Announcement

Collapse
No announcement yet.

PipeWire Should Be One Of The Exciting Linux Desktop Technologies For 2019

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts
    Weasel
    Senior Member

  • Weasel
    replied
    Originally posted by pininety View Post
    That has nothing to due with a cheap or broken DAC but is just the nature of digital to analog sampling. Every DAC is bandwidth limit, in audio usually with a lower frequency of 0. So every signal you give it will be folded into that. If you have a digital signal width a higher bandwidth than that, you need to resample that first to get it even into the DAC and hence you have to adjust the frequency range with a digital low pass. If you do not do that, you will get the same aliasing you would get from a ADC with wrongly configured anti aliasing filter because the two signals are indistinguishable from a low bandwidth perspective.
    Again, I'm not talking about the digital aliasing. The analog filters (low pass) are not "exact" and neither is the conversion. That's why the quality of the DAC even matters.

    Leave a comment:

  • oiaohm
    Senior Member

  • oiaohm
    replied
    Originally posted by Weasel View Post
    And I will drop the bomb which is based on 100% fact, whether you (anyone who reads) like it or not: If you hear stuff above 20khz, you are either lying, or you hear aliasing from the DAC or speakers. Period.
    This I am sorry to have to say is right in most cases there is one special case where what you said is wrong..

    There is only one case where a human will be hearing above 20Khz. To hear above 20khz you have to have had surgery to treat glue ear and it was decide to do a particular form of the operation resulting in a resonance tube being created in side your ear the result is a hear able 1khz block somewhere between 20khz to 25khz due to that being resonated into a lower range. I have had this done I have hearing 21khz to 22khz this is particularly bad when someone blows a silent dog whistle rated at 100db inside that range and basically drops me to ground in pain..

    Something people are not aware of who don't have this is high power 150db silent dog whistles are 20khz to 22khz you do not want to hear in that range and go anywhere a dog park because some idiot will be using them wrong. Basically hearing higher than 20khz suxs big time as it restricts where you can go due to usage of dog whistles .. Yes this also explains when people don't know what they are doing they will be blowing and blowing a silent dog whistle and the dog is staying 100 metres away from them as the dog is not mad enough to come up to point blank on a 150db noise source..

    Ultrasonic tape measures are 22.5 to 25khz can can be over 160db. Ultrasonic cleaners are normally 25khz at times well over 200db. Basically above the area of normal human hearing is quite a nosily place.

    So hearing over 20Khz is a pain in the ass and will place restrictions on where you can go. The process is done because in particular cases of glue ear it results in getting 99% of normal hearing range functional instead of 60/70 without the built in resonance tube.

    Most people who claim they can hear about 20Khz are lies. The ones like me who really can wish we could not and absolutely would not want over 20khz out our sound systems as hearing a tone up in that range is normally go to high alert and find the source to move away from sound source to avoid getting hurt by it..

    Leave a comment:

  • pininety
    Senior Member

  • pininety
    replied
    Originally posted by Weasel View Post
    No, aliasing means that frequencies alias some other region that they aren't supposed to
    While this is true, the person you are throwing it at did not contest that. What that person said is "Increasing the bandwidth cannot result in more aliasing than lower bandwidth". And that is true.

    Originally posted by Weasel View Post
    Your DAC may alias high frequency content if it's cheap, or broken, or whatever.
    That has nothing to due with a cheap or broken DAC but is just the nature of digital to analog sampling. Every DAC is bandwidth limit, in audio usually with a lower frequency of 0. So every signal you give it will be folded into that. If you have a digital signal width a higher bandwidth than that, you need to resample that first to get it even into the DAC and hence you have to adjust the frequency range with a digital low pass. If you do not do that, you will get the same aliasing you would get from a ADC with wrongly configured anti aliasing filter because the two signals are indistinguishable from a low bandwidth perspective.

    Leave a comment:

  • Weasel
    Senior Member

  • Weasel
    replied
    Originally posted by rewik View Post
    Sir, I have an issue with you using the term "aliasing". In signal processing aliasing means artifacts caused by sampling a signal at a lower frequency than the Nyquist theorem states as the minimal frequency required to reproduce the signal in question. Therefore it's mathematically *impossible* to *introduce* aliasing by *increasing* the sampling frequency for the same input. Even moreso since outside highly-specialized lab equipment (Brüel & Kjær) I've never seen a microphone that can reliably work beyond 20 kHz (we tended to use Behringer ECM8000 - very nice up to 20kHz, useless after that).
    No, aliasing means that frequencies alias some other region that they aren't supposed to. It's in the name. Your DAC may alias high frequency content if it's cheap, or broken, or whatever. Same with speakers/headphones/monitors.

