Audyssey Room EQ Review

[DanGuitarMan]

I'm using rePhase to mannualy create my filters and while it doesn't impose any limits pushing dips more than 6-8dB is simply counterprorductive and may lead to undesired effects. That is the reason why automated EQ solutions like Acourate, Dirac etc never do it.

What hardware or software do you use to implement your rePhase filters?

 

[QMuse]

What hardware or software do you use to implement your rePhase filters?

BruteFIR plugin for Volumio. It's running on a fanless Intel based PC.

 

[capt.s]

Well behind you isn't really possible(I might have exaggerated there if my description came off like that) but right above you, 90 degree to the side of you and right beside your head etc. is possible but you need a very good room. I've been in rooms that are way better than mine and it's truly breathtaking what is possible. But with all that said the majority of recordings are just too poor.

You can come very far by making your room as symmetrical as possible and be really precise when placing and toeing in your speakers(use laser or similar) and take care of first reflections. But even when doing this the room will be the ultimate limitation - most rooms it's just not possible regardless how good your speakers and placement is. This is getting off topic but the gist of it is that spending days to fine tune speaker and listening position is worth it! Just a few cm will make a difference!

Try the album Roger Waters - Amused to Death

Edit: Not until all that stuff is taken care of(first reflection points, days of fine tuning placement of speakers AND listening position) I would even think about room correction. Room correction is the icing on the cake but it can make a massive difference in the bass region, which in combination with the other stuff taken care of can result in a big difference.

The album is on Tidal so I took a listen - possibly the most expansive sound stage recording I've heard! I have a pretty optimal symmetrical setup; speaker 6 feet from side walls, first reflection points taken care of etc etc. My list of priorities relating to quality and sound stage have always been (withinn reason) room setup / room eq ability - speakers - amp - source equipment. However, listening to that reminded me source material really should be a the top of the list. I can't believe just how poor/flat a lot of recordings are and you forget what a really good stereo mix is capable of. Of course the answer is most people don't care or notice so it's not a priority.

 

[peng]

I will try it later when my house is awake and ready to hear test tones all day. Why does this happen when measuring the summed output of a pair?

May be your of of the speaker has its tweeter's polarity's reversed.

Why does this happen when measuring the summed output of a pair?

If this only happens when measuring both channels together then you may have a tweeter wired with polarity reversed. If it is due to a defective tweeter or crossover then you would have the same dip even with one speaker, that is, left or right only.

 

[doodlebro]

I have a question about the high frequency dip. It is also present in my measurements with a U-MIC and REW. Calibration does very little to change it using the minidsp calibration file. I use an old boom mic to make measurements. Why does this happen? I know it is not a big deal but it irritates my graph sensibilities.

Calibration-
View attachment 59885
No calibration-
View attachment 59886

You need to click the "All SPL" tab. This is normal after you first measure when you are still on the "SPL & Phase" tab.

 

[mitchco]

Nice review @amirm The number one issue I have with (all) H/W solutions like this is the limited number of FIR filter taps to actually provide real low frequency room correction. Whether it is Audyssey, miniDSP, Trinnov, DEQX, or other H/W processors they all suffer from the same issue - not enough FIR filter taps below 100 Hz to really be called room correction.

As a maths example, the miniDSP 2x4 HD product datasheet talks about 4096 taps. But this is the total number. For 4 channels you have 1024 taps each channel available. The frequency resolution of a 1024 taps filter @ 48 kHz samplerate is 48000/1024 = 46.875 Hz. So e.g. below 100 Hz there are just 2 frequency bins at 46.875 and 93.75 Hz. This clearly means that you have no control over the lower frequency range.

