1. This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
    By continuing to use this site, you are consenting to our use of cookies.

    Dismiss Notice

Slow VS Fast roll off and Minimum VS Linear phase

Discussion in 'Sound Science' started by windowsx, Feb 16, 2017.
3 4
  1. pinnahertz
    That should be "reply". Darn autocorrect.
  2. WindowsX
    How about you try to put effort into making comprehensible post so other people can understand? Well, you can choose to end your discussion here too.
  3. pinnahertz

    Fair enough.

    If anyone else thinks my posts are incomprehensible, let me know and I'll make more effort.
  4. cjl

    Most DAPs on the market are these: http://www.apple.com/ipod/ , or  these: http://www.apple.com/iphone/ , or these: http://www.samsung.com/us/mobile/galaxy/ or similar, none of which have digital filter options. The DAPs on that page that you linked make up a vanishingly small minority of DAPs sold.
    Tennessee and Joe Bloggs like this.
  5. shadow_shooter
    Are you telling me after sending thousands on headphone gear, you're sourcing your music from a basic iPhone? It's been long since this thread started but I had the same question so it made sense to me. Apart from reading your painful, cynical answers, I've learned that the OP shouldn't ask a question because you're using the worst option for a source out there. If people have nothing to offer on the subject, why the heck to bother and bash others.
    1Q84 and Tennessee like this.
  6. pinnahertz
    Interesting....thread dead for 4 months, then you resurrect it with a complete misquote of me.
    No, read my post again...accurately.
    Why do you say that? And how do you know what I'm using?
    I don't recall bashing anyone. However, the above might just qualify as "bashing".
  7. bigshot
    I don't mind a slow roll off if the frequencies being rolled off are above my ability to hear. I don't mind a sharp brick wall if it's above my ability to hear and doesn't cause any problems down in the audible range. I don't generally worry about things I can't hear.
  8. castleofargh Contributor
    it certainly looks like an unfair grudge against WindowsX, but what you're not seeing in this thread are all the other threads he started with loaded questions that made many people here expecting a trap of sort. his growing resentment toward this sub section or some idea of objectivism also doesn't help him getting a warm welcome. maybe this one thread was genuine honest curiosity and this is all just a misunderstanding? maybe sometimes we get "lost in translation" and it creates problems? maybe if our answers had been different it would have turned into yet another wild goose chase? but it's a typical case of "fool me once shame on you, fool me 5 times, shame on me".

    as for the topic, when focused on a DAC the question has IMO fairly simple answers:
    -if you can hear a difference, you select the one you like most. the end
    -if you can't hear a difference, does it even matter which kind of low pass is used? no it doesn't.

    from a "scientific" perspective, the signal can be defined using several variables, and it's challenging to just go and decide that one variable is more important than another one. soon enough it comes back to what is the most likely to be heard, or the more likely to please a listener and we're back into personal hearing and preferences.
    because pre-ringing when it happens in the audible range can feel less "natural" than post-ringing, some decide that using a filter without pre-ringing is "obviously" better even when used at a frequency we fail to notice. this is in most applications, a laughable "inaudible that sounds better" paradox.
    some people like the guys behind Pono and Meridian only know to talk about how horrible ringing is(despite where it happens), and how the time domain is what makes a great sound(their opinion!). to the point that they don't mind degrading other variables and introducing a lot more aliasing just to marginally improve timing. on the other hand most of the industry when making albums will decide to go with brick wall filters that will ring in the area where the filter is applied(logically outside the audible range). in exchange it keeps the audible range with as little roll off as possible. and by better fitting the Nyquist theorem, will also result in as little aliasing as possible. so it's the method that probably agrees most with the guys who made digital audio possible. but in any case, we're mostly talking about small stuff and it should concern people the way small stuff do.
    now is ringing worst than aliasing or the other way around? hard to give an objective answer when dealing with an apple and oranges kind of question.
  9. shadow_shooter
    Thanks for your respective. I wasn't aware of the situation here. I have a good quality dap, which actually never mentioned digital filters as a thing. Dx200 but still has the options Windows X asked. I know Windows X from that sub. He didn't seem impolite or pushing. I think he was asking this question to figure out the setting to use for his rom. To me brick wall is the neutral one, reference or high freq heavy, I can't tell the difference between other filters always and when I can it is subtle that my brain could accept both as a signature of the headphone I am using.
  10. WindowsX
    I always ask out of curiosity. I believe the replies will be a lot different if asked by different members. Thank you for sharing your thoughts about my question by the way.

    After modding iBasso, Fiio, Cayin, and other hires DAPs for a while, I noticed most of them have digital filter options. I myself prefer slow linear phase ones for full body coherent sound and I always leave that thing to default for users to choose on their own. I'm just curious about what people in here pick for scientifically hoping to learn more about digital filters. I didn't know such question is inappropriate in sound science forum.

  11. bigshot
    Shadow Shooter, I'm very serious about sound and I use an iPhone as my portable player. Its performance is as good or better than most standalone CD players.

    Theory is great, but at some point it steps off into ideas without any relationship to reality. Whether or not one type of filter is better than another is moot, because even with cheap DACs you aren't going to be able to hear a real difference. If a DAC is properly designed and made it should be completely transparent however it chooses to roll off at the top. Sometimes options are just options. It doesn't mean they make a real difference.

