Audyssey's Next Generation of Room Correction (MultEQ-X)

[GalZohar]

You could also just create your own MRC with the 20$ app after disabling the default one, can't you?

 

[peng]

You could also just create your own MRC with the 20$ app after disabling the default one, can't you?

Yes, but probably more work than using the X, if I want to replicate the same kind of dip MRC enabled would do especially if I draw. It would be much easier if I use Ratbuddyssey, that is, type instead of draw.

 

[HarmonicTHD]

In my opinion, the Denon/Audyssey video is way better if the goal is to understand how the software works. The Joe N one seems biased to me and has a lot of his personal understanding/interpretation of how MultEQ X works. He seems like a better presenter and they could be one reason why his video may be the more popular one.

The D+M/Audyssey one is naturally biased too but at least you will get the correct (should be...) interpretation/explanation of the the features and how they work, right from the manufacturers.

What import option would you like to have?

Hi Peng.

You inspired me some 4 to 6 months ago to try Ratbuddyssey to see how much some finetuning of the automated XT32 results could further improve the room response. (You nicely demonstrated this with your REW postings).

Now we know, what a tedious process that is to use REW and Ratbuddyssey and iteratively feedback the results via the IOS App into the AVR.

The interview with the Denon developer and JoeNTell at least suggested that there might be an easier way in the future as opposed to the above mentioned process (via Ratbuudyssey) to get the REW corrections into the „Housecurve“ adjustments. (And yes of course these interviews are biased and contain marketing, but most here are mature enough to take it with a grain of salt and take out the useful info to make an informed decision whether to buy the MSApp or not).

This is what I am hoping for, just to clarify.

 

[peng]

Hi Peng.

You inspired me some 4 to 6 months ago to try Ratbuddyssey to see how much some finetuning of the automated XT32 results could further improve the room response. (You nicely demonstrated this with your REW postings).

Now we know, what a tedious process that is to use REW and Ratbuddyssey and iteratively feedback the results via the IOS App into the AVR.

The interview with the Denon developer and JoeNTell at least suggested that there might be an easier way in the future as opposed to the above mentioned process (via Ratbuudyssey) to get the REW corrections into the „Housecurve“ adjustments. (And yes of course these interviews are biased and contain marketing, but most here are mature enough to take it with a grain of salt and take out the useful info to make an informed decision whether to buy the MSApp or not).

This is what I am hoping for, just to clarify.

Thank you for the clarification. I would just want to emphasize that the way I did it with Ratbuddyssey and the App is indeed very tedious, but it was only because I had too much time to try and get a near flat line. If I were to accept +/- 2 to 3 dB in the 20 to 200 Hz range, with no smoothing (okay may have leave 1 or 2 narrow dips alone) it would take me less than two hours to achieve, including the time to run REW a few times. Again, what I did was necessary unnecessary and in my opinion/experience had no audible benefits, it could have made the other seats, or impulse response worse, though fortunately that wasn't the case, upon checking with REW. Still, that's just my opinion based on my own experience.

JoeNTell would be right for sure, if he said there might be an easier way in the future. There is always a better way.

 

[HarmonicTHD]

Thank you for the clarification. I would just want to emphasize that the way I did it with Ratbuddyssey and the App is indeed very tedious, but it was only because I had too much time to try and get a near flat line. If I were to accept +/- 2 to 3 dB in the 20 to 200 Hz range, with no smoothing (okay may have leave 1 or 2 narrow dips alone) it would take me less than two hours to achieve, including the time to run REW a few times. Again, what I did was necessary and in my opinion/experience had no audible benefits, it could have made the other seats, or impulse response worse, though fortunately that wasn't the case, upon checking with REW. Still, that's just my opinion based on my own experience.

JoeNTell would be right for sure, if he said there might be an easier way in the future. There is always a better way.

Absolutely also my experience. After putting in a lot of work with Ratbuddyssey the results were, let’s say a bit less than overwhelming. Recently we redid the living room and I didn’t want to go through it again. But it was fun initially and it proved the concept and taught me sth. After all it’s a hobby… right.

 

[bkdc]

Should be tied to the product serial number and transferable. Count me out.

 

[anotherhobby]

Should be tied to the product serial number and transferable. Count me out.

I really wish it wasn't licensed the way that it is. Of all the ways Audyssey could have licensed it, they went with the most restrictive and profitable model, and it feels like a money grab. That said, MultEQ-X is not made by and sold by Denon and Marantz like the mobile app is. This is made by Audyssey, so it's essentially a cross-sell. Either way, I feel that after you've bought the software once, a reduced fee of maybe $50 or something per additional AVR would be far more reasonable for owners of multiple units.

