soongsc said:
What makes that curve to have such ends will come from frequency components above 20Khz. But humans have a built in 20Khz (some has lower) low pass filter, their ears. The choice made to use 44.1Khz sampling is related to how high frequencies humans can hear. So if you had taken that wave form which such abrupt change from zero level to a sinusoid of one period and then again to zero and filtered it with a 20Khz a theoratical brickwall filter and listened to it, it would sound as same as if yo had played it without filtering it. Because the human ear doesn't hear anything above 20Khz, so whether the signal has components above 20Hz or not doesn't make any difference. If human's could hear up to 30Khz, then a sampling period of at least 60Khz would be required not to loose any information "that is audiable". The idea is to keep and reconstruct information that is audiable.
BTW Werner's post earlier on this thread points to real problems with a sampling rate of 44.1Khz, but it is not what you are pointing out. I would suggest reading a book on linear systems and signals to get a better understanding on how these are working. There is an online book at www.dspguide.com, which is very good at explaining things without using not too technical and math language.
Another side note 16bit provides 90db dynamic range, which is not something to sneeze at. Do you ever get a listening room that has less noise than 90db for instance?
janneman said:
Any single musical instrument cannot create that, I agree. But if you take two instruments ans shift the time around somewhat, that combination can occur.
gmarsh said:
As afore mentioned, two or multiple instruments can end up with some segment of music similar to the red curve. And this kind of signal cannot be preserved.
I recall that the 20K Hz human limit is based on steady wave testing. I also recall somewhere mentioning that humans can detect transients equivalent to 50KHz.
Euphase said:
Good link. I qoute:
"In addition to EFM, the data are encoded in a format called two-level Reed-Solomon coding. This involves combining the left and right stereo channels along with data for error detection and correction. Digital errors detected during playback are either: corrected by using the redundant data in the encoding scheme, concealed by interpolating between adjacent samples, or
muted by setting the sample value to zero. These encoding schemes result in the data rate being tripled, i.e., 1.4 Mbits/sec for the stereo audio signals versus 4.3 Mbits/sec stored on the disc.
After decoding and error correction, the audio signals are represented as 16 bit samples at a 44.1 kHz sampling rate. In the simplest system, these signals could be run through a 16 bit DAC, followed by a low-pass analog filter.
However, this would require high performance analog electronics to pass frequencies below 20 kHz, while rejecting all frequencies above 22.05 kHz, ½ of the sampling rate. A more common method is to use a multirate technique, that is, convert the digital data to a higher sampling rate before the DAC.
A factor of four is commonly used, converting from 44.1 kHz to 176.4 kHz. This is called interpolation, and can be explained as a two step process (although it may not actually be carried out this way). First, three samples with a value of zero are placed between the original samples, producing the higher sampling
rate. In the frequency domain, this has the effect of duplicating the 0 to 22.05 kHz spectrum three times, at 22.05 to 44.1 kHz, 41 to 66.15 kHz, and 66.15 to 88.2 kHz. In the second step, an efficient digital filter is used to remove the newly added frequencies."
Jan Didden
Euphase said:
I think if the theory meets reality, then a difference test sending the same signal through two different sample and playback systems would result in zero error, in reality I have yet to see two systems that do that.
The world was flat in theory at some time in the past.
🙂
soongsc said:
Why should it be the same? It will not be the same, after all, even if the digital part is identical, there is still the difference in filtering. Especially if there is a difference in sampling rate, there is a difference in filter types and characteristics.
An other question is, will they sound the same? That may well be, even with the different technical characteristics. One example: wildly different phase characteristics are virtually undetectable by the human ear.
Jan Didden
Two instruments together cannot create extra frequencies that a single instrument cannot create. The two base signals (instantaneous-starting sine wave) that summed to create the red curve cannot exist in reality. Thus the red curve can't exist.soongsc said:
Regarding the 50KHz thing, maybe. Perhaps this is why DVD Audio sounds better to me. 😀
janneman said:
You will never get the same digital part with the same music signal. 😀
On good systems, I can hear the difference between non-inverted signals and inverted signals if the instruments played are of the percussion type.
gmarsh said:
I guess you are right that this cannot be naturally recorded due to the acoustic waves.🙂
I hadn't really sat down with a good DVD audio system, but in THEORY it should be better? Or maybe if the 44.1 KHz is already perfect, there is no room for improvement.😀
soongsc said:
I think the main problem with 44.1KHz isn't the loss of high frequency content, it's the actual packing-to-44.1-and-unpacking-it process; The (possibly audible) effect of antialiasing/antiimaging filters, the compression and shaped dither used to get the 'best' dynamic range out of 16 bits, and so forth.
One thing that I find odd is the amount of 'clipping' you find on most compact discs. You'll often find several +32767/-32768 samples per second on a 'loud' CD track. Makes me wonder what sort of musical detail may have been lost in compression and clipping.
A lot of these demands are relaxed with the higher sampling rate and 24-bit dynamic range of DVD Audio.
Another question worth asking oneself is, "Has anyone demonstrated the audibility of inserting a functioning 16/44.1 A/D->D/A into a signal path in any reasonably controlled test?"
Well, there have been experiments that show that the difference between 16 and 24bit is audible, and I think I also remember seeing some showing the same for the difference between even 44.1 and 48kHz, so it seems reasonable to assume that converting to/from 16/44.1 will be audible compared to the original analogue.SY said:
Gee guys, lots of noise here.
Werner posted already on nr. 27 the real reason: its the sampling proces!
What he said is: if you sample at 44,1 kHz, you can at max reconstruct a SINE-wave of 22,05 kHz; the circuitry cannot "see" if it is anything else than a sinewave.