    If you're looking for a way to improve actual sound quality, you'll do better to focus on the quality of your transducers and achieving a balanced frequency response. If you aren't happy with the sound you're getting, those two things will get you closer to your goal. If you're already satisfied with your headphones and response curve, then it's time to sit down and listen to some great music. Tweaking the unbearable won't get you better sound, but if you enjoy doing stuff like that, go ahead and have fun with it.
    Last edited: Jun 18, 2017
  12. Psalmanazar
    My experience is that linear phase, fast roll off filters on well-implemented DACs tend to be the most natural/neutral; the impulse ringing is masked by other noise and basically impossible to hear unless it is causing IMD in downstream gear which it probably isn't thus it is just the "best" as in "higher fidelity to the source" almost all of the time.

    Slow roll off tends to collapse staging a bit as castleofargh noted and sometimes have audible roll off depending on the filter. The stock Windows remixer filter sounds super rolled off compared to bypassing it with WASAPI or ASIO into neutral gear. Some slow roll-off filters sound much better than others though like the "F2 Linear phase, slow roll off" on the Grace M900 sounds much more like Grace fast roll-off filter than the stock AKM ak4490 slow roll off one sounds like the AKM linear phase, fast roll-off. Smoother sound is not always clearer/better sound.

    Minimum phase tends to destroy staging and soften the sound, even in the bass region, not just the rolled off treble. It never sounds right with speakers on DACs with selectable filters if the linear phase ones are well done. It is used as many DAC makers cut corners. They will stick caps in the signal path or use DAC chips for marketing purposes that are very vulnerable to electrical interference and noise like Wolfsons with their built in op-amps or ESS Sabres (new 9038 is not supposed to be as vulnerable but who knows) with both the vulnerability and timbre issues. They will choose these chips instead of older tried and true ones (AD1955, CS4398, various older TI chips) for marketing purposes and tech specs instead of being able to implement them well (Wolfsons and Sabres notoriously difficult). To cut the time/cost/skill/effort instead of overbuilding the power supply with enough filtration capacity or writing their own DSP filters (this skill takes years to develop) if the stock ones aren't any good, will often just use a minimum phase or slow roll off filter that's often built into the chip to make an "acceptable" or "pleasant" sounding product to sell to the consumer audiophile market.

    Non-oversampling DACs or NOS emulation filters (AKM's "Super slow" filter emulates NOS impulse response) tend to horrifically mangle the staging, separation, tone, and timbre even more so than minimum phase oversampling filters even if the measured roll-off looks similar. There are DACs though where the NOS low pass filter does actually sound more "coherent" than the linear phase oversampling one but that is due to how poorly done the oversampling filter (or the entire DAC) actually is like the Holo Spring DAC or certain ESS 9018 based DACs which have tons of treble timbre issues that a NOS filter conveniently just shoves under a rug as described in the previous paragraph.
  13. bigshot
    What is the size of the phase error you're talking about? If it's audible, it must be pretty large- over 1-3ms. Also what do you mean by staging? You mean the channel separation? I've never found channel separation to be a problem, even on cheap equipment. How can phase error show up more on speakers than headphones when the speakers and room are introducing much larger timing errors themselves? I'm no expert on all this stuff, but I do know what humans can and can't hear. If it's audible, it's a large error. We're talking the millisecond range right?

    I use a Sony blu-ray player, Yamaha AVR, iMac, Mac Mini, a recent iPhone, iPod Classics, an Oppo BDP103D and an Oppo HA1. I've compared them all, and they all sound exactly the same. I would expect any DAC to be transparent, but I only know the ones I have. I guess I've been lucky and haven't ever run across a bum DAC. I know I've had several Apple branded DACs, Wolfson (iPod) and Sabre (Oppo). I don't know what's in the Sony or Yamaha, but they sound exactly like the Apples and Oppos. I know there used to be lousy DACs in PCs, but I've been told that isn't the case any more because DACs are well designed now and mass produced inexpensively to much higher standards than in the past.
    Maybe they're still using crappy DACs in brands I haven't tried.
    Last edited: Jun 18, 2017
  14. Psalmanazar
    Minimum phase and NOS analog filters produce noticeable errors in stereo separation as they are uneven in the time domain, which is why they typically aren't used.

    Please don't derail the thread. Minimum phase filtering is measurably inferior in both the time and frequency domain and it is used to make a more sellable, "better sounding" product in many DACs. I'm not going to debate you on this and am not the person to do so as I wouldn't buy any DAC with a minimum phase filter myself. Ask Ayre, Meridian, Meier, C-Entrance, or any other manufacturer who makes DACs with minimum phase filters if you want to get to the bottom of why they have minimum phase filtering.
  15. castleofargh Contributor
    well one clear variable to consider is our own hearing. the higher we can hear in frequency, the more likely we are to be able to notice some variations from filters. at least when sending 44.1khz to the DAC. I remember as a kid going crazy over the high pitched noises from PSU, lights, so called mosquito repellents that made ultrasonic sounds and at the time only ever repelled me and our cat from the room. I'm fairly confident that I would have had a different opinion about how noticeable some filters were back in those days. but now that 16.5khz is but a quiet tone at normal level and 17khz almost total silence for me, obviously whatever happens at or above those frequencies doesn't really move me.
3 4

Share This Page