|
|
Post by igorzep on Jun 5, 2015 7:19:22 GMT -5
I addressed one or two of these items in another post - and I'll cover one or two more here. Unfortunately you haven't clarified it above, just the topic was slowly substituted from Audio to Video settings..  3) The way Dirac "integrates with bass management" is quite simple - it doesn't. Thank you for honest answer... But... They are both IN ONE UNIT! How it come to the point that they are not supposed/designed to work together? All this deserves a side note in the feature list under Dirac Live, something like "*only for use with full-range speakers and audio only content; not designed for use together with Bass Management, and not designed for use with video content, if you decide to use that - do it for your own risk"... Playing the hit and miss game. Even Audyssey is integrated with bass management (how reliably it works considering a lack of user-controllable parameters is another issue, but in Pro version it have all necessary tools and in default version it equalizes satellites to the slope that is expected by Bass Management of the receiver, unfortunately only to one frequency that it decides is a natural speaker roll-off). Dirac then does its thing to try and match the frequency response of each speaker to the appropriate Target Curve and, at the same time, to compensate for anomalies in phase response and timing that it has detected. (The details of how it does that are part of Dirac's proprietary "secret sauce". There is no need to know how. There is only a need to know what are the target aims. And this is pretty natural and common thing - their target is minimum phase response. So, the target phase response is completely defined by the amplitude response. And amplitude response of the target is (by default / 'automated') - 'EQ flat as low and as high as it can then drop sharply'. This is all that is needed to know for proper integration and actually very convenient for 'perfect' bass management filter alignments (assuming no rolloff from speakers), which XMC-1 doesn't have (instead assuming a specific rolloff from speakers which Dirac does not provide, at least by default, and have no convenient tools to enforce the expected roll off to the target curve). They have a few white papers on their site about what Dirac Live does and doesn't do - and how it decides what errors to correct and what ones not to.) Great papers, I've read them. But they are again - not related to the topic I am asking. 4) The asymmetrical setting in our crossover does assume that the speaker itself will be exhibiting a natural rolloff. However, the reality is that most speakers produced these days don't specifically roll off at the THX recommended frequency, or necessarily at the expected slope. This really stops to matter once speakers are equalised. You can equalize them to any target response you like, including the desired roll off, and there is no practical difference (in terms of linear response) if this roll off comes from speaker's acoustical properties or from the EQ target. Even if the physical unequalised speakers requires specific BM curves and exactly one crossover setting is valid (with some hope) for them does not mean an equalized one should necessarily have the same flexibility limitations. Dirac Live for Emotiva is intended as a more or less automated room correction solution. The LE version is calibrated to work with the microphone we provide; the Full version will work with different microphones, and allows you to adjust the Target Curve and frequency limits, but still doesn't allow you to "get inside the process" much beyond that. Likewise, the bass management options we provide are intended to be relatively simple to use and set up. (The filter types we offer are "preprogrammed" into the high-end audio DSP we use - we didn't write them.) I know you are using BM from the chip vendor, but, what about "We control 100% of what's going on now"... May be it's time to stop pointing to each-other (Dirac vs Emotiva), but sit together and make some sensible strategy, not just throw it in and hope it works? With PEQ and Loudness features... are they too 'preprogrammed' into the DSP you use? If you want a lot more measurement flexibility, and wish to do a significant degree of customization, then you should consider using the extensive manual adjustment and EQ capabilities of the XMC-1. You might also consider using REW instead of Dirac. I am talking about the basics here, not about a 'significant degree of customization'... Adding one biquad in the chain to pass the 'perfect' response done by Dirac through BM also close to 'perfect' should not be too costly (as long as you have the advertised '100% control'). I know XMC-1 have a comparatively powerful PEQ capabilities, but calibrating 5 speakers individually... sorry... I don't want to spent my life doing it. REW does something but it is far from sophisticated frequency response calibration tool. It is a good tool to analyze room problems though. |
|
cawgijoe
Emo VIPs 
"When you come to a fork in the road, take it." - Yogi Berra
Posts: 5,035
|
Post by cawgijoe on Jun 5, 2015 9:02:11 GMT -5
You keep saying that Keith does not answer your question. it appears that he is really trying, but there is some mis-communication.
I have tried to read through all of this, but it's giving me a headache.
Can you summarize what your main question is?
