Listening to DSD256 native audio files, on a proper DAC, on my active without DSP studio monitors, that have -3db response until 50khz, I can say that DSD256 it is the best digital I heard. Can´t wait to hear native DSD512 audio files! Can´t you record some DSD512? You already have the technology!
since medium (air) is always changing by heat expanding, humidity, volume. it would be difficult to get near perfect audio reproduction using any of digital technology since amplifier is controlled by solid state mechanism (digital). maybe perfect sound reproduction can be reached by using analog voice drivers that are not powered and driven by digital, for instance by using very controlled lightning or spark produced by tesla coil.
As i understand it, i guess this 5uS affair is true only for stereo source, since the ability of human hearing is to resolve this time delay between left and right ear. So, in mono this isn't problem. (??)
Not exactly. In stereo it's about stereo imaging how truly can system reproduce the information about space. But still it's about pulse response in the each chennel...think about it like about how fast can electronics react on signal chenges...44khz have different "grid" then 192khz....
what George says in this video, essentially confirms what i was suspecting all along... we can probably only hear and identify consciously up to some 20khz for a prolonged period, but subconsciously our brain still captures the higher pitched frequencies and does process them accordingly, and that gives it a "feeling" of hearing something more than 20khz?
@@slofty First of all, that's not true I'm afraid. Dispersion is dependent on wavelength (and therefore frequency assuming a relatively constant speed of sound), for instance, imagine a speaker, high frequencies will propagate forwards, whereas because low frequencies have a wavelength that's long relative to the size of the speaker, it will spread around the size of the speaker. This is why bass can be loud behind a speaker, but higher frequency sounds get quieter (in an anechoic environment - no echoes, we actually wouldn't hear any high frequencies, assuming the cabinet also wasn't vibrating). As for my initial statement. This is an altogether different matter. This is about the conversion from digital to analogue and therefore resolution. If we had a sample rate of 20kHz then the highest frequency it would be able to represent is 10kHz, because you need at least two points to plot within a cycle of sound in order to reproduce that frequency. However, just using the Nyquist frequency (double the frequency we are attempting to reproduce), doesn't mean that it's being reproduced accurately. A sample rate of 40kHz gives us two points to plot within a 20kHz cycle, but these two points barely represent the shape of the wave at all, draw a line between them and you end up with a straight line, not a sine wave. Plot many points and you get closer to the sine wave. This can be mitigated to a certain extent through interpolation, but this is just smoothing relative to the rest of the signal and not actually putting the detail back in. There's the other matter that these two points might happen to line up on a silent part of the cycle rather than the peaks, meaning that the signal isn't actually converted (there is also everywhere in between). This is why we don't use a minimum sample rate of 40kHz, but slightly higher at 44.1kHz or higher. Sample rates above this will represent the wave even more accurately. Now whether a speaker is able to resolve such high frequency information accurately is a different matter altogether, and whether we'd be able to discern it at all is a totally different matter. Hope this helps
@@Elliott-Designs "but these two points barely represent the shape of the wave at all, draw a line between them and you end up with a straight line, not a sine wave" Watch videos on the Fundamental Theorem of Calculus. If you used calculus during your time at uni, you would understand that d/dx can be sufficiently expressed such that an oscilloscope of a converted 16/44khz (or even lower bit depths!) signal will match an analog source-generated signal of, say, any given fundamental (at a minimum, I am not looking to give examples that would be described by _g' of x_ etc. or other notational forms). This is what AES (or other worldwide equivalent) engineers took into account when developing hardware and other standards to fit these devices' tech parameters well within the audible threshold for any given human being.
@@Elliott-Designs Can you give any iterative layout of how the sigma notation used in the development of the IC layout in 16/44khz systems to fully approximate an analog waveform along slope f(x) such that dy/dx _CANNOT_ provide a sufficient value of the tangent of that particular slope at any _h?_ At present moment mathematics says otherwise. Because in doing so you would show that Lagrange and everyone after him was mistaken. Keep in mind Fourier, Poisson, and Babbage cited Lagrange in their work and those that followed cited their work in the development of nuclear/particle physics, semiconductor tech, and GPS, among many other things!
Listening to DSD256 native audio files, on a proper DAC, on my active without DSP studio monitors, that have -3db response until 50khz, I can say that DSD256 it is the best digital I heard. Can´t wait to hear native DSD512 audio files!