    Here educate yourself: https://www.soundonsound.com/sound-a...what-causes-it

    I'm talking about the aliasing that happens in the DAC, when it converts to analog, not digital aliasing. For example almost every integrated DAC/soundcard in motherboards will alias above 16khz. Yes you "hear" a tone above 16khz alright, so you think they're super good, but it's not higher, it starts to sound like Shepard Tone actually, due to aliasing. That doesn't mean it's "quality" but casual normies won't even know the difference.

    And I will drop the bomb which is based on 100% fact, whether you (anyone who reads) like it or not: If you hear stuff above 20khz, you are either lying, or you hear aliasing from the DAC or speakers. Period.

    Don't come with or "golden ears", you're not a dog, this is a biological fact. There are tests you can perform, you know, if you truly think you can hear like a dog, I'm sure you will break the Guiness World Record and become famous so go do the tests, unless you know it's pure nonsense and you're deceiving yourself.
    Weasel
    Senior Member
    Last edited by Weasel; 04 February 2019, 12:54 PM.

    Leave a comment:

  • ssokolow
    Senior Member

  • ssokolow
    replied
    Pinging Michael. Spurious spam flag on a linkless post again.

    Leave a comment:

  • ssokolow
    Senior Member

  • ssokolow
    replied
    Originally posted by rewik View Post
    Sir, I have an issue with you using the term "aliasing". In signal processing aliasing means artifacts caused by sampling a signal at a lower frequency than the Nyquist theorem states as the minimal frequency required to reproduce the signal in question. Therefore it's mathematically *impossible* to *introduce* aliasing by *increasing* the sampling frequency for the same input. Even moreso since outside highly-specialized lab equipment (Brüel & Kjær) I've never seen a microphone that can reliably work beyond 20 kHz (we tended to use Behringer ECM8000 - very nice up to 20kHz, useless after that).
    While I cannot rule out that every single 96kHz record out there is mastered using microphones that can pick up noises at 48kHz and higher and somehow the people doing the recording didn't know about it and forgot to use an anti-aliasing filter, I find that higly unlikely. Heck, I'd find it surprising to find aliasing on a regular 44kHz-sampled audio file, unless there was incompetence involved.
    It's more likely that the problem is being caused by either a resampling algorithm never having been tested for ultrasonic frequencies (eg. trying to use something only ever tested for 48KHz to downsample 192KHz to 96KHz and triggering a bug) or speakers not rated for ultrasonics behaving badly when driven with certain combinations of frequencies.

    (Reminds me of that explanation I once received for the myth that it's impossible to get "warm" audio out of a CD and digital amplifier. Apparently it got its start when, in the early days of CDs, audio engineers forgot to omit the "boost the high frequencies to compensate for vinyl limitations" step when mastering CDs, resulting in a harsh sound when the CD accurately reproduced what was put into it. That said, I'd love to see someone design and code/build a distortion filter to offer the vinyl+tube sound on any digital audio with the push of a button.)
    ssokolow
    Senior Member
    Last edited by ssokolow; 04 February 2019, 08:42 AM.

    Leave a comment:

  • rewik
    Junior Member

  • rewik
    replied
    Originally posted by Weasel View Post
    You are completely wrong. Interference does not change frequency.

    Mind you, I trust you that you hear a difference here, but it's not because it's "higher quality". In fact, it's the opposite. It's probably because of aliasing. In short, the signal is worse at 96khz than 44khz or whatever.

    So the difference you hear is because the 96khz is degraded and has aliasing from higher frequencies.