Thank you.
|
|
|
|
Post by igorzep on Jun 5, 2015 10:02:30 GMT -5
You keep saying that Keith does not answer your question. it appears that he is really trying, but there is some mis-communication. I have tried to read through all of this, but it's giving me a headache. Can you summarize what your main question is? Thank you. There are three of them here, I'll try to summarize each in one sentence. 1) Missing configuration option for correcting an audio delay coming from how far from the listener speakers are placed (absolute one, so in relation to the video, not in relation between speakers themselves). 2) SPL level calibration so user should not re-adjust levels after every Dirac curve loading (relatively minor one, but taking some time from the user)... At least if they were not reset each time this would be much more convenient already. 3) Dirac ('default') assumptions and XMC-1 Bass Management assumptions in relation to amplitude and phase response are incompatible with each other... |
|
KeithL
Administrator 
Posts: 10,273
|
Post by KeithL on Jun 5, 2015 11:32:00 GMT -5
OK, here goes.....
Dirac Live is a software based room measurement and correction system. We LICENSE it from Dirac, and we've integrated it into the XMC-1. (In practical terms this means that, while the original Dirac Live used software modules both to take the measurements and calculate the filters, and to actually "run" the filters, we have a firmware module in the XMC-1 that runs the filters in hardware instead. While we've obviously worked very closely with Dirac to do this, Dirac Live is still actually a separate product, and we don't have unlimited options to redesign it. Therefore, Dirac's correction process and our bass management process remain separate.
As for the Lip Synch settings, perhaps I need to be a bit more direct with my answer. The VAST majority of people use the Auto setting for Lip Synch and find it to work very well. The manual Speaker Presets have an option to manually enter speaker distances (since they're not measured automatically); the Dirac preset doesn't include this option because it DOES measure the distances automatically). Since Dirac measures those distances automatically, and measuring them using the actual acoustic delay is more accurate in a practical sense than doing it by ruler anyway, we simply didn't see a need to include the option to manually tweak them.
You seem to be under a misconception as to how Dirac provides equalization. Dirac does NOT "EQ flat as low and as high as it can then drop sharply". What it does is "EQ flat as low and as high as it can then IGNORE EVERYTHING OUTSIDE THE FREQUENCY LIMITS". Dirac does NOT impose any sort of roloff outside the range of frequencies it is correcting - rather it simply doesn't EQ them and leaves them exactly as they were. (You'll usually see a roll off past the cutoff points because the cutoff points were chosen to be the points at which your speaker is starting to roll off. If you manually move the cutoff curtains - which you can on the Full version - you may end up with your speaker's response rising, or remaining flat, outside the range which Dirac corrects; in either case Dirac simply leaves the response outside the limits as it was.)
However, I should also make it perfectly clear that Dirac Live is designed primarily to do its best to produce a reasonable overall frequency and phase response; it is NOT designed to allow the user to create electronic crossovers. One consequence of this is that Dirac Live on the XMC-1 is limited to applying boosts of up to 10 dB and attenuation up to -20 dB. So, for example, the Full version of Dirac Live will indeed let you create a Target Curve that rolls your speaker off below 80 Hz at 12 dB/octave, it can only do so until it hits the maximum attenuation limit of 20 dB. (In other words, it works like what it is - something intended to provide a reasonable amount of EQ, but not as an electronic crossover filter.) Since this is what Dirac was designed to do, and why we included it in the XMC-1, we don't see this as a limitation or a problem.
We do not suggest that you recalibrate the absolute loudness level "every time after running Dirac". The main purpose of setting the speaker levels is to adjust all of your speakers to play at the same level. We don't consider it essential that anything be adjusted to any specific absolute level.
In "the old days" (twenty or so years ago) "a DSP" was a more or less general-purpose computer for which each user had to write their own code (the manufacturers typically provided "programming kits" to help you do this). However, as the required features became more and more complex, this model simply became unworkable. Therefore, modern DSP chips contain a large set of "building blocks" which the designer can then use to create the functionality they need. There are a very few companies producing these DSP chips, and they have chosen and included the features that they deem useful and necessary. So, for example, we can choose between any one of several filter modules that is available in the DSP we use, and we can set the parameters for that module - within the limits stated, but we can't redesign the module itself... and asking them to produce a custom version of that DSP chip just for us would be a project that starts in the millions of dollars. (So, for example, our programmers created the interface that allows you to use the parametric EQs, and we decided on the frequency and gain limits for them, but the actual PEQ code module we use is the one that is included in the DSP we use.... so we can't rewrite the actual code that executes the filters themselves. So, to take your example, if the filter code in the DSP offered the option of "another biquad" which could be chained when invoked, we could enable and use it, but we can't rewrite the filter code itself.)
