Understanding the state of the DSP market

[Robbo99999]

It’s not entirely without measurement. I use a UMIK-1 and the WiiM Home App’s RC function to check the overall frequency response of the system. That’s how I level matched and adjusted the crossover from the sub to the speakers to simulate a Harman curve. I just don’t use the resulting calculated EQ. I have a large null between 100-200Hz that I cannot cure despite trying all sorts of speaker placements, subwoofer placements, and crossover configurations. Including Final Assessment just in case you want to see the result if I were to leave the EQ enabled (I always disable it immediately after running so as not to stack RC on top of RC):
View attachment 447994

View attachment 447995

I am actually quite happy with the overall sound. I simply wish I could resolve the big room null in the lower midrange, more for peace of mind than anything.

In theory, I could remove the Anti-Mode 8033S-II from the equation as well as the bass management of the WiiM from the equation if I could adjust the crossover point in the Anti-Mode X2D. Supposedly the crossover point is adjustable on the X2D, however it involves connecting a USB flash drive formatted in FAT32 to the X2D, rebooting it to save configuration data to the flash drive, then plugging the flash drive into a computer and using DSPeaker's web-based console to change the crossover settings, saving the files over, plugging the flash drive back into the X2D, and rebooting it to take effect. Unfortunately, the theory is not working for me in practice so far as I am having a hard time getting the X2D to save the initial configuration data to my flash drive. Not sure if it is an issue with the formatting of the flash drive. I need access to a PC to know for sure, as I have a feeling my Mac is not letting me pick the right version of FAT when I format it using the Mac. However this part is off topic.

I think I remember having a null in the same place (but not quite as wide) as you have there when I first setup my speakers in the previous place I lived. I think I managed to get rid of it by changing the height of the speakers. I'm a bit vague on my memory on this but might give you a place to start if you wanted to google it to see if it's valid & maybe experiment.

 

[SoySalsaPicante]

I think I remember having a null in the same place (but not quite as wide) as you have there when I first setup my speakers in the previous place I lived. I think I managed to get rid of it by changing the height of the speakers. I'm a bit vague on my memory on this but might give you a place to start if you wanted to google it to see if it's valid & maybe experiment.

Thanks for this advice! I will give it a try somehow.

It could also be induced by a huge coffee table that sits between the speakers and my sweet spot. It is my concession with the wife, so it cannot be removed.

 

[AlmaAtaKZ]

What a pain in the arse a 93750 Hz sample rate would be. I don't know of any FIR generating software that could use it.
What's Hypex thinking, I wonder? Other than confine everything into itself?

Hypex did respond on that question: https://audiosciencereview.com/foru...plate-fusionamp-amplifiers.18537/post-1338143

 

[gnarly]

Hypex did respond on that question: https://audiosciencereview.com/foru...plate-fusionamp-amplifiers.18537/post-1338143

Yep, saw all that from 3dII3d00d's link....
Sounds like a microscopic techno focus....that confines everything to itself ! (other than maybe for folks who use VCad to generate FIR filters for it)

 

[levimax]

I know it's rooted to be measurement based.......but I simply doubt the validity of in-room measurements that lead to corrections beyond VERY broad brush strokes.

I have pretty much given up on in-room measurements beyond nearfield and MMM. For FR adjustments I find MMM fairly reliable and repeatable below Schroeder and nearfield sweep measurements using loopback also seem reliable and repeatable. Of course these measurements have limited uses but anytime I try sweeps at the listening position, even with gating and all the rest, I never get anything I can trust.

 

[masterhw]

I have pretty much given up on in-room measurements beyond nearfield and MMM. For FR adjustments I find MMM fairly reliable and repeatable below Schroeder and nearfield sweep measurements using loopback also seem reliable and repeatable. Of course these measurements have limited uses but anytime I try sweeps at the listening position, even with gating and all the rest, I never get anything I can trust.

assuming MMM = moving mic method, is this not essentially averaging an infinite amount of measurements around the MLP? Taking full measurements at every point (to some discretion) and averaging them should yield a very similar result. calculus!

 

[levimax]

Assuming MMM = moving mic method, is this not essentially

assuming MMM = moving mic method, is this not essentially averaging an infinite amount of measurements around the MLP? Taking full measurements at every point (to some discretion) and averaging them should yield a very similar result. calculus!

Yes when I average multiple sweeps taken at different positions around the MLP I get almost identical results to MMM. The difference is MM takes less than 1 min and averaging multiple sweeps takes 10 X to 100 X more time. If averaging multiple sweeps would give me phase information that I could rely on it would be well worth it but I have never been able to get it to work. With MMM there is no phase information to futilely try to make sense of so this again saves time and aggravation

 

[Keith_W]

There is certainly something quite peculiar about the loopback results there, though difficult to say what that might be. I'm happy to look at the mdat file to see if that gives any clues, though the measurement of the loopback channel itself is not retained, which may be where the problem arises. A measurement of the loopback channel without any timing reference could be helpful there.

