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.
First
 
Back
1 2
4
Next
 
Last
  1. bigshot
    I don't know if any of you have ever done slow tone sweeps on your ear, but it's likely you have a dead spot in your hearing. My ears both have a super narrow dead zone. When the sweep reaches a certain point, first one ear goes dead then comes back and then the other one does the same. Maybe it's similar to a blind spot in the eye. Dunno.
     
  2. bigshot


    It's derailing the thread to ask how phase error in a DAC on such a minute scale makes any difference when speakers and rooms have a huge amount of similar kinds of errors? Why is it derailing the thread to ask if "measurably inferior" performance is actually audible? Are we entertaining ourselves here by talking purely in theory, or are we shopping for a nice DAC for a bat?

    Quite frankly, if I can't hear it, I refuse to worry about it. I'm not going to pay more money and jump through hoops unless it actually makes a difference to human ears. That's why I don't pay any attention to jitter (inaudible), lossless vs high bitrate AAC (inaudible), high sampling rates (inaudible) and high bitrates (inaudible). From the tone of your response, I'm guessing I can add
    minimum phase filtering to the list. The thing I don't understand is why people give so much attention and energy to things that don't matter while not addressing the elephant in the corner- speakers sound better than headphones, getting better speakers are usually the best way to improve sound quality, balancing your frequency response is the best tweak you can do without buying new equipment, and treating your room to avoid unwanted reflections is the best way to improve the timing of your system.
     
    Last edited: Jun 19, 2017
    Sinarca likes this.
  3. pinnahertz
    Doing this with headphones or speakers?
     
    castleofargh likes this.
  4. Psalmanazar
    Believe what you want to believe about everything you just said yet they measure different, use different parts that sound different, and people with more experience than you can hear the difference. Just because you can't or choose not to, doesn't mean they can't.
     
  5. bigshot
    Headphones. Oppo PM-1s

    It's interesting that you seem to think that experience can improve on perceptual thresholds. I'm quoting best case figures for just detectable differences established by peer reviewed scientific studies. I don't know of any studies that show that someone can hear super audible things if they concentrate hard enough or study real hard. A threshold is a threshold, isn't it?

    If the error is clearly measurable, like you say it is, it's also quantifiable. Why would asking what amount of error we're talking about be considered "derailing the thread"? Just because you don't want to admit that it's pretty clearly inaudible doesn't mean you can hear it.
     
    Last edited: Jun 19, 2017
  6. Psalmanazar
    You can hear minimum phase and slow roll off filters collapse the stage by paying attention to the locations of instruments when abing between filters and DACs. This is much easier to hear than the roll-off most of the time. It's not hard if you aren't deaf in one ear, focus on the location of one instrument, and actually use a setup that can reproduce panning and get the samples to the amp relatively unmangled. If you're using USB and do not have a DAC that has exclusive mode enabled in the firmware than tough luck.

    PM-1s are planar headphones which are bad at reproducing pannings that aren't hard so the effect of a minimum phase filter will be much less. Try it with a lossless, Redbook or higher recording with lots of pannings (orchestral recordings, old jazz, and twin guitar rock/metal are good choices) and a pair of Sennheiser HD 600/650/800 or even better a well set up pair of nearfield speakers (minimizes the room) and then switch between the filters on a DAC with switchable ones (or use an plugin with a low-pass minimum phase filter if your DAC only has linear phase like most DACs if you want to get the general effect) and you can hear the time domain being out of phase very easily with minimum phase or NOS filters.
     
    Last edited: Jun 19, 2017
  7. castleofargh Contributor
    what technical aspect could possibly lead you to that idea?
     
  8. pinnahertz
    This implies that it's impossible to create two identical minimum phase filters that match closely enough not to affect localization. Localization of a virtual sound source is accomplished by means of a differential in amplitude and/or a differential it arrival time. So, for the apparent location of a sound to change there must be a difference between the two channels in amplitude, arrival time, or both, that is universally attributable to the use of minimum phase or NOS filters.

    I just don't think so.
     
