Pre and Post Effin Ringing and shit like that

Discussion in 'Blind Testing and Psychoacoustics' started by ultrabike, Aug 17, 2016.

  1. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,989
    Trophy Points:
    113
    Location:
    Irvine CA
    LOL! Done.

    Well, they are calling the non-zero delay filters "linear phase". Linear phase does not skewer or smear the phase. In fact, in the video they kind of mumble jumbled about how linear phase does not f**k up the phase (kind of over-complicated-ly).

    If the simple/low taps/whatever EQ filter is not linear phase, then why call it linear phase in the video? It is not. Specially when you are making a case about the benefits of minimum phase vs linear phase. To me this is too misleading to ignore.

    They go overboard to talk about how the pre-ringing creates some sort of reverb. That's AFAIK complete BS that I double checked in that Octave/Matlab script there. If the case is about subjective evaluation, I checked the processed files. I don't hear that reverb.

    The EQ program might not be a total failure. But if the linear phase settings are generating reverb artifacts, maybe they need to revisit the implementation.

    Maybe I'll do a few linear and min phase notches to compare. But bare in mind that notches may indeed generate reverb like sound (in either type of filter). That's is not because of the linear/minimum phase deal, that's due to the notch itself. It's coloring the signal.

    Furthermore, note that their comments extend to low pass and high pass filters. A low pass is what I did. I can lower the pass band to half the bandwidth (10 kHz) and see what you guys think too. Could do a high pass as well. Let me know what you guys want to check. I don't have a horse in the race BTW.
     
    Last edited: Feb 24, 2017
  2. Armaegis

    Armaegis Friend

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    8,181
    Trophy Points:
    113
    Location:
    Winnipeg
    Ah, y'see, mentally I was still being kinda terminology ignorant. You're arguing for the proper usage of the terms. To me, the things they were using might as well have been called "the bastardized filter we happen to call linear phase" and "the bogun filter we call minimum phase but mebbe we don't know what those words mean".
     
  3. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,989
    Trophy Points:
    113
    Location:
    Irvine CA
    LOL! Yup.

    But it's not just the proper terminology IMO. Folks looking at that video can get the wrong impression about linear phase filters (and their "pre-ringing" evilness as I see over and over again). Just trying to set the story straight.
     
    Last edited: Feb 24, 2017
  4. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,989
    Trophy Points:
    113
    Location:
    Irvine CA
    BTW, nothing against bi-quads. My entry level mini-DSP (w/o the Dirac goodness) does a great job, and love it.

    Also impressed @Hrodulf stuff implements a linear phase filter for highs and possibly mids. That is IMO high-end-ish.
     
  5. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,989
    Trophy Points:
    113
    Location:
    Irvine CA
  6. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,989
    Trophy Points:
    113
    Location:
    Irvine CA
    f**k!!!! You guys don't know how pissed I'm right now!

    "As a side-effect of changing the level balance of different bands of frequencies, conventional EQ also changes their phase relationships. So low frequencies can be delayed with respect to high, or vice versa, depending on how the EQ parameters are set. We call this 'phase shift'."

    No it f'ing doesn't!!! f**k!!!! A linear phase filter will only result in delay and transcients. Once the transients die, all phase relationships across the audio spectrum should be f'ing aligned!

    "'Minimum phase' means that the amount of phase shift that happens is as little as possible."

    DAMIT NO! Minimum phase means the latency will be kept at a minimum and transients will also be kept at a minimum. But phase relationships after the transients will change, unlike the case for linear phase.

    "The expression 'transient smearing' is sometimes used, meaning that when there is a sudden attack in the signal, such as a kick or snare drum, then since different frequencies will be delayed by different amounts, then the transient will not attack in the same way as it did before filtering or EQ"

    WTF!!! There is no "transient smearing", there are just f'ing transients. Read a f'ing book on circuit analysis. Hell, any f'ing random book on differential equations should explain what a f'ing transient is. It's due to the initial conditions of the circuit! If a bandlimited signal is applied and the filter BW is larger than the BW of the signal there is no f'ing problem.

    "Linear phase filters and equalizers have the unfortunate side effect of creating audible 'pre-ringing' in some types of signal, such as the attack of a snare drum."

    TOTAL AND COMPLETE BULLSHIT! pre-ringing affects transient behavior ONLY! The audible part of a drum is f'ing bandlimited and once the transients die and the system is in steady state if you hear a difference it's because the minimum phase filter fucked up!

    Not reading that shit anymore.

    Get it guys. This is bullshit. And is depressing to see large companies like Cirrus spread this crap.
     
  7. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,989
    Trophy Points:
    113
    Location:
    Irvine CA
    Here, which one is more accurate?

    filter_stuff.jpg

    For this example the original is random shit passed through a filter at 0.45 fs/2 to bandlimit the input signal. Both linear and equivalent minimum phase filters have a passband of 0.5 fs/2.

    Now again, f'ing tell me if the red crosses are not right inside the little blue circles and the black crosses all over the f'ing place due to phase distortion?