Thanks for your interest, John. I am planning to do another round of timing measurements in REW to find out why the timings are so far off. I will make a post in the AVNirvana forums when I am done.

 

[ernestcarl]

I never get anything I can trust.

How does one set up a sub/multi-subs or multichannel setup etc. if you can't "trust" in-room measurements by even a little bit? I don't think the majority here think anechoic measurements are equivalent to in-room measurements at all. That would be naive. You only need to keep in mind that room acoustics, SBIR, modes, speaker directivity etc. are always at play -- there's always a potential of misinterpretation, user or other system error will occur... The calibrated omni mic is not equivalent to a pair of ears and brain. I mean, we all know that. Well, at least that's what I hope.

 

[SoySalsaPicante]

How does one set up a sub/multi-subs or multichannel setup etc. if you can't "trust" in-room measurements by even a little bit? I don't think the majority here think anechoic measurements are equivalent to in-room measurements at all. That would be naive. You only need to keep in mind that room acoustics, SBIR, modes, speaker directivity etc. are always at play -- there's always a potential of misinterpretation, user or other system error will occur... The calibrated omni mic is not equivalent to a pair of ears and brain. I mean, we all know that. Well, at least that's what I hope.

Well, one thing I've already noticed is that (and this came from pure experimentation) one aspect these measurements don't necessarily account for is directionality of the sound.

A good test would be placing a subwoofer far to the left or far to the right in a 2.1 setup, and setting a very high crossover point (for example 160Hz). Many, I dare say most, algorithms would not detect the heavily off-center nature of sound at the crossover frequency range. Such a test requires either stereo microphones (like our actual ears) or scanning at multiple locations and being able to recognize the off-center crossover frequency that way.

 

[ernestcarl]

algorithms would not detect the heavily off-center nature of sound at the crossover frequency range. Such a test requires either stereo microphones (like our actual ears) or scanning at multiple locations and being able to recognize the off-center crossover frequency that way.

Yes, it's wise to take multiple measurements across a space. Checking if the crossover/EQ actually holds up over a wider area than just a single point...

 

[JohnPM]

Thanks for your interest, John. I am planning to do another round of timing measurements in REW to find out why the timings are so far off. I will make a post in the AVNirvana forums when I am done.

I noticed the speaker was measured up to 300 Hz. Measuring a system below its bandwidth will produce a result which has a correspondingly altered IR, with a strong acausal component. The speaker would be better measured using all the available bandwidth (i.e. to half the sample rate) to properly capture its IR.

 

[gnarly]

I have pretty much given up on in-room measurements beyond nearfield and MMM. For FR adjustments I find MMM fairly reliable and repeatable below Schroeder and nearfield sweep measurements using loopback also seem reliable and repeatable. Of course these measurements have limited uses but anytime I try sweeps at the listening position, even with gating and all the rest, I never get anything I can trust.

Me too !!!!!!!
Although every time I read about some new/better measurement technique, I get excited and try it and measure a sub in the room to see......so far, hope springs eternal. Lol

 

[levimax]

How does one set up a sub/multi-subs or multichannel setup etc. if you can't "trust" in-room measurements by even a little bit? I don't think the majority here think anechoic measurements are equivalent to in-room measurements at all. That would be naive. You only need to keep in mind that room acoustics, SBIR, modes, speaker directivity etc. are always at play -- there's always a potential of misinterpretation, user or other system error will occur... The calibrated omni mic is not equivalent to a pair of ears and brain. I mean, we all know that. Well, at least that's what I hope.

I trust MMM below Schroeder and nearfield measurements and from there you can get most of what you need. I do use multiple LP sweeps and averaged they are fine for FR. Where I struggle to find consistency is the timing/ phase measurements taken at the LP. About the only thing I can get to work somewhat reliably is the "Delay a driver and look for 2 impulse responses method" to confirm timing, @Keith_W h uses this method as does Mini-DSP https://www.minidsp.com/applications/rew/measuring-time-delay . Even this is hard though as it is hard to tell what is a reflection and what is the delayed impulse. I have to take multiple sweeps with different delays and watch which "peak" moves to make sure I am looking at it correctly. I usually get the same answer using Nearfield measurements and a tape measure and when I do I feel confident and if not I go back to square one and figure out what I did wrong.

 

[Keith_W]

I noticed the speaker was measured up to 300 Hz. Measuring a system below its bandwidth will produce a result which has a correspondingly altered IR, with a strong acausal component. The speaker would be better measured using all the available bandwidth (i.e. to half the sample rate) to properly capture its IR.

