Hi-res music is not about over sampling.
It is an original performance that was recorded with a higher sampling rate.
2 Likes
The benefit of upsampling is to avoid the use of a complexed low pass filter at the output of D/A.
A complexed filter introduces phase errors that can impact the quality of music.
1 Like
Remember that the 44.1khz is based on information theory developed by Nyquist and Shannon.
You need to sample to at least the double of the highest frequency you want to record.
So at 44kHz, you will find the 19kHz component.
The problem is that there can be an amplitude or a phase error since at this frequency, the waveform is sampled twice.
1 Like
Looks like you don’t know what is the difference between over-sampling and up-sampling
Do you think they are the same? If you think they are the same, I suggest you to check more about their differences.
Yes, agreed. It is ONE of the benefits of upsampling.
Yes, possible but not a must.
Agreed
I am confused. Do you mean that for a 1kHz component 44kHz would sample it 44 times?
Yes
At 44.1 kHz, sampling is every 22,7 microseconds.
1kHz component lasts 1 millisecond or 1000 microseconds.
1000/22,7 = 44
For hi res music, I don’t talk about over or up sampling.
It is just the fact that in the recording studio, they use a higher sampling rate.
James, for the purposes of what we are talking about, you can consider them the same, there is no need to be getting pedantic on it:
1 Like
As @jussi_laako mentioned that up- and over- sampling is the same from the DAC’s point of view.
For this, I 100% agree. i.e. it means the DAC cannot tell if a 192/24 digital input is from Hi-Res (192/24, i.e. over-sampling as 44kHz is good enough to capture 22kHz audio signal) or from a up-sampling source (i.e. up-sampling from a 44kHz source to 192/24 by HQPlayer)
One thing we are pretty sure is that the two “192/24” are different to our ears as the over-sampled 192/24 “contain more data from the original analog audio signal”
i.e.
The up-sampled 192/24 source was interpolated from the 44kHz input source, i.e. 44k sampled points per second.
The over-sampled 192/24 source would have more sampled points from the original audio signal, i.e. 192k sampled points per second.
If we look at the original question I asked:
We were talking about Hi-Res music (e.g. 192/24). In this context, Hi-Res music is about over-sampling. Hi-Res music is NOT about up-sampling.
Therefore, sorry, I cannot agree with you that over- and up- sampling are the same, especially with what we were talking above.
The pedantics is getting too side tracked from bigger issues:
There is a tonne of fake hi-res music out there. It is there on Qobuz, Apple Music Hires, Tidal HiRes.
Using crap upsampling/oversampling/resampling filters.
New fakes coming out each and every week.
With HQPlayer Client app and v5 you can see in real-time. No need to check in Audacity like I used to do.
My feature request is what brought this to HQPlayer !
This has long been known from the days of HDTracks, before hi-res streaming went mainstream
1 Like
Let’s focus on the ADC process (not DAC) here:
If your focus is to capture audio signal (i.e. 22k Hz or less), using any sampling rate higher than 44k Hz is known as over-sampling. Agree?
Based on your early reply (as stated below), I assume you agree. (Please let me know if you are not):
Based on the same sampling theory you stated above, using higher bitrate sampling (e.g. 192k Hz) would capture signal up to (96k Hz). Correct?
With our understanding, most people cannot hear anything more than 22kHz. As a result, a lot of people in ASR claimed Hi-Res (over-sampling) is useless. It is a pseudo science claim, I debunked in my blog ( 768kHz for digital music sampling? Can you hear 384kHz? You’re crazy… no-pseudo-science.blogspot.com)
Of course, if your reason for creating 192kHz sampled data file is to capture analog signal with frequency up to 96kHz input signal (i.e. most people would consider such input as not “audio” analog input signal as it is higher than 22k) then it would be a different story. In that case, it is not over-sampling as you need 192k to capture that.
Agree. Over-sampling vs Up-sampling is a bit of off topic here.
Agree, but I still don’t want to walk away with incorrect info. That’s my bottom line.
I think you want to emphaized there are a lot of fake Hi-Res music (i.e. those up-sampling from CD or MP3 sources). For this, I 100% agree with you. Those Hi-Res (to me) is rubbish. These fake Hi-Res blocks the accpetance of real Hi-Res and slow down the whole Hi-Res industry (same as the pseudo science claims from ASR). They are basically trying to kill Hi-Res.
Getting much better recently though. At least I have not encountered much fake recently.
And then places like nativedsd.com state the original recording format separately.
There’s plenty of good hires too. And more coming all the time.
2 Likes
I’m more talking mainstream than you will find at nativeDSD
I encounter fake albums each week
I do like see more 48kHz albums coming out though.
RBCD can sound amazing with great mix/master but the more 24/48kHz (with great mix/master!) the better !
1 Like
Agree, many “mainstream Hi-Res” are questionable.
That’s why I prefer to upsampling CD myself with HQPlayer. It is much much better. 
Thank you @jussi_laako for such great product 
BTW, I have a question regarding dsdnative.com <== it was a typo. Should be nativedsd.com:
I knew that they use HQPlayer for upsampling. If we use the same input source and upsampling ourselves (assuming we are using the same filter, shaper, etc), would we be able to get the same result as they did?
Do they have any special trick to make their up-sampled music sound better than I do at home with HQPlayer?
I agree that people cannot hear over 22kHz.
My point is that the sampling rate has an impact on the accuracy of the digital information.
For 44.1 kHz sampling the number of samples per cycle are:
19 kHz 2 samples.
15 kHz 3 samples
10 kHz 4.4 samples
5 kHz 8,8 samples
1 kHz 44,1 samples
For 192 kHz sampling the number of samples per cycle are:
19 kHz 10 samples
15 kHz 13 samples
10 kHz 19 samples
5 kHz 38 samples
1 kHz 192 samples
Not really. The “impact on the accuracy of the digital information” is little or nothing, both in frequency and amplitude.
As long as the reconstruction filter is sufficient, a 19 kHz sine wave is reproduced accurately with roughly two samples per cycle. The same holds for 20 kHz, 21 kHz, even 22 kHz sine waves.
AJ
2 Likes
As shown in this graphic, if the waveform is sampled at its peak, the phase difference will be 90 degrees.
A few points of note…
Sampling peak to trough (or positive peak to negative peak) is not necessary at any frequency.
As depicted, a sine wave sampled at exactly and only peak and trough in a 44.1 kHz system would be 22.05 kHz, which is the Nyquist frequency and not very interesting because it is outside of the 20 kHz audio band.
Most modern DACs do not operate on sample and hold principles as depicted in the graphic. A better and more accurate graphical representation of a linear phase digitally filtered DAC output would be to show the summation of positive and negative sinc function approximations that peak at the sample points.
Last but not least, just look at some measurements. Exhibiting no phase shift across the passband is standard procedure for linear phase oversampling DACs.
AJ
Honestly… no. Your blog has its target audience, which wants to believe that not only their ears are golden, but they also know more about anything than some stupid scientist possibly could. Next thing, you’d suggest arguing on HeadFi that fuses do not affect the sound
Eh… since when? ADC is its own process, it has nothing to do with up- or over-sampling.
Saying “those scientists, they don’t know nothing!” is not quite the same as debunking.
It does not, to Nyquist frequency, that’s the whole point.
2 Likes