    This would not be as annoying if it wasn't the nth time this was brought up, and large companies did not created freaking marketing banners in their webpages about it.

    Cirrus WTF!!!!?
     
  8. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,989
    Trophy Points:
    113
    Location:
    Irvine CA
    Ok. I'm more calm down now. Sorry for losing my marbles here. I'm just tired of reading the same stuff over and over again. And now even from large respected companies such as Cirrus Logic. It's just a bit disappointing.

    https://www.cirrus.com/solutions/playback/hifi/

    Nobody in the industry cares obviously. Costumer is always right I guess. *sigh*
     
  9. Armaegis

    Armaegis Friend

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    8,181
    Trophy Points:
    113
    Location:
    Winnipeg
    There's also this crossroads known as marketing that buggers things up, compounded by the fact that they don't talk with the engineers, again compounded by the fact that the engineers generally suck at communicating in the first place.

    I'm not into the game deep enough to be able to dig into the math, and I don't know all the proper terminology... but in some cases I think that actually helps because the terms all get abused and literally misused, so I wind up getting the base concepts of how some things happen, but I don't get thrown off when someone uses the wrong word to describe something.
     
  10. monacelli

    monacelli Friend

    Pyrate
    Joined:
    Jul 23, 2017
    Likes Received:
    1,650
    Trophy Points:
    93
    Location:
    The Land of Enchantment
    You raise a good point. We've seen manufacturers use the term "minimum-phase," when what they mean is something like "linear phase with short delay" or "slow roll-off." This is just one example. If you're familiar with the technical vernacular, this can be quite confusing. So when reading marketing materials, it's definitely helpful to keep in mind that it more than likely wasn't written by someone with engineering expertise. And the converse isn't necessarily any better. A lot of the sciency white paper stuff released by companies is just as useless as the ad copy. Schiit is in a unique position that one of their lead engineers is also a marketing expert. I imagine it's hard to strike the right balance even for someone who does have experience in both worlds.
     
  11. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,989
    Trophy Points:
    113
    Location:
    Irvine CA
    Nah guys. Those morons are definitively talking about minimum phase. A linear phase filter is symmetrical. Those Cirrus plots on minimum phase are most definitively not symmetrical. Those marketing folks drank the kool-aid and believed some opinionated rando more that probably their own engineers.

    I can even see things going like this:

    Larry Marketing: Hey. John.

    John Engineer: What now Lary?

    Larry Marketing: Some vocal folks at such-and-such forum are all about this minimal face filter stuff. Like their are outraged that we are not using it.

    John Engineer: Wait. What? You mean minimum phase filters?

    Larry Marketing: Yeah. Whatever. That. It like they are all bend out of shape because of this phone ringing. I don't get it. What do phones have to do with it?

    John Engineer: Maybe they mean the sync function approximation filter's pre cursor taps. They are optimal filter taps that get used to interpolate using samples from the future, along with the post cursor taps that mix in samples from the past. We can interpolate using samples from the past, but can't interpolate using samples that have not been played from the future. So we delay all the taps and now all samples are in the past. There is a latency in the output, but the system is now causal and realizable. Now, all taps before and after the main tap look like they are oscillating because of the sync filter behavior. But that's what you need to have optimal reconstruction.

    Larry Marketing: OK John. That made no sense to me. What's a sink filter? Hell. What's a filter? You mean like the coffee filters in the break room?

    John Engineer: *sigh* No Larry. It's basically a filter to optimally fill the gaps between samples when upsampling to make it easy on the DAC analog reconstruction filters. The digital and analog reconstruction filters reproduce or reconstruct the signal in continuous time from recorded samples in discrete time.

    Larry Marketing: John, you lost me completely at discrete time. Why does time have to be discrete about anything?

    John Engineer: OK Larry. In a nutshell, what those randos at such and such forum want is going to hit performance. It makes no sense.

    Larry Marketing: They are paying the bills John. So is there a way to give them what the want? Like give them options and stuff. I mean we could sell that right?

    John Engineer: Well. I guess we could do that. Give them both an optimal causal and windowed filter with finite samples, and it's equivalent minimum phase filter.

    Larry Marketing: Is it a lot of effort from your team John?

    John Engineer: Hell no. It's not even going to consume hardly any ROM. And if you want to put that in the latest part with internal flash, it's like a slam dunk.

    Larry Marketing: Then do it John.

    And the rest is history.

    That's fine. But now it's all the sex in their website. It's depressing.
     
    Last edited: Aug 1, 2018
  12. Armaegis

    Armaegis Friend

    Pyrate BWC
    Joined:
    Sep 27, 2015
    Likes Received:
    8,181
    Trophy Points:
    113
    Location:
    Winnipeg
    Sex is what pays the bills Larry. When someone asks you for minimum phase, the correct answer is "how wide?"
     
  13. famish99

    famish99 Friend

    Pyrate BWC
    Joined:
    Dec 26, 2017
    Likes Received:
    1,789
    Trophy Points:
    93
    Location:
    Austin, TX
    This conversation needs to be more brotastic and it would be more accurate.
     