As a graphic example, here is a Trinnov Altitude 16 versus Audiolense. Audiolense is a software based DSP/DRC product that is not limited by the number or FIR filter taps for low frequency control. In this case, Audiolense has generated a FIR filter with 65,536 taps, hosted in a software based convolution engine on a PC. We can clearly see the difference with the measured Trinnov correction on top and the measured Audiolense correction on the bottom using the same speakers same room, mic, etc:

Here is another example using Audiolense with a FIR filter length of 131,072‬ taps for the ultimate in low frequency control with 600ms of excessphase correction at 10 Hz:

Note that this measurement was taken at 9ft at the listening position using REW's default 500ms window and no smoothing. Meaning low frequency room reflections (i.e. standing waves, resonances) are getting into the measurement. But as one can see in the phase response, it follows the speakers minimum phase response, with no excessphase (i.e. no low frequency room reflections) disturbing the bass response. The result is crystal clear, even sounding bass response. To learn more, I wrote an article on the subject or one can hear my talk on it.

The point I am making is that not all DRC products are the same. Some have serious limitations on how much low frequency room correction can realistically be accomplished and folks should be aware of that.

 

[mansr]

The album is on Tidal so I took a listen - possibly the most expansive sound stage recording I've heard!

That's QSound at work.

 

[bobbooo]

That would require measuring a time span instead of a moment in time. The complexity of that will definitely give a certain Klippel owner a headache.

Yes to fully characterise the perceived sound effect from the Leslie speaker it would, but even a standard measurement would partially capture this, as the Doppler effect acts to shift the frequency response (in a time-varying manner in the case of this speaker).

This is the same argument as to why small movements of the mic (e.g. when holding it instead of using the supplied stand) when taking Audyssey measurements could possibly affect the result, by shifting the frequency response - any relative motion between sound source and measuring apparatus (mic or ears) will result in frequency shifts via the Doppler effect.

 

[SIY]

Unless you’re really flailing the mike, Doppler should be negligible. Frequency shift goes linearly with the ratio of the mike speed to the speed of sound, so something like .001.

 

[Robbo99999]

Well, that is just one thing. As soon as the wavelength of audio becomes close to width of your face, then one microphone is not measuring what two ears are hearing. Measurements can quickly become useless than way unless great care is taken. One of these days I will write the REW tutorial and explain these things.

I think that would be an excellent thing to do Amir, I've found EQ to be very important (recent experience), and I used REW too....to have had some proper resource from a trusted source outlining the main points to focus on and 'gaining perspective' would have been very useful for me, as I think with measurement tools like REW with UMIK things can get very detailed in terms of measurements and the quest to achieve a perfect looking measurement curve can be a bit of a rabbit hole of folly!

Then I don't understand how unpredictable movement of the mic during recording of each single chirp would not affect the result (I was thinking via slight Doppler frequency shifting due to the relative motion between sound source and mic), unless this effect is very small, or Audyssey accounts for this (not sure how that would be possible). By the way, maybe you could test a Leslie speaker (which utilises the Doppler effect) using your Klippel system at some point

I also have the same intuition as you bobbooo, and that's the approach I used for my EQ'ing measurements using UMIK & REW, but it's possible intuitions are wrong.

The amount of correction needed in the time domain would depend upon how good the speakers are to begin with. Why I asked is because a lot of people just see music listening as you listen to a wall of sound and spend most of their time thinking about if the bass is good, is the treble neutral etc... It is so much more than that with proper speakers, setup correctly, in a treated room. With good recordings I have music from floor to ceiling and all of front to back of the room. It sounds like surround sound basically- but with stereo.

I have been surprised when people have listened and asked me to turn off the center channel when playing a song with mainly vocals..yeah it's already off, it's the phantom image. These have been people who own proper speakers at home. So sometimes I wonder what people are doing.

I've had this same surround sound experience with my 2 channel system after I EQ'd it using REW & UMIK. I've not got involved in "Phase" at all and don't really understand that element at this point anyway, but I did EQ my system to account for 'small' speaker distance differences (20cm) in terms of applying speaker delay to closest speaker (0.6ms) as well as a small attenuation of speaker volume of the closest speaker based on UMIK measurements so that both speakers theoretically have the same volume at the listening position (0.4dB measurement difference)...not sure how significant those efforts were.....but I also of course EQ'd it based on the frequency sweep of course (from 20-550Hz). Yeah, I experienced that surround sound effect in terms of feeling like the sound is coming from all around you rather than the individual speakers and perfect phantom centre with my setup after I EQ'd it....but the positioning of a 30 degree angle from listening position to speaker was the biggest element of this phantom centre and surround effect, but I think the EQ enhanced that variable too as well as sorting out the general frequency response.

