
