    Sorry to burst your bubble. Just physics and math. No amount of snake oil will change that.
    Sir, I have an issue with you using the term "aliasing". In signal processing aliasing means artifacts caused by sampling a signal at a lower frequency than the Nyquist theorem states as the minimal frequency required to reproduce the signal in question. Therefore it's mathematically *impossible* to *introduce* aliasing by *increasing* the sampling frequency for the same input. Even moreso since outside highly-specialized lab equipment (Brüel & Kjær) I've never seen a microphone that can reliably work beyond 20 kHz (we tended to use Behringer ECM8000 - very nice up to 20kHz, useless after that).
    While I cannot rule out that every single 96kHz record out there is mastered using microphones that can pick up noises at 48kHz and higher and somehow the people doing the recording didn't know about it and forgot to use an anti-aliasing filter, I find that higly unlikely. Heck, I'd find it surprising to find aliasing on a regular 44kHz-sampled audio file, unless there was incompetence involved.

    Leave a comment:

  • Weasel
    Senior Member

  • Weasel
    replied
    Originally posted by vsteel View Post
    I will respectively disagree with you. As mentioned in my other comment, there are constructive and destructive interference that effect the waves we can hear. I also have done testing on the very thing you have mentioned and I can hear a difference up to 96Khz 24bit and and others that I have tested. I have not tested anyone including myself that could tell a difference between 192Khz and 96Khz though maybe I don't have the equipment for it. I didn't tell them before which levels they were hearing and they could tell. Yes it must be mastered correctly but rather than read internet articles a person needs to actually listen and see if they can tell a difference. A person might not have good enough equipment, listen close enough or just not have the hearing that can tell a difference.
    You are completely wrong. Interference does not change frequency.

    Mind you, I trust you that you hear a difference here, but it's not because it's "higher quality". In fact, it's the opposite. It's probably because of aliasing. In short, the signal is worse at 96khz than 44khz or whatever.

    So the difference you hear is because the 96khz is degraded and has aliasing from higher frequencies.

    Sorry to burst your bubble. Just physics and math. No amount of snake oil will change that.

    Leave a comment:

  • vsteel
    Phoronix Member

  • vsteel
    replied
    Originally posted by ssokolow View Post

    If you've spent that much money, I doubt you'll be swayed but, for others, you don't need bit-perfect 192KHz/24bit audio.

    The reason 192/24 sounds better is because they generally put more care into mastering when they sell you 192/24 audio and, surround vs. stereo aside, it'd sound just as good if they sold you a CD-quality (44.1/16) downmix made with the same degree of care from the same master. Aside from being an indicator that they probably put more effort into mastering properly, 192/24 is just intended as a temporary format for mastering and mixing to give the equipment and filters more wiggle room to avoid rounding errors and other kinds of degradation as the audio engineers apply various different stages of processing on the recording.

    Monty from Xiph.org (the guy who came up with Ogg Vorbis) wrote a lengthy and heavily-cited article on the details of this.
    I will respectively disagree with you. As mentioned in my other comment, there are constructive and destructive interference that effect the waves we can hear. I also have done testing on the very thing you have mentioned and I can hear a difference up to 96Khz 24bit and and others that I have tested. I have not tested anyone including myself that could tell a difference between 192Khz and 96Khz though maybe I don't have the equipment for it. I didn't tell them before which levels they were hearing and they could tell. Yes it must be mastered correctly but rather than read internet articles a person needs to actually listen and see if they can tell a difference. A person might not have good enough equipment, listen close enough or just not have the hearing that can tell a difference.

    Leave a comment:

  • vsteel
    Phoronix Member

  • vsteel
    replied
    Originally posted by Brisse View Post

    You don't actually need to bypass PA to get bit perfect. PA is smart enough to choose whatever sample rate the first detected audio stream has, so as long as nothing else is playing when you start playing music, you are getting bit perfect audio. Normally it will only choose between 44.1khz and 48khz, but this can be reconfigured to accept any sample rate if you want to play ultrasonic music for your dog, which seems to be a thing that you're into judging from the rest of your comment.
    I might have to try that again, last time I tried there were issues and it didn't seem to work very well but that was a while ago.

    It is about having enough headroom, I know a person can't hear the highest frequencies that you can reproduce but you can hear the constructive and destructive interference and how that effects the music along with how it effects the speakers. I know everyone loves to quote the only need to sample twice the frequency, the issue is you can't tell the shape of the wave. Another real world example is when you buy an Oscope for work you always sample at least 10x the frequency you are using. On a side note though I can't really hear the difference between 92Khz and 192Khz but I can tell the difference between those and 44.1.

    Leave a comment:

Working...
X