 

[bobbooo]

Unless you’re really flailing the mike, Doppler should be negligible. Frequency shift goes linearly with the ratio of the mike speed to the speed of sound, so something like .001.

True, it will likely be a small effect in most cases. I'm not sure if this frequency error might propagate up to something larger via the Audyssey compensation curve calculations though.

 

[KKoen]

 

[Putter]

@amirm : " I noticed that Dynamic EQ was on by default. I turned it off and the sound became a lot less pleasant (I used the "Reference" mode)."

Dynamic EQ, as stated by Audyssey, should work gradually as you decrease the master volume from reference 0dB where is completely off. A home theater system automatically calibrated by Audyssey MultEQ will play at reference level when the master volume control is set to the 0 dB position. At that level you can hear the mix at the same level the mixers heard it.

Audyssey Dynamic EQ is referenced to the standard film mix level. It makes adjustments to maintain the reference response and surround envelopment when the volume is turned down from 0 dB.

Considering that for movie I never listen at 0dB, I always have DEQ on and I find it much better; for music depends, maybe i will modify the offset level; who else has experimented with this?

more info about DEQ here: https://audyssey.zendesk.com/hc/en-us/articles/212347383-Dynamic-EQ-and-Reference-Level

I like to use it for music and find that it need to be set at different levels for different recordings. With my current speakers, JBL E20, it's set at -15db which is the minimum setting and will occasionally will find some recording where I switch it off although in most cases I'm OK with a little too much bass . With other speakers, it's varied anywhere between -5 to -15db again depending on the source.

 

[amirm]

Nice review @amirm The number one issue I have with (all) H/W solutions like this is the limited number of FIR filter taps to actually provide real low frequency room correction. Whether it is Audyssey, miniDSP, Trinnov, DEQX, or other H/W processors they all suffer from the same issue - not enough FIR filter taps below 100 Hz to really be called room correction.

Thanks for the informative post. Indeed cost constraints are huge as far as DSP power and with so many channel to process.

They have another restriction though: latency. If you have 48,000 points at 48 kHz, you incur a full second of latency. For an AV Receiver, it would then need to buffer full 1 second of video (24 frames of 4k video for example) which is not something they can or willing to do.

For straight music playing, this is not a problem of course since the audio playing a second later is not a big deal.

 

[ace_xp2]

I was under the impression that dirac at least did a combination of IIR and FIR, is this tap limitation possibly one of the reasons why?

 

[amirm]

I was under the impression that dirac at least did a combination of IIR and FIR, is this tap limitation possibly one of the reasons why?

Correct. IIR filters are recursive so don't need lot of memory and signal processing power and have low latency. They are more difficult to design and keep stable however.

 

[HighImpactAV]

Nice review @amirm The number one issue I have with (all) H/W solutions like this is the limited number of FIR filter taps to actually provide real low frequency room correction. Whether it is Audyssey, miniDSP, Trinnov, DEQX, or other H/W processors they all suffer from the same issue - not enough FIR filter taps below 100 Hz to really be called room correction.

As along time Audiolense user I agree with what you are saying. However, both Trinnov and Dirac use IIR filters for the bass. Trinnov lets you set the transition frequency. The default is 150 Hz. Various Dirac processors have a different number of taps to use. The maximum is 3048 taps per channel for 32 channels and is used by StormAudio. The cheaper processors using Dirac have less taps. I think Trinnov uses 4800 taps has a maximum based on the maximum filter length of 100 ms.