As with many things, terms like "control" have rather flexible meanings. I would say that we do indeed have 100% control over every single line of code in the XMC-1. We chose the particular DSP chip we use because it includes all the features we require, and allows us exactly the level of individual control over each which we require. (We CHOSE to use a DSP that provides built-in filters, based on the fact that it provides the filters and options we need, and accepting that our choice limits our ability to rewrite them to suit ourselves, rather than pay a few million dollars to have one custom written for us - which would have given us the ability to rewrite each filter as we liked. We made this choice because we considered the filters provided to be exactly what we were looking for, and because the extra cost of "reinventing the wheel" would have forced us to charge you several thousand dollars more for your XMC-1.) If you really need the ability to choose different types of filters, then there are several dedicated electronic crossover devices available that give you that capability - but I'm not going to apologize that the XMC-1's bass management capabilities aren't quite as extensive as a dedicated active crossover device.
Incidentally, if you really don't want to, you DON'T have to calibrate each channel separately. The XMC-1 can export and import all of the PEQ settings in the Speaker Presets. This means that you CAN simply calibrate one channel, using REW or the controls in the XMC-1, save those settings to a file, then re-import that information and apply it to the remaining channels. You can then use it as is, or use it as a starting point for the calibration of those other channels. "Repurposing" the configuration saved from one channel for another is a simple matter of copying and renaming a few files, and changing the text in a few plain text fields that identify which channel it goes with.
|
|
tubby
Emo VIPs
Route 2 in Weekapaug!!!
Posts: 408
|
Post by tubby on Jun 5, 2015 12:10:03 GMT -5
You keep saying that Keith does not answer your question. it appears that he is really trying, but there is some mis-communication. I have tried to read through all of this, but it's giving me a headache. Can you summarize what your main question is? Thank you. There are three of them here, I'll try to summarize each in one sentence. 1) Missing configuration option for correcting an audio delay coming from how far from the listener speakers are placed (absolute one, so in relation to the video, not in relation between speakers themselves). 2) SPL level calibration so user should not re-adjust levels after every Dirac curve loading (relatively minor one, but taking some time from the user)... At least if they were not reset each time this would be much more convenient already. 3) Dirac ('default') assumptions and XMC-1 Bass Management assumptions in relation to amplitude and phase response are incompatible with each other... For #1. Where is this setting in a entry level avr as you suggest. I have never seen it. |
|
Deleted
Deleted Member
Posts: 0
|
Post by Deleted on Jun 5, 2015 12:24:17 GMT -5
While I dont like it (but have decided to work around it by using a minidsp) the bass management and intesgration with multiple subwoofers is in fact a DIRAC thing...the minidsp 88A with DL in it also needs to have the subs EQ'd as one...and time aligned seperatley
so its not an XMC problem
|
|
|
|
Post by sandiway on Jun 5, 2015 13:00:22 GMT -5
You keep saying that Keith does not answer your question. it appears that he is really trying, but there is some mis-communication. I have tried to read through all of this, but it's giving me a headache. Can you summarize what your main question is? ... 3) Dirac ('default') assumptions and XMC-1 Bass Management assumptions in relation to amplitude and phase response are incompatible with each other... If Dirac Live on the XMC-1 is a separate "black box" component inside the box that is the XMC-1, is there any advantage of the XMC-1 over a separate box solution like the MiniDSP nanoAVR DL Dirac Live, where clearly no bass management is done? |
|
|
|
Post by igorzep on Jun 5, 2015 13:13:46 GMT -5
For #1. Where is this setting in a entry level avr as you suggest. I have never seen it. It is called 'Distance' and you can find it in Speaker settings. I've opened a manual of the first entry level AVR model I remember - Onkyo TX-SR313 and it is described on the page 41. |
|
|
|
Post by monkumonku on Jun 5, 2015 13:38:01 GMT -5
For #1. Where is this setting in a entry level avr as you suggest. I have never seen it. It is called 'Distance' and you can find it in Speaker settings. I've opened a manual of the first entry level AVR model I remember - Onkyo TX-SR313 and it is described on the page 41. Every modern day AVR and processor I am aware of has this feature. As it is described on page 41 where you referenced, it is simply a manual adjustment. Or now it is often done automatically during setup by whatever program the processor has. Why should we need more than that? |
|
bootman
Emo VIPs
Typing useless posts on internet forums....