I bought MultEQ-X, and I do feel I've got my value out of it it for my AVR, but everybody values different things. I also only own one AVR, so I'm not currently affected by the licensing limitations. I will be very annoyed when I upgrade some day though. The $200 premium on every single AVR you own is a tough pill to swallow.

 

[peng]

Absolutely also my experience. After putting in a lot of work with Ratbuddyssey the results were, let’s say a bit less than overwhelming. Recently we redid the living room and I didn’t want to go through it again. But it was fun initially and it proved the concept and taught me sth. After all it’s a hobby… right.

For clarity, I made a typo, meant to say the tedious way I did it was "unnecessary", not "necessary".

 

[Ruspamen]

hello everyone, when you perform the subwoofer level before the measurement, what level should it reach in slow e super slow graph? 75 decibels or higher than 80 db so that the trims go negative?

 

[thewas]

 

[Sonic icons]

It's as if this product is not from Denon/Marantz but instead from a 3rd party

True - Audyssey MultEQ-X software is not from Denon/Marantz (a division of Sound United), rather this software is from Audyssey Laboratories (https://audyssey.com/news/)

 

[Sonic icons]

Audyssey does not actually use PEQs. As mentioned by others, and it was explained in that video Audyssey/Denon video (From around the 50 minute mark). ...... When you use the "PEQ" using MultEQ X, that PEQ data entry format, that is the frequency, gain and Q values, Audyssey would use the entered values to figure out the FIR filters accordingly.

I've wondered about the nitty-gritty of designing a Finite Impulse Response (FIR) filter. I found a practical description. This is for software called "Micromodeler DSP", a "web-based digital filter development environment". (Not specifically for audio, digital signal processing has many other applications.) See: https://www.micromodeler.com/articles/IntroductionToDSP/FIR.jsp, especially the sections "Creating an FIR Filter", "Editing the filter's frequency response", "Impulse Response". I gather from this description that the frequency response described by a single FIR filter can be simple or complex. The level of complexity of the FIR filter is determined by the total number of "control points". Each control point is specified by a few parameters: frequency, magnitude, "ripple" factor, and interpolation ("shape of curve used to join this control point with the next"). Since a single FIR filter with sufficient "control points" can describe any frequency response curve, no matter how complex, only one FIR filter is needed to specify the correction curve for one speaker in Audyssey. (This is unlike PEQ based correction schemes where many PEQs must be added to describe a complicated response curve.)

But how many control points are in an Audyssey FIR filter that describes the correction curve for a single speaker? In Audyssey MultEQ XT32, the maximum number of "control points" is 512X for each speaker (satellite or subwoofer) (this ASR post). The most primitive version of Audyssey, called 2EQ, provides X control points for each satellite and 0 for the sub. Audyssey MultEQ (still sold today in entry-level AVRs) provides 2X control points for each satellite and 128X for the sub. But how many is X? Audyssey has said, "we won't tell you the number X, that's proprietary information". If X is at least 20 (my guess), that means MultEQ XT32 provides more than 10,000 control points per speaker. If we were given full access to the specifications for an Audyssey FIR filter with thousands of control points, I don't think anyone would want to manually edit that filter (!).

 

[Sonic icons]

thanks, so can you confirm that with mult-eq x there is no longer the problem of raising the trim level by the same amount as the boost applied? if so, do you think it would be enough to simply create this curve or, combined with it, raise the trim levels anyway?

According to the avsforum thread on MultEQ-X, yes there is still the problem of raising the trim level (of the subwoofer channel). This post quotes a response from Audyssey support, which I take to be the definitive answer.

Audyssey tech suppport comment, as quoted on avsforum:

This is by design - the digital headroom is managed such that we keep the level the same - this also avoids a large on/off level difference.

Currently, to compensate for the level, you should apply the same target curve to subwoofer and satellite speakers - then you can choose a frequency - when the curve is the same for sats and sub, any frequency will do, but if you look at, say, 250Hz, then you will usually find the target curve is down several dB there for the subwoofer. If you use that offset and add it to the trims it will get you what you're looking for. For example, say it's -4.5dB on the sub, while the satellite is at 0dB, then you dial in +4.5dB to your subwoofer trims.

This is expected behavior, and we anticipate that a future version of MultEQ-X will do this for you automatically.