One thing rarely shown or discussed is that the phase is not corrected correctly in the bass by any system including Audiolense. The impulse measurement is measuring the energy of the highest frequency of the subwoofer and not the energy at the crossover. Using a dual channel measurement system like SMAART reveals this issue. REW can't show it. The Dirac Live Bass Control does a much better job and implements all-pass filters to address this issue. I've suggested a manual intervention for advanced calibrations when using Dirac Live Bass Control.

The frequency resolution of a 1024 taps filter @ 48 kHz samplerate is 48000/1024 = 46.875 Hz

It is also important to realize that the higher the sample rate, the less accuracy in low frequency correction with FIR filters if the number of taps stays the same. For 1024 taps, 192 kHz has a frequency resolution of 187.5 Hz. This is one reason why I prefer to use 48 kHz for all content. The article FIR Filter for Audio Practitioners states that frequency resolution of the filter is worse than you say: For a quick and rough estimation, we can multiply the frequency resolution by 3 (three) to predict the effective low frequency limit of the filter. 46.875Hz x 3 = 141Hz. This means an FIR filter with fs = 48kHz and N =1024 will be effective at 141Hz and above.

The largest Audiolense calibration I've done so far is 15 channels. I'm building my own new theater and hope to compare 24 channels of Audiolense vs 24 channel of Dirac correction, but might not get around to it for another year or so.

 

[AudioJester]

Dirac have just launched a bass management module including multiple subs, as an add on for compatible AVRs.
JBL synthesis wll now have it all!

 

[audioBliss]

I think that would be an excellent thing to do Amir, I've found EQ to be very important (recent experience), and I used REW too....to have had some proper resource from a trusted source outlining the main points to focus on and 'gaining perspective' would have been very useful for me, as I think with measurement tools like REW with UMIK things can get very detailed in terms of measurements and the quest to achieve a perfect looking measurement curve can be a bit of a rabbit hole of folly!


I also have the same intuition as you bobbooo, and that's the approach I used for my EQ'ing measurements using UMIK & REW, but it's possible intuitions are wrong.


I've had this same surround sound experience with my 2 channel system after I EQ'd it using REW & UMIK. I've not got involved in "Phase" at all and don't really understand that element at this point anyway, but I did EQ my system to account for 'small' speaker distance differences (20cm) in terms of applying speaker delay to closest speaker (0.6ms) as well as a small attenuation of speaker volume of the closest speaker based on UMIK measurements so that both speakers theoretically have the same volume at the listening position (0.4dB measurement difference)...not sure how significant those efforts were.....but I also of course EQ'd it based on the frequency sweep of course (from 20-550Hz). Yeah, I experienced that surround sound effect in terms of feeling like the sound is coming from all around you rather than the individual speakers and perfect phantom centre with my setup after I EQ'd it....but the positioning of a 30 degree angle from listening position to speaker was the biggest element of this phantom centre and surround effect, but I think the EQ enhanced that variable too as well as sorting out the general frequency response.

Even what might seem like small changes can have huge impact of what you hear. Time-aligning the distances and volume will for sure have a positive effect on imaging if that is off. Theoretically this would not be needed in a symmetrical, treated room with identical response from each speaker etc.. But no one has a perfect room with perfect placement so some correction is usually needed.

I've found that this is a good starting point for very good imaging if the room and speakers allow for anyone that is interested to play around with placement.

 

[mitchco]

Cheers @amirm. With respect to audio/video latency, there are a couple of options to consider:

If using JRiver for video playback, the convolution engine takes into account the linear phase FIR filter latency, so you get perfect lipsync.

If using the Win10 Netflix app that supports Dolby Digital, Atmos, etc. One can route the audio through Hifi cable & ASIO bridge and on the output, select Audiolense Convolver which then hosts a minimum phase version of the correction filters. @Juicehifi 's convolver is very low latency, so I do not notice any lipsync issues.

If using a Media player that supports DirectShow (like Windows Media Player or VLC for example), one can install audio/video codecs (scroll down to see features) and using the 2nd option above, pretty much decodes everything...