Posts: 9,358
|
Post by bootman on Jun 5, 2015 13:45:35 GMT -5
For #1. Where is this setting in a entry level avr as you suggest. I have never seen it. It is called 'Distance' and you can find it in Speaker settings. I've opened a manual of the first entry level AVR model I remember - Onkyo TX-SR313 and it is described on the page 41. It's on page 52 in the XMC manual also under speaker settings. But only available under presets 1 and 2 since Dirac sets distance on it own during setup. |
|
|
|
Post by igorzep on Jun 5, 2015 14:07:37 GMT -5
The manual Speaker Presets have an option to manually enter speaker distances (since they're not measured automatically); the Dirac preset doesn't include this option because it DOES measure the distances automatically). Since Dirac measures those distances automatically, and measuring them using the actual acoustic delay is more accurate in a practical sense than doing it by ruler anyway, we simply didn't see a need to include the option to manually tweak them. Can you explain HOW Dirac is able to measure the distances automatically? It connects to the XMC-1 through WiFi+router+Ethernet on one side, through unknown delay accuracy mic and operating system drivers on the other side... It simply have no hardware to do this measurement automatically with any accuracy contrary to the usual case with mic connected directly to an AVR that contains all the necessary calibrated hardware inside. How can Dirac account for all those largely unreliable variables? All it can do is to measure the DIFFERENCES in distance and this is why it is is doing the additional chirp of the front left at the end of a measurement cycle. If it would be able to measure distances accurately it would not need to do this additional chirp. But this chirp still does not let it any more information about absolute delay value, it only helps to improve the difference calculation... |
|
|
|
Post by igorzep on Jun 5, 2015 14:11:04 GMT -5
Every modern day AVR and processor I am aware of has this feature. As it is described on page 41 where you referenced, it is simply a manual adjustment. Or now it is often done automatically during setup by whatever program the processor has. Why should we need more than that? We need exactly that, but we don't have it... I am not talking about something more than that. |
|
|
|
Post by monkumonku on Jun 5, 2015 14:15:11 GMT -5
Every modern day AVR and processor I am aware of has this feature. As it is described on page 41 where you referenced, it is simply a manual adjustment. Or now it is often done automatically during setup by whatever program the processor has. Why should we need more than that? We need exactly that, but we don't have it... I am not talking about something more than that. I would venture to say that for all practical, reasonable, real world situations (which would then exclude abnormal listening distances such as 300 feet away), the XMC-1 has it. Anything else is needlessly academic and I doubt it is worth Emotiva spending any money or time trying to "fix" what doesn't need to be fixed in the first place. To me it is in the same category as fussing over speaker hiss that you can faintly hear with your ear right up to the cone but you can't hear from more than 6-12" away, and your normal listening position is several feet away from the speaker. |
|
|
|
Post by igorzep on Jun 5, 2015 14:16:22 GMT -5
We do not suggest that you recalibrate the absolute loudness level "every time after running Dirac". The main purpose of setting the speaker levels is to adjust all of your speakers to play at the same level. We don't consider it essential that anything be adjusted to any specific absolute level. If it is not re-calibrated then features such as Loudness and Midnight do not work correctly / as they are designed... So, the above statement is inaccurate. I use Loundess extensively and consider it a very essential feature. |
|
|
|
Post by millst on Jun 5, 2015 14:19:16 GMT -5
1) Missing configuration option for correcting an audio delay coming from how far from the listener speakers are placed (absolute one, so in relation to the video, not in relation between speakers themselves). 2) SPL level calibration so user should not re-adjust levels after every Dirac curve loading (relatively minor one, but taking some time from the user)... At least if they were not reset each time this would be much more convenient already. 3) Dirac ('default') assumptions and XMC-1 Bass Management assumptions in relation to amplitude and phase response are incompatible with each other... 1) Dirac is already computing the absolute delay/distance, same as Audyssey and other systems. Overriding it is likely to do more harm than good so it's not configurable. Yes, it's less flexible than other products, which do provide the ability to override it, and Emotiva could spend money to change this, but I don't see the value. 3) This is fairly standard and I've never seen this in any "cheap" AVR. I believe these are are all the same or worse: NAD M12, Anthem D2V, Yamaha CX-A5000, Marantz 8802. You'd probably have to go Krell Foundation or Theta Casablanca to get that crossover alignment. Audyssey Pro uses a very similar system of drawing the target curve. I haven't seen anything in the manuals that provide for an easily selectable 12dB low-frequency rolloff. -tm |
|
|
|
Post by millst on Jun 5, 2015 14:26:30 GMT -5