To further illustrate the issue, here are MultEQ-X graphs for the Surround and Subwoofer 1 channels using a "target curve" that I created, in demo mode. Reference is set to my "target curve", and Flat is unchanged (flat). (I captured this image with the Print Screen button, followed by paste to a graphics app, crop, and vertical stretch x2). Comparing the graphs, the target curve level is 0 dB at 1 kHz for the Surround and -5 dB at 1 kHz for Subwoofer 1. That means a subwoofer trim of +5 dB must be dialed in manually to achieve a smooth blend between the subwoofer and satellite (front, center, surround) responses.

 

[Dj7675]

According to the avsforum thread on MultEQ-X, yes there is still the problem of raising the trim level (of the subwoofer channel). This post quotes a response from Audyssey support, which I take to be the definitive answer.

Audyssey tech suppport comment, as quoted on avsforum:


To further illustrate the issue, here are MultEQ-X graphs for the Surround and Subwoofer 1 channels using a "target curve" that I created, in demo mode. Reference is set to my "target curve", and Flat is unchanged (flat). (I captured this image with the Print Screen button, followed by paste to a graphics app, crop, and vertical stretch x2). Comparing the graphs, the target curve level is 0 dB at 1 kHz for the Surround and -5 dB at 1 kHz for Subwoofer 1. That means a subwoofer trim of +5 dB must be dialed in manually to achieve a smooth blend between the subwoofer and satellite (front, center, surround) responses.

View attachment 202702

This is correct and laid out very well. I learned this from a post @Jon AA made on either AVS or Audioholics (can’t recall). He showed the same thing. I know when the app came out there were a lot of complaints about lack of bass when using a sub curve. Post measurements show if you boost the sub trims the same amount as your target curve it will match the mains curve as long as you use the same curve. I always assumed it was a flaw in Audyssey but it sounds like it is just working as it was originally designed which was ruler flat bass.

 

[Head_Unit]

Measurement with detailed frequency responses for individual channels

Are these real and detailed responses, like REW? Or those fake-looking "targets" like Audyssey/ARC?
- No Mac = NO.
- Not per-unit and thus transferable = greedy and stupid = NO
I'd be far more interested in a "Must-EQ NSMD"* for free...and 3-4 subs would be cool.
*No Stupid Midrange Dip

 

[Dumdum]

I’ve seen a few people saying this puts audessey on the same playing field as Dirac… not really, Dirac uses FIR for timing as well as freq response, this just uses FIR for freq response… Dirac corrects and adjusts the impulse response to effectively focus all freqs timing at the listening position and correct for drivers and how they effect the impulse and also the room to a lesser degree

 

[GalZohar]

I’ve seen a few people saying this puts audessey on the same playing field as Dirac… not really, Dirac uses FIR for timing as well as freq response, this just uses FIR for freq response… Dirac corrects and adjusts the impulse response to effectively focus all freqs timing at the listening position and correct for drivers and how they effect the impulse and also the room to a lesser degree

So far I've yet to see someone with measurements showing this clear advantage. Although I'm still curious as I did see some comparisons with differences in phase, it just wasn't clear if/how what Dirac did was any better.

 

[Dumdum]

So far I've yet to see someone with measurements showing this clear advantage. Although I'm still curious as I did see some comparisons with differences in phase, it just wasn't clear if/how what Dirac did was any better.

Dirac delays any early arriving frequencys to make the phase flat

A speaker doesn’t send all frequencys evenly (in terms of time) when it creates the pressure, the low bass and high freq preceed the mid band

For example like this phase plot of my workshop active setup

The freq response is roughly flat from 80-20k, it’s a 13l ported bookshelf design with a 6.5” midbass and a 19mm tweeter

But as you can see the phase shows the late arrival (further down in the top graph is later)

What Dirac does (as well as correcting the freq response with FIR filters) is it adds delay with FIR filters to the early arriving frequencys to align the phase correctly and create a flat phase response, this is also why it corrects the impulse response, it reduces the lag of certain freqs therefore shortening the impulse response to make it nearer to perfect

 

[Newman]

Yes but he’s asking you (or someone) to show him the above measurement that shows the flat phase response after Dirac.

 

[Dumdum]

Yes but he’s asking you (or someone) to show him the above measurement that shows the flat phase response after Dirac.

So he’s disputing what Dirac does and how it works… I’m not in a position to provide that as I don’t own a Dirac processor… but it’s well publicised what it does and how it does it