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

[DuncanTodd]

MultEQ-X is the only thing that has attempted a fix or workaround for this to date. Without it, set your speaker distances to metric and multiply the values by 0.875 to get a more accurate impulse timing.

Thanks. Do you mean measure manually with a tape and enter the results after the x 0.875 on the Audy settings, or manually correct the automated distances set by Audy by x 0.875?

 

[Chromatischism]

Thanks. Do you mean measure manually with a tape and enter the results after the x 0.875 on the Audy settings, or manually correct the automated distances set by Audy by x 0.875?

The latter.

 

[Galz]

MultEQ-X is the only thing that has attempted a fix or workaround for this to date. Without it, set your speaker distances to metric and multiply the values by 0.875 to get a more accurate impulse timing.

Are you sure? From what I understand Audyssey sets correct delays and you shouldn't modify manually, and the issue is only relevant if you tape measure, or try to manually enter values based on MultEQ-X measurements. Although I had a hard time finding where I saw this, so I could be wrong...

 

[peng]

Are you sure? From what I understand Audyssey sets correct delays and you shouldn't modify manually, and the issue is only relevant if you tape measure, or try to manually enter values based on MultEQ-X measurements. Although I had a hard time finding where I saw this, so I could be wrong...

I did a search for you. There seemed to be more than one threads on AVSF that discussed the topic but I would focus on the following as it included a response from Audyssey:

https://www.avsforum.com/threads/audyssey-time-alignment-is-off-according-to-rew-impluse-readings.3243684/page-9#post-61661422

 

[Galz]

I did a search for you. There seemed to be more than one threads on AVSF that discussed the topic but I would focus on the following as it included a response from Audyssey:

https://www.avsforum.com/threads/audyssey-time-alignment-is-off-according-to-rew-impluse-readings.3243684/page-9#post-61661422

I read that one already but it says nothing conclusive about calibration with the mobile app or the receiver.

 

[peng]

I read that one already but it says nothing conclusive about calibration with the mobile app or the receiver.

Did you read this one too:

https://www.avsforum.com/threads/audyssey-multeq-x.3228468/page-57#post-61666474

Note the part ".....He's not experiencing the issue because he's using the licensed MultEQ-X app, which compensates for the bug for you. But you can do the same yourself by running the built-in or mobile apps, and manually scaling by the 0.874 (12.5%). Just be sure to scale the metric values and not feet values because there are yet more bugs in their conversion to feet. Bugs, bugs! Bugs everywhere......"

I seems to me if he is right and I think he is, then unless you have the MultEQ-X, if you want to have the delay more accurate you would to use the 12.5% correction factor but use metric (based on his finding reported on his other post). That means you need to apply the correction to the results if you use the mobile app or the AVR itself to run Audyssey.

I think to get everything confirmed, one would have to contact Denon, quoting Audyssey's response that confirmed the error but they did not want to address it because it was Denon (presumably Marantz too) who introduced the error.

 

[BigVU's]

It can be a simple as connecting it to an access point that is capable of Wireless to wired bridging.. Not all are capable.

Many old Linksys are so capable.e.g the very popular WAP54G can do it; other more modern AP are able to play the role of wireless to wired bridge. Involves a bit of configuration but nothing very arcane..
Here's an example

Peace

Thankyou. going to give it a go. just didn't to waste too much time if it had already been tried with no success.

 

[Chromatischism]

Are you sure? From what I understand Audyssey sets correct delays and you shouldn't modify manually, and the issue is only relevant if you tape measure, or try to manually enter values based on MultEQ-X measurements. Although I had a hard time finding where I saw this, so I could be wrong...

It does accurate measurements - the problem is that internally, the speed of sound constant that is being used is slightly off. Who knows why, but it was discovered by a user at AVS and confirmed by others. The workaround adjusts the timing to close the gap.

 

[Dj7675]

It does accurate measurements - the problem is that internally, the speed of sound constant that is being used is slightly off. Who knows why, but it was discovered by a user at AVS and confirmed by others. The workaround adjusts the timing to close the gap.

The flaw is really surprising it was there at all, and not really shown to exist until now with all the people that take pre/post Audyssey measurements. I haven’t followed it too closely, but has Denon/Marantz acknowledged the issue yet?

 

[Chromatischism]

The flaw is really surprising it was there at all, and not really shown to exist until now with all the people that take pre/post Audyssey measurements. I haven’t followed it too closely, but has Denon/Marantz acknowledged the issue yet?

I'm not aware of any communication with Denon, just with Audyssey.

 

[KMO]

I read that one already but it says nothing conclusive about calibration with the mobile app or the receiver.

So far only the MultEQ-X app has correction. It will show real physical distance measurements in its UI, but it will put them into the receiver with the 0.875 adjustment.

If using the normal Audyssey app, or the receiver itself, or a tape measure, you should make the 0.875 adjustment yourself.

It does accurate measurements - the problem is that internally, the speed of sound constant that is being used is slightly off. Who knows why, but it was discovered by a user at AVS and confirmed by others. The workaround adjusts the timing to close the gap.

Yes, that speed of sound constant means an inaccurate conversion of entered distance to delay.

I believe the reason why is just that 1ft = 1ms = 30cm was a reasonable approximation when older D+M AVRs only had 1ft/1ms/30cm distance adjustment, maybe 20 years ago. It didn't cause any appreciable error within the adjustment granularity available at the time, and it kept the maths simple.

When they refined the adjustment granularity to 0.1ft/0.1ms/3cm, that approximation was no longer good enough - lots of people will end up with 0.5ms error, thus not achieving the expected 0.1ms accuracy.