  14. bazelio

    bazelio Friend

    Pyrate
    Joined:
    Nov 16, 2015
    Likes Received:
    3,423
    Trophy Points:
    113
    Wait, so what constitutes a "conventional EQ" circuit? Is there consensus on this? Ignorantly I'd think a simple LC that combines the phase shifted waveform with the original signal. But I dunno, and am not trying to argue. Just that the original statement doesn't seem soooo outlandish.
     
  15. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,989
    Trophy Points:
    113
    Location:
    Irvine CA
    Based on the context, I think they meant conventional DAC digital filters, which are linear phase AFAIK.

    But yes, there are no real conventional EQs. There are all kinds of them.

    BTW, a simple LC analog filter is equivalent to a second order IIR filter (two poles, and one or no zeros). All those conventional digital filters are FIR. There is actually a "bilinear" transformation that can be used to go back and forth from s (analog) and z (digital) domain. It does not always works 100% though.

    IIR filters (infinite impulse response), have poles and zeros. FIR filter (finite impulse response), only have zeros. The advantage is that FIR filters are unconditionally stable (will not blow up or oscillate). IIRs can oscillate. Furthermore, it's hard to make an IIR filter with linear phase. You can only approximate it. FIRs can be linear phase. In many applications linear phase is highly desirable as it reduces phase distortion in the passband.
     
  16. bazelio

    bazelio Friend

    Pyrate
    Joined:
    Nov 16, 2015
    Likes Received:
    3,423
    Trophy Points:
    113
    Ok I dunno, but what I do know is my ADD won't permit me to sift through all of it to figure it out. I once worked for a Cirrus competitor. We killed them in our market, and I never thought much of their designs. But, at least back then, they weren't this stupid. :)
     
  17. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,989
    Trophy Points:
    113
    Location:
    Irvine CA
    LOL! I don't think Cirrhosis are stupid. But their marketing made a boo boo there. It happens.

    Their CODECs are pretty good. And their TOTL DACs and ADCs seem fairly respectable. I think Prismsound uses them. One could argue about the sound of their products. But specs and measured performance is IMO not shit.

    I would be more bend out of shape if all they offered was that minimum phase filter. But they do have the linear one. And to be fair, I think both AK's Mom and Tits and Instruments drank from the same kool-aid and offer the minimum phase awesomizer as well. You just change some settings through I2C and boom! Music fucked! Or unfucked.

    It's not necessarily that big of a deal though. The minimum phase filter is probably small potatoes compared to shit speakers and headphones which are not in short supply.

    I feel the whole industry's marketing drank from the same mind fuckifing concoction though.

    That said, there are a few things were the minimum phase filter excels at though: Lowest latency. In some cases maybe that's the holy grail. Maybe marketing should just change the spin of it.
     
    Last edited: Aug 2, 2018
  18. Erikdayo

    Erikdayo Friend

    Pyrate
    Joined:
    Apr 29, 2017
    Likes Received:
    424
    Trophy Points:
    93
    Location:
    Boston, MA
    This thread has been a good read and also has me re-evaluating my filter choice on the RME ADI-2 DAC. I went through them quickly earlier to see if there is anything I can hear. Tried this when I first got the DAC as well. I have a really difficult time hearing differences outside of the very noticeable roll off on 'NOS' and the much more slight roll off on 'Slow' and 'Short Delay Slow'.

    It's all very interesting whether the differences in ringing are actually audible to me or not. I'll try to do some critical listening over the weekend to see if I can pick up more than the frequency response changes.

    My current preference is 'NOS', but they all sound pretty good to me.
     
  19. ultrabike

    ultrabike Measurbator - Admin

    Staff Member Pyrate MZR
    Joined:
    Sep 25, 2015
    Likes Received:
    8,989
    Trophy Points:
    113
    Location:
    Irvine CA
    Yeah. Like I said, relatively speaking the filter choice its sort of small potatoes IMO.

    My tantrum there has more to do with the claim that the "pre-ringing" is the end of the world, which is a complete bunch of non-sense.
     
  20. purr1n

    purr1n Desire for betterer is endless.

    Staff Member Pyrate BWC
    Joined:
    Sep 24, 2015
    Likes Received:
    92,279
    Trophy Points:
    113
    Location:
    Padre Island CC TX
    I didn't realize upon initial read that the audio master class article conflated (confused) pre-ringing with group delay of signals above the xover point for the HPF.

    I'd like to see an analysis of the waveforms from the EQ program that he used. A lot of EQ software is screwy and doesn't necessarily do what they claim. It's possible the EQ's linear phase EQ setting intentional mimicked the group delay in the passband of an analog HPF. Still, this should not result in such a drastic perceptual difference, if any at all. Bass wavelengths are super long and a shift of an inch here or there isn't going to make a difference. People who build subwoofers know this.

    EDIT: @ The software EQ is probably IIR to reduce latency. Still doesn't explain the massive perceived group delay.
     
    Last edited: Aug 3, 2018

Share This Page