... 3) Dirac ('default') assumptions and XMC-1 Bass Management assumptions in relation to amplitude and phase response are incompatible with each other... If Dirac Live on the XMC-1 is a separate "black box" component inside the box that is the XMC-1, is there any advantage of the XMC-1 over a separate box solution like the MiniDSP nanoAVR DL Dirac Live, where clearly no bass management is done? The advantage of the XMC-1 over the nanoAVR DL is that the Dirac filters are applied after the crossover. Either that or you need an extra box to do the bass management first. Comparing the XMC-1 against the 88A, you eliminate an extra ADC/DAC conversion. Everything is more tightly integrated with XMC-1, too. One box versus multiple. Tradeoff is less flexibility. -tm |
|
KeithL
Administrator
Posts: 10,273
|
Post by KeithL on Jun 5, 2015 14:45:33 GMT -5
Actually, since it communicates with the XMC-1 over the network, Dirac is privy to significantly more information than you realize. For example, since Dirac communicates with the XMC-1, and the XMC-1 responds, Dirac has a pretty good idea how long network communications between the two take. So, yes, Dirac will be most accurate in its determination of the RELATIVE delay/distance between the various speakers. And this timing is the one that's critical because because any error there will adversely affect imaging. However, Dirac is also able to determine the overall delay reasonably accurately as well (I don't know precisely what method they use or how accurate it actually is). But, in general, the time it takes for Dirac to send a packet to the XMC-1, across a local network, should be on the order of a fraction of a millisecond. And, since it would take a 16 foot (16 millisecond) error to cause the equivalent of a single frame synch error between the audio and video, the absolute timing simply isn't all that critical. (The audio would have to be off for more than one full frame before you'd even potentially notice it.) You must also remember that crossovers and other electronics introduce electrical delay to the signal. You can use a ruler to measure the PHYSICAL distance to the speakers, but, because of that extra electrical delay, the value you measure may not accurately describe the "acoustic delay" anyway (the actual delay between when you send the signal and when the sound arrives at the microphone). So, not counting potential errors, measuring the actual acoustic delay is by far the more accurate method for getting the correct setting there. (And using a ruler to physically measure the distance isn't especially accurate at all.) The manual Speaker Presets have an option to manually enter speaker distances (since they're not measured automatically); the Dirac preset doesn't include this option because it DOES measure the distances automatically). Since Dirac measures those distances automatically, and measuring them using the actual acoustic delay is more accurate in a practical sense than doing it by ruler anyway, we simply didn't see a need to include the option to manually tweak them. Can you explain HOW Dirac is able to measure the distances automatically? It connects to the XMC-1 through WiFi+router+Ethernet on one side, through unknown delay accuracy mic and operating system drivers on the other side... It simply have no hardware to do this measurement automatically with any accuracy contrary to the usual case with mic connected directly to an AVR that contains all the necessary calibrated hardware inside. How can Dirac account for all those largely unreliable variables? All it can do is to measure the DIFFERENCES in distance and this is why it is is doing the additional chirp of the front left at the end of a measurement cycle. If it would be able to measure distances accurately it would not need to do this additional chirp. But this chirp still does not let it any more information about absolute delay value, it only helps to improve the difference calculation... |
|
|
|
Post by millst on Jun 5, 2015 15:11:23 GMT -5
I assumed Dirac was performing some kind of time synchronization between the XMC-1 and itself during the measurement process. I'd imagine that could be sub millisecond accuracy. I'm sure the latency for things like the XMC-1 outputting the tone to the analog outs and the USB mic's digitizing of the input are known constants. Use of exclusive mode when recording in Windows would give you a known latency.
All of the above is speculation, of course. Would be interesting to hear from Dirac. Even if it were 10ft off, it would be less than 1 frame out of sync. In the end, the important part is that the relative delay between the channels is accurate.
-tm
|
|
|
|
Post by igorzep on Jun 5, 2015 16:34:39 GMT -5
I'm sure the latency for things like the XMC-1 outputting the tone to the analog outs and the USB mic's digitizing of the input are known constants. Use of exclusive mode when recording in Windows would give you a known latency. All of the above is speculation, of course. Of course it is. I've seen measurements done by people trying to time align their speaker crossovers with an USB mic... They failed with tens milliseconds difference in readings when just measured multiple times the same speaker with the same USB mic. Drivers reporting arbitrary delays that have nothing to do with reality are also pretty common issue. Networks are not symmetrical, random network delays up to hundreds milliseconds are the normal activity... In the end, the important part is that the relative delay between the channels is accurate. Actually too not as accurate as I hoped... |
|
|
|
Post by millst on Jun 5, 2015 16:40:05 GMT -5
Of course those people are having problems. REW has a system for doing that kind of alignment and those that choose to ignore it are doomed to fail. Sonos uses Wi-Fi transmission and keeps multiple speakers time-aligned. It can be done. The question is whether Dirac has done the proper engineering.
-tm
|
|