If you really want to reconstruct the waveform from a sampled signal you want at least 10 samples (measured defining points), which means that a normal CD produces an "undistorted" signal till about 4500 Hz. 😀
Some people think you need 20 samples, so then the situation is even worse ......
However there are also people that think that in practice 5 samples will do, as out brains will do the rest ...... so let's say we have 9 kHz from a CD "undistorted".
That why "the vinyl boys" keep burning the CD!
The only way out will be a very high sampling rate ic. SACD. 😉
The rest of the discussion has more to do with s/n ratio's and the audible effects of rambling in the digital signal, which may eventually also be heard as distortion.
Werner posted already on nr. 27 the real reason: its the sampling proces!
What he said is: if you sample at 44,1 kHz, you can at max reconstruct a SINE-wave of 22,05 kHz; the circuitry cannot "see" if it is anything else than a sinewave.
If you really want to reconstruct the waveform from a sampled signal you want at least 10 samples (measured defining points), which means that a normal CD produces an "undistorted" signal till about 4500 Hz. 😀
Some people think you need 20 samples, so then the situation is even worse ......
However there are also people that think that in practice 5 samples will do, as out brains will do the rest ...... so let's say we have 9 kHz from a CD "undistorted".
That why "the vinyl boys" keep burning the CD!
The only way out will be a very high sampling rate ic. SACD. 😉
The rest of the discussion has more to do with s/n ratio's and the audible effects of rambling in the digital signal, which may eventually also be heard as distortion.
What is the effective band width of a vinyl disk?
I seem to remember that the RIAA curve has very little output (-20dB) after 20 kHz
It is interesting that that some people can hear a difference between a recording on vinyl and CD, when the master tape for both is digital 48 kHz 16 bit….. I find it a bit confusing.
\Jens
I seem to remember that the RIAA curve has very little output (-20dB) after 20 kHz
It is interesting that that some people can hear a difference between a recording on vinyl and CD, when the master tape for both is digital 48 kHz 16 bit….. I find it a bit confusing.
\Jens
JensRasmussen said:
You're quite right.
For some 25 years the cutting machines were able to put up to 45 kHz in the groove ...... That top-range was not only for audio but also used for some FM-signal to derive 4 channel reproduction.
Anyhow the problem is not getting it in but to get it out. 🙄
Very good vinyl records were made directly or thru a wide-band low-noise tape recorder.
Indeed: at the end of the seventies a lot of recordings were made on 44.1 or 48 kHz. Lots of these are not worth the trouble to put them on SACD (or another wide-band medium).
So you have to be very selective when buying SACD!!
It IS confusing because some CD's sound very good to the ear ..... I think that has to do with the fact that in many (non classical) recordings have artificially generated overtones already in the source.
marconist said:
I really don't understand what you are trying to say. In MATH world, where you can have ideal filters etc., one can sample any shape of signal that is band limited to 20Khz, with a sampling frequency of 40Khz, and then reconstruct it EXACTLY as it was. May be I am reading what you are saying wrong, but from what you wrote it seems to me that you don't agree with this (or you believe that humans are easily able to hear beyond 20Khz). But it is just the way it is, even if you don't buy it or not. May be you should reread to that Werner's post. I don't think your arguments hold any water, sorry no offense but have to say what is right. Moving from MATH world to real world with its practical limitations, there are some problems that arise, but nothing like you said. There is a reason why the CD format was chosen to use 44.1Khz, not just 40Khz, and it is to be able to avoid some of those practical problems to implement the math world solutions. A higher sampling rate will make things easier, but let's not mix apples with oranges.
You can only reconstruct the part of the signal that is below half the sampling freq
If you sample a perfect 20 khz square wave using a 40 khz sampling freq, you will get a nice 20 khz sine wave (the fundemental freq) after the A/D converter and filter.
\Jens
If you sample a perfect 20 khz square wave using a 40 khz sampling freq, you will get a nice 20 khz sine wave (the fundemental freq) after the A/D converter and filter.
\Jens
hallo guys
it's been a good read..
one more thing to discuss here I think. The higher sampling rate makes an ADC less prone to analog anti-aliasing filter imperfection. Not only wide-band noise of filter aliases, but also harminic distortion and possibly products of HF instability. Don't forget they are made of op-amps and quite complex in terms of phase response within high-feedback loop!!!
It could be useful to use high sampling rate for ADC.
for DAC it is the same thing again-the 'stairway' signal from adc has wide bandwidth and HF components can easily intermodulate in interpolation filter. Actually we don't know how signals intermodulate in op-amp based filter. Op-amps are measured for IM with non-reactive impedances within feedback.
What do you think?
best regards
it's been a good read..
one more thing to discuss here I think. The higher sampling rate makes an ADC less prone to analog anti-aliasing filter imperfection. Not only wide-band noise of filter aliases, but also harminic distortion and possibly products of HF instability. Don't forget they are made of op-amps and quite complex in terms of phase response within high-feedback loop!!!
It could be useful to use high sampling rate for ADC.
for DAC it is the same thing again-the 'stairway' signal from adc has wide bandwidth and HF components can easily intermodulate in interpolation filter. Actually we don't know how signals intermodulate in op-amp based filter. Op-amps are measured for IM with non-reactive impedances within feedback.
What do you think?
best regards
JensRasmussen said:
That's no different than what I wrote, which was:
"In MATH world, where you can have ideal filters etc., one can sample any shape of signal that is band limited to 20Khz, with a sampling frequency of 40Khz, and then reconstruct it EXACTLY as it was."
I don't know what is being discussed here, this is just known for years good old math theory.....
I looked into this part of the DiyAudio the first time, after seeing the title of the thread on the home page, which seemed interesting, hoping to see some interesting stuff. But I guess I was mistaken. May be I should better leave it here...🙂
- Status
- Not open for further replies.