  9. bigshot


    Pinnahertz has a good point. I was assuming you were talking about a timing error. But maybe you're talking about channel separation. I can't imagine there being a problem with channel separation, but I can't imagine a timing error in the audible range either.

    Can you measure the phase distortion? What does it measure? Does it affect all frequencies equally? Or are some more affected than others? Answers to questions like these will help me figure out for myself if it's audible.
    I hope I'm not being rude by not just taking your word for it. But your suggestion that I go out and buy all new headphones and speakers to be able to hear it makes me suspect what you're talking about isn't the least bit audible.

    By the way, I listen to classical and jazz all the time, and I find that the placement of instruments in soundstage is entirely determined by the original mix, not by my equipment. Some recordings have a very defined soundstage, some don't. It doesn't matter what player I use. I've directly compared every player I have and they all sound the same.

    My guess is that you've got some normal human expectation bias going on. That's common when comparing two aurally identical sounds. It's perfectly fine if you want to improve purely theoretical things in your system because doing that interests you. But my purpose is to put together a system that sounds as perfect as possible to human ears. It's a big enough job dealing with the clearly audible aspects. I don't have time to worry about theoretical sound. And I'm sure not going to go out and buy new headphones and new speaker system to theoretically be able to hear the theoretical sound you're talking in theory about.
     
    Last edited: Jun 20, 2017
    seanwee and Sinarca like this.
  10. pinnahertz
    If "phase distortion" (perhaps actually "group delay"?) were identical in both channels the image would not be affected. There's zero chance that one entire channel is out of time with the other. Channel separation is never affected within the digital stream, so it would have to be an analog cross-channel coupling, but that would have to get pretty horrible to change the perceived image placement, on the order of 35dB or less. I would never expect that in today's equipment. Group delay has to be pretty significant to become audible, so I'm just no sure what he's on about.
     
  11. bigshot
    Yeah, I was figuring he was talking about group delay, but he hasn't described it as that. I'm unclear about what he means by "pannings" too. I suppose he means the kind of thing you'd run into in cheap analogue graphic equalizers, but I've never encountered that sort of thing in digital audio.
     
    Last edited: Jun 20, 2017
  12. Psalmanazar
    That is the downside of minimum phase filters; the change in arrival time of different frequencies is not uniform and greatly affects the perceived stereo image. Yes DACs with minimum phase filters will apply it to both channels and yes it collapses the stage almost always when compared to the linear phase one in the same DAC. There does not need to be a difference between channels, the stereo image is already messed with. Try ABing it yourself on something with switchable filters and you can hear instruments in a multitrack recording no longer be as well delineated and will usually move closer to the center. NOS DACs tend to be much much worse things than ones with minimum phase filters though.

    There are some youtube videos that show the benefits of minimum phase filters applied to single instrumental tracks in mixing but they are not good for stereo recording or listening overall. The one benefit for consumers is that they can tame overly forward/bright DACs like ESS Sabre 9018 or poorly implemented Wolfsons. They are also good for brighter headphones like the stock HD 800 or Beyerdynamics. Yes it's not very high-fidelity to mangle the phase but it can result in more pleasing sound. Best thing to do though is simply not use super bright gear like that you have to work around the flaws like that though.
     
    Last edited: Jun 24, 2017
  13. Arpiben
    As far as I am concerned, it is hard to figure out how stereo image can be affected regardeless of filter type linear/minimum phase used in DACs.
    Especially when assuming that:
    • stereo has been properly implemented before reaching DAC level,
    • Left @ Right Channels are receiving same symetrical digital and analog treatment inside DACs
    Do you mind elaborating more, please?
    I am more prone to rather suspect DACs` digital signal processing for such behavior....
     
    Last edited: Jun 25, 2017
  14. pinnahertz
    I"m more prone to suspect expectation bias. At least some testing of linear phase vs minimum phase audibility was done here. It was not exactly conclusive or the results some would expect.
     
  15. bigshot
    I think it's repetition of snake oil sales pitch used by some DAC manufacturer to make potential customers worry about aspects that no human can actually hear.
     
First
 
Back
1 2
4
Next
 
Last

Share This Page