The ability to reproduce a 20 kHz sine wave does not guarantee the ability to reproduce transients from drums, piano, cymbals and hand claps, etc, in a clean, non distorted way with a sharp leading edge chock wave. Even though our ears does not hear a 21 kHz (15 kHz if you are as old as me) sine waves, we can still differentiate between a sharp crack from a hand clap or a hard edge strike on a drum, from a muffled and low passed reproduction of the same source material.
The ability to reproduce clean, fast and sharp transients is much more important to the perceived sound quality then bandwidth, distortion and flat tone curves. Our ears and brain can to a quite remarkable extent compensate for room acoustics effecting the tone curve (peaks and dips in the response) as long as early reflections does not muddy the transient response.
Transients is what differentiates music from tones. No amount of steady state sine waves can synthesize a piano, Cello or drums.
This is why true 24/96 or 24/192 records sounds much better then 16/44,1 if the rest of the reproduction chain is able to handle the resolution and bandwidth.
The ability to reproduce clean, fast and sharp transients is much more important to the perceived sound quality then bandwidth, distortion and flat tone curves. Our ears and brain can to a quite remarkable extent compensate for room acoustics effecting the tone curve (peaks and dips in the response) as long as early reflections does not muddy the transient response.
Transients is what differentiates music from tones. No amount of steady state sine waves can synthesize a piano, Cello or drums.
This is why true 24/96 or 24/192 records sounds much better then 16/44,1 if the rest of the reproduction chain is able to handle the resolution and bandwidth.
Yes, that is true!
I find it much easier to hear the difference between 16/44,1 and 24/96 with point source speakers like Unity/Synergy-horns or full range drivers, then from "normal" multiway hifi loudspeakers with their crossovers and multiple drivers spaced apart on a baffle.
It is also very important to minimize early reflections that interfere with the transient reproduction (edge diffraction, multiple drivers etc etc).
I find it much easier to hear the difference between 16/44,1 and 24/96 with point source speakers like Unity/Synergy-horns or full range drivers, then from "normal" multiway hifi loudspeakers with their crossovers and multiple drivers spaced apart on a baffle.
It is also very important to minimize early reflections that interfere with the transient reproduction (edge diffraction, multiple drivers etc etc).
That's nonsense. What you hear is reproduced. What you can not hear is not reproduced. And that's the way it has to be. Have you ever read the theories of Nyquist in math lessons? Transients above 20kHz are inaudible and below 20kHz they are reproduced.
There are discussions on forums on fast transients, but they are mainly related to loudspeaker and how bad they are at reproducing fast transient.
I'm quite on disagreement with the statement "No amount of steady state sine waves can synthesize a piano, Cello or drums.", Fourier Analysis actually demonstrates that it is indeed true, any signal can be synthesised as a sum of sinusoidal waves with proper amplitude and phase.
Human hearing is indeed more sensitive to transient, but it has mostly to do with the need for a quick localisation of the source. I'm not an expert on the subject but I think the quick localisation is based on the phase rather than the bandwidth of the sound. And again loudspeakers are not very accurate on the signal phase.
I'm quite on disagreement with the statement "No amount of steady state sine waves can synthesize a piano, Cello or drums.", Fourier Analysis actually demonstrates that it is indeed true, any signal can be synthesised as a sum of sinusoidal waves with proper amplitude and phase.
Human hearing is indeed more sensitive to transient, but it has mostly to do with the need for a quick localisation of the source. I'm not an expert on the subject but I think the quick localisation is based on the phase rather than the bandwidth of the sound. And again loudspeakers are not very accurate on the signal phase.
Myth. Such 'shock waves' from handclaps, drum beats etc do not exist in real life. You might think they do based on your hearing. Analysis of the real waveforms shows no such thing.
oh oh
So you're telling that all the people that insist to say to use a microphone to do measurements because FR, phase, diffraction, refraction, cannot be heard as well as DDR, Hom, metal domes ringing, break-up in membranes, impedance -related issues etc. cannot be heard and those people are wrong ?!
The only argument I can think of is relaxing the filter requirements. But the stupid thing is that no vendor uses these high FS recording for that but still chase the last 0,00001 dB at 0,1 below the FS.
Crazy.
//
Fourier Series -- from Wolfram MathWorld
24/96 falls quite short of "infinite sum of sines and cosines" as that would also mean infinite bandwidth....
Everything is possible in the theoretical real of mathematics when you don't care much about the need for infinite bandwidth. But it is a little harder squeezing "infinite bandwidth" out of a 16/44,1 CD record.
YouTube
I did not say "supersonic" chock wave. There is a very easily heard and seen sharp and steep wave leaving the hands. It is sharp enough to startle people of you clap your hands suddenly and behind their back. It is quite hard for normal hifi speakers to reproduce the sound of a hand clap with any realism. Large horns like Danley Synergy horns do it much better then normal hifi loudspeakers.
Looking in DAC and ADC chip datasheets, you often see transition bands ranging all the way from 20 kHz to 156.4 kHz at 176.4 kHz sample rate (so the ADC messes up the 20 kHz to 88.2 kHz range with aliases and the DAC fills up 88.2 kHz to 156.4 kHz with images) and passband ripples much greater than 0.00001 dB.
Not quite, it just means that the frequency spacing between one sine wave and the next tends to zero, in the case of a Fourier transform rather than a Fourier series.
The link you posted includes "Monty's demos" which includes this 24 minute video which references
the article in the link. This video is referenced frequently on HydrogenAudio forums as an excellent
explanation of digital audio. If you haven't yet seen it I highly recommend it.
Xiph.Org Video Presentations: Digital Show & Tell
G²
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.