It's an active 4 way speaker. Each driver needs to be aligned to the tweeter, so can not be measured up to 20kHz. This was why I was asking you about the tweeter timing chirp on the AVNirvana forum. I want to measure the distance between the chirp and the impulse peak of the DUT to see if I can determine the timing that way. But you said it's not part of the same IR.

 

[gnarly]

I noticed the speaker was measured up to 300 Hz. Measuring a system below its bandwidth will produce a result which has a correspondingly altered IR, with a strong acausal component. The speaker would be better measured using all the available bandwidth (i.e. to half the sample rate) to properly capture its IR.

Hi, watched your interview last night. Thx for doing it, interesting to hear about upcoming 5.4.
And great to hear wavelets are already in the spectrograph...will explore asap. Surprised I've missed it.

In designing and tuning linear-phase active multi-ways,
My practice for subs <120Hz, or lows < 500Hz, or even mid range drivers <1000Hz,
has evolved into low passing each section with a linear-phase low pass when taking raw measurements. I typically set a higher order lin-phase low-pass two octaves above the intended passband.

I've found I get measurements that take less final adjustment of fixed delays with the 'measurement assist' low-passes in place than without, along with FIR filters having less work to do, giving cleaner electrical impulses.. Seems to help truncate needless out of band FIR mag and phase corrections.

Do you see anything inherent wrong with this practice?
I should probably ask on your REW forum... but since the discussion has touched on how to best measure low frequency drivers...
thx !

 

[JohnPM]

Each driver needs to be aligned to the tweeter, so can not be measured up to 20kHz.

If you measure a driver over a range that is less than the driver's bandwidth there is in effect a frequency domain truncation of the driver's response, which will result in ringing in the time domain in the IR corresponding to that measurement and a broadening of the IR commensurate with the bandwidth reduction. It will not be the IR of the driver.

There are other ways to align speakers/subs or individual drivers, Pat Brown has a set of videos illustrating an alternative approach here.

 

[Purité Audio]

Ecellent videos, clear and concise.
Keith

 

[Robbo99999]

Re those videos, they were to the point & well explained I think. I think I did kind of the same thing but all in the frequency domain & just using acoustic measurements when I combined my sub & mains using miniDSP 2x4 and REW (along with UMIK). First I measured physical difference with tape measure between subwoofer and the horizontal axis of the 2 speakers (if you'd drawn a line between the 2 speakers) whilst travelling in direction of the listening position (sub is behind speakers). I applied that physical distance difference as a delay to the main speakers (miniDSP software allow a ms delay to be calculated from physical distance difference). I flipped polarity on subwoofer to see if I got an increased frequency response at crossover position and didn't so I left it at default. Then I adjusted phase of the subwoofer using the SVS app until I got the highest summed frequency response at crossover position. I then did RoomEQ to take down the peaks below 300Hz, applied into miniDSP 2x4 of course. So I think I tried to achieve the same thing that was being done in those videos but all done in the frequency domain. They were good videos, but I lost it a bit in the last two videos.

 

[Oddball]

How does one set up a sub/multi-subs or multichannel setup etc. if you can't "trust" in-room measurements by even a little bit? I don't think the majority here think anechoic measurements are equivalent to in-room measurements at all. That would be naive. You only need to keep in mind that room acoustics, SBIR, modes, speaker directivity etc. are always at play -- there's always a potential of misinterpretation, user or other system error will occur... The calibrated omni mic is not equivalent to a pair of ears and brain. I mean, we all know that. Well, at least that's what I hope.

It is a long thread by now and have not been following all of it, so might be repeating or contradicting the previous posts.

To the best of my knowledge, there is no mic that would be equivalent to our ears+brain, so there is always going to be a disconnect. There are various mics in the business and they will do the job as good as they can. UMIC-1/2 and Audyssey mics are probably most common, with the Audyssey calibrated mic being in the tight calibration range despite its toy looks. Then there would be Lyngdorf mic and Trinnov mic, with the latter potentially being more accurate.

Calibrating for the large HT system is completely different than for the 2 channel system even if the latter is reinforced with subs. Good thing is that multi channel will mask lots of faults that would be more apparent in 2 channel content, but bad news is that integrating them is going to be more difficult.

For the large HT system the tools you have at your disposal are limited to build in room correction and potentially sub DSP, unless you want to complicate your life with adding more DSPs in the chain to correct additional things that might or might not need correction.

It is unlikely that in the large HT setup you will be able to get showcase frequency response, phase response, group delay response and sound delay response. You might be able to get close to that if you overcomplicate your setup, but than the question is if that is really worth it.