Can´t you record some DSD512? You already have the technology!
since medium (air) is always changing by heat expanding, humidity, volume. it would be difficult to get near perfect audio reproduction using any of digital technology since amplifier is controlled by solid state mechanism (digital). maybe perfect sound reproduction can be reached by using analog voice drivers that are not powered and driven by digital, for instance by using very controlled lightning or spark produced by tesla coil.
As i understand it, i guess this 5uS affair is true only for stereo source, since the ability of human hearing is to resolve this time delay between left and right ear. So, in mono this isn't problem. (??)
Not exactly. In stereo it's about stereo imaging how truly can system reproduce the information about space. But still it's about pulse response in the each chennel...think about it like about how fast can electronics react on signal chenges...44khz have different "grid" then 192khz....
Brilliant
what George says in this video, essentially confirms what i was suspecting all along... we can probably only hear and identify consciously up to some 20khz for a prolonged period, but subconsciously our brain still captures the higher pitched frequencies and does process them accordingly, and that gives it a "feeling" of hearing something more than 20khz?
He said absolutely nothing to that effect.
??
I don't believe that this old man still have 20 Khz hearing. No one can keep 20 kHz hearing at this white hire age.
He's not talking about that. He's talking about the time domain. A 1MHz sampling rate can more accurately represent the time domain
@@Elliott-Designs How? Sound propagates through air all the same.
@@slofty First of all, that's not true I'm afraid. Dispersion is dependent on wavelength (and therefore frequency assuming a relatively constant speed of sound), for instance, imagine a speaker, high frequencies will propagate forwards, whereas because low frequencies have a wavelength that's long relative to the size of the speaker, it will spread around the size of the speaker. This is why bass can be loud behind a speaker, but higher frequency sounds get quieter (in an anechoic environment - no echoes, we actually wouldn't hear any high frequencies, assuming the cabinet also wasn't vibrating).
As for my initial statement. This is an altogether different matter. This is about the conversion from digital to analogue and therefore resolution. If we had a sample rate of 20kHz then the highest frequency it would be able to represent is 10kHz, because you need at least two points to plot within a cycle of sound in order to reproduce that frequency. However, just using the Nyquist frequency (double the frequency we are attempting to reproduce), doesn't mean that it's being reproduced accurately. A sample rate of 40kHz gives us two points to plot within a 20kHz cycle, but these two points barely represent the shape of the wave at all, draw a line between them and you end up with a straight line, not a sine wave. Plot many points and you get closer to the sine wave. This can be mitigated to a certain extent through interpolation, but this is just smoothing relative to the rest of the signal and not actually putting the detail back in. There's the other matter that these two points might happen to line up on a silent part of the cycle rather than the peaks, meaning that the signal isn't actually converted (there is also everywhere in between). This is why we don't use a minimum sample rate of 40kHz, but slightly higher at 44.1kHz or higher. Sample rates above this will represent the wave even more accurately.
Now whether a speaker is able to resolve such high frequency information accurately is a different matter altogether, and whether we'd be able to discern it at all is a totally different matter.
Hope this helps
@@Elliott-Designs "but these two points barely represent the shape of the wave at all, draw a line between them and you end up with a straight line, not a sine wave"
Watch videos on the Fundamental Theorem of Calculus. If you used calculus during your time at uni, you would understand that d/dx can be sufficiently expressed such that an oscilloscope of a converted 16/44khz (or even lower bit depths!) signal will match an analog source-generated signal of, say, any given fundamental (at a minimum, I am not looking to give examples that would be described by _g' of x_ etc. or other notational forms). This is what AES (or other worldwide equivalent) engineers took into account when developing hardware and other standards to fit these devices' tech parameters well within the audible threshold for any given human being.
@@Elliott-Designs Can you give any iterative layout of how the sigma notation used in the development of the IC layout in 16/44khz systems to fully approximate an analog waveform along slope f(x) such that dy/dx _CANNOT_ provide a sufficient value of the tangent of that particular slope at any _h?_ At present moment mathematics says otherwise.
Because in doing so you would show that Lagrange and everyone after him was mistaken. Keep in mind Fourier, Poisson, and Babbage cited Lagrange in their work and those that followed cited their work in the development of nuclear/particle physics, semiconductor tech, and GPS, among many other things!