The flaw is really surprising it was there at all, and not really shown to exist until now with all the people that take pre/post Audyssey measurements. I haven’t followed it too closely, but has Denon/Marantz acknowledged the issue yet?

The problem is independent of Audyssey. And it doesn't really show up on any frequency response, which is what most people are looking at. Also many pre/post comparisons will be done by turning Room EQ on and off, so keeping the timing...

Don't think D+M have said anything.

 

[Galz]

I've asked this before but got no answers yet...
Anyone tried to set the cutoff frequency to the crossover frequency (keeping the default 2nd order slope) and measure how that improves the subwoofer integration (over the default which is often lower than 80Hz)? Preferably comparing after using REW to manually adjust subwoofer distance for optimal integration in both cases.

 

[DuncanTodd]

The latter.

Forgot to ask. How does that calculation works with subs? I assume it's not a straight x0.875 like with passive speakers.

 

[KMO]

0.875 applies for any active or passive speaker or sub.

An active speaker/sub will show a bigger distance as it incorporates some virtual distance for the processing delay. But that virtual distance is subject to the same distance->delay conversion error and still needs the 0.875 to get the right delay compensation in the AVR.

 

[Chromatischism]

Forgot to ask. How does that calculation works with subs? I assume it's not a straight x0.875 like with passive speakers.

It technically is, however you may still want to do sub distance tweaking on your own.

 

[DuncanTodd]

0.875 applies for any active or passive speaker or sub.

An active speaker/sub will show a bigger distance as it incorporates some virtual distance for the processing delay. But that virtual distance is subject to the same distance->delay conversion error and still needs the 0.875 to get the right delay compensation in the AVR.

Thanks.
I stumbled on a post on different thread on avsf that went into quite a confusing explanation (maybe not so confusing, but I've yet had time to read it), how tuning the subwoofer distance is different in this case, so wanted to verify.

 

[KMO]

It's simple if you're using a distance measurement from (non MultEQ-X) Audyssey that was based on delay, like the speakers' was. Then 0.875 for everything gets you to what Audyssey was intending to do. (Which I believe is a pure impulse alignment, not addressing phase integration for crossover).

But if after initial Audyssey measurement you did any sort of crossover integration through frequency response analysis, changing the sub distance, then putting the 0.875 change in would mess with that integration. The complicated posts in there are trying to talk people through what to do to their sub distance number to preserve the sub/speaker offset and hence their crossover integration, without having to redo the measurements and sub timing alignment. You could just redo it after the 0.875 adjustment.

Actually - the 0.875 adjustment is so small compared to the sort of changes made during sub alignment, it probably doesn't really matter anyway. The timing alignment tool in REW uses 0.5ms steps by default, and that's a very small change to phase at typical crossover levels, starting from an aligned position. The 0.875 tweak will likely not end up shifting stuff by even two of those steps. So just doing the 0.875 to everything will only ever so slightly disrupt a crossover integration.

 

[Galz]

I remote connected to a friends PC to try and help him tune his system and saw something strange:

I added some low shelf filter to account for some room gain, trying to boost the bass a few decibels. However, the effect on the "after" curve seemed much smaller than the decibel amount I entered in the filter (a 3db boost seemed to boost by a very small amount, 1db or less, compared to no filter). Even setting the maximum of 10db boost showed the "after" curve only a few db higher than the baseline level. Any idea if this is just a display error or if the filters work differently than what I expect them to do? Unfortunately, this was done remotely so I had no way to measure what was actually happening.

On another note, it seems like the MultEQ-X "after" curve is much more realistic-looking than the android/iphone app "after" curve, which strengthens the claim that the android/iphone app "after" curve is complete BS even if you give it (probably false) credit for being some kind of spatial averaging result.

 

[peng]

I remote connected to a friends PC to try and help him tune his system and saw something strange:

I added some low shelf filter to account for some room gain, trying to boost the bass a few decibels. However, the effect on the "after" curve seemed much smaller than the decibel amount I entered in the filter (a 3db boost seemed to boost by a very small amount, 1db or less, compared to no filter). Even setting the maximum of 10db boost showed the "after" curve only a few db higher than the baseline level. Any idea if this is just a display error or if the filters work differently than what I expect them to do? Unfortunately, this was done remotely so I had no way to measure what was actually happening.

On another note, it seems like the MultEQ-X "after" curve is much more realistic-looking than the android/iphone app "after" curve, which strengthens the claim that the android/iphone app "after" curve is complete BS even if you give it (probably false) credit for being some kind of spatial averaging result.

May be within the range you picked there are several room mode dips that Audyssey couldn't do much about.

 

[KMO]

You can't effectively boost your way out of low-frequency nulls. Almost all extra energy from your gain just goes to the other places in the room, probably making the balance in those places even worse.

Your best bet is moving the subwoofer / speakers. Or listening position.

You can see similar effects from reflections cancelling at higher frequencies - you boost the reflections too, so they still cancel.

Most of the time a +3dB gain filter will provide +3dB output, but not when nulls/cancellations are happening.

If you mess with REW or similar you'll see people wittering on about "minimum phase" regions of response - that's what they're talking about. "Minimum phase" is the good parts of the frequency response that respond to EQ - an "excess phase" graph shows spikes in regions where it won't respond to EQ.

The simple Audyssey app will be assuming everything is minimum phase and will respond to EQ perfectly, and shows what the after would look like if that was the case. (It's just literally summing the "before" and the "EQ" level, ignoring phase).

If the MultEQ-X app is doing better, then it is likely paying attention to the phase issues.