Tsuyoshi Yamamoto Trio – Sweet For - Is vinyl record losing its quality? Comparison between vinyl record, SACD and streaming

[Jean.Francois]

Thank you for this work and these additional measurements.
We can indeed see that DSD64 to PCM conversion involves filtering in the high frequencies, a filtering that is more important with JRiver, which cuts above 20 kHz, unlike DBPoweramp, which does not filter (or only slightly) below 48 kHz, as can be seen from the start of the noise rise above 35 kHz.
This is the limit of DSD64, with a very strong rise in noise, which is generally reduced by filtering in the high frequencies like JRiver. I don't know if the filter frequency is adjustable in JRiver.
This is why we now have DSD128 or DSD256 formats, which push noise higher up the frequency range.
At the lower end of the spectrum, the signal is identical, it's just that, as you point out, the FFT size is the same for 96 kHz and 352.8 kHz, so you lose precision in the low frequencies with the higher sampling frequencies.

 

[dualazmak]

Thank you for this work and these additional measurements.
We can indeed see that DSD64 to PCM conversion involves filtering in the high frequencies, a filtering that is more important with JRiver, which cuts above 20 kHz, unlike DBPoweramp, which does not filter (or only slightly) below 48 kHz, as can be seen from the start of the noise rise above 35 kHz.
This is the limit of DSD64, with a very strong rise in noise, which is generally reduced by filtering in the high frequencies like JRiver. I don't know if the filter frequency is adjustable in JRiver.
This is why we now have DSD128 or DSD256 formats, which push noise higher up the frequency range.
At the lower end of the spectrum, the signal is identical, it's just that, as you point out, the FFT size is the same for 96 kHz and 352.8 kHz, so you lose precision in the low frequencies with the higher sampling frequencies.

Yes, thank you for your nice summary; I essentially/generally agree with you.

My post here would be also of your interest and reference regarding objective comparison between;
1. DSD256(4x), DSD128(2x), DSD64(1x),
2. DXD 352.8 32-biut WAV, DXD 352.8 kHz 24-bit WAV,
3. FLAC 352.8 kHz 24-bit, FLAC 192 kHz 24-bit, FALC176.4 kHz 24-bit, FALC 96 kHz 24-bit FLAC 88.2 kHz 24-bit,
4. PCM 192 kHz 24-bit WAV, PCM 176.4 kHz 24-bit,WAV, PCM 96 kHz 24-bit WAV, PCM 88.2 kHz 24-bit WAV, PCM 44.1 kHz 24-bit WAV

of carefully prepared and well QC-ed one identical digitally recorded sampler track kindly provided by SOUND LIAISON company.


Edit:
Let me add a caution for all of you.

In many of poorly QC-ed (or actually no carful QC practice at all?) SACD=DSD64(1x) and/or download-purchased DSD64(1x) DSD128(2x) SDS256(4x) and even DSD512(8x) music tracks, sometimes (or I may say "often") unacceptable (maybe harmful to our audio gears and/or beloved pets I mean dogs, cats, birds, etc.) huge amount of quantization UHF (ultra-high frequency) noises are included even though they are inaudible when and if transparently reproduced by our audio system with excellent tweeters and/or super-tweeters. You can find here a typical example case of DSD128(2x) tracks containing such bad UHF noises which I download purchased.

In the same post, you can also find an example case of well QC-ed DSD256(4x) track I purchased in which the UHF quantization noises are well suppressed.

This issue of "UHF quantization noise in DSD format", therefore, would be also one of the important QC matters, I believe.


In any way, I essentially do not care about it since now I have -48 dB/Oct low-pass (high-cut) filters at 25 kHz in my DSP configuration, as well as I always convert all of my digital music tracks (including DSD tracks) into 88.2 kHz or 96 kHz 24-bit PCM (i.e. up to 44.1 kHz or 48 kHz in L&R channel) "on-the-fly" by JRiver MC to feed into system-wide DSP center "EKIO", as repeatedly shared here. (My latest total setup can be found here.)

At least in my case, therefore, I will/shall never purchase music disks/tracks (except for rare experimental purposes liken in this thread) exceeding 96 kHz 24-bit, since now I have -48 dB/Oct low-pass (high-cut) filters at 25 kHz in my DSP configuration.

 

[Jean.Francois]

Yes, thank you for your nice summary; I essentially/generally agree with you.

My post here would be also of your interest and reference regarding objective comparison between;
1. DSD256(4x), DSD128(2x), DSD64(1x),
2. DXD 352.8 32-biut WAV, DXD 352.8 kHz 24-bit WAV,
3. FLAC 352.8 kHz 24-bit, FLAC 192 kHz 24-bit, FALC176.4 kHz 24-bit, FALC 96 kHz 24-bit FLAC 88.2 kHz 24-bit,
4. PCM 192 kHz 24-bit WAV, PCM 176.4 kHz 24-bit,WAV, PCM 96 kHz 24-bit WAV, PCM 88.2 kHz 24-bit WAV, PCM 44.1 kHz 24-bit WAV

of carefully prepared and well QC-ed one identical digitally recorded sampler track kindly provided by SOUND LIAISON company.


Edit:
Let me add a caution for all of you.

In many of poorly QC-ed (or actually no carful QC practice at all?) SACD=DSD64(1x) and/or download-purchased DSD64(1x) DSD128(2x) SDS256(4x) and even DSD512(8x) music tracks, sometimes (or I may say "often") unacceptable (maybe harmful to our audio gears and/or beloved pets I mean dogs, cats, birds, etc.) huge amount of quantization UHF (ultra-high frequency) noises are included even though they are inaudible when and if transparently reproduced by our audio system with excellent tweeters and/or super-tweeters. You can find here a typical example case of DSD128(2x) tracks containing such bad UHF noises which I download purchased.

In the same post, you can also find an example case of well QC-ed DSD256(4x) track I purchased in which the UHF quantization noises are well suppressed.

This issue of "UHF quantization noise in DSD format", therefore, would be also one of the important QC matters, I believe.


In any way, I essentially do not care about it since now I have -48 dB/Oct low-pass (high-cut) filters at 25 kHz in my DSP configuration, as well as I always convert all of my digital music tracks (including DSD tracks) into 88.2 kHz or 96 kHz 24-bit PCM (i.e. up to 44.1 kHz or 48 kHz in L&R channel) "on-the-fly" by JRiver MC to feed into system-wide DSP center "EKIO", as repeatedly shared here. (My latest total setup can be found here.)

At least in my case, therefore, I will/shall never purchase music disks/tracks (except for rare experimental purposes liken in this thread) exceeding 96 kHz 24-bit, since now I have -48 dB/Oct low-pass (high-cut) filters at 25 kHz in my DSP configuration.

Hello,

indeed, DSD has a noise rise in the high frequencies, with DSD64 (used for SACD), the noise starts above 25/30 kHz. This frequency increases if you switch to DSD128 or DSD256. Hence the interest in using DSD256 rather than DSD64.
But, indeed, there are DSD256 that are similar to DSD64s. So we might wonder whether it's not a simple DSD64 that's been converted to DSD256?
As always, it depends on the professionalism of those who offer these high-resolution versions.
I've already encountered this problem several times, but without any explanation from the vendor.

 

[dualazmak]

Hello,

indeed, DSD has a noise rise in the high frequencies, with DSD64 (used for SACD), the noise starts above 25/30 kHz. This frequency increases if you switch to DSD128 or DSD256. Hence the interest in using DSD256 rather than DSD64.
But, indeed, there are DSD256 that are similar to DSD64s. So we might wonder whether it's not a simple DSD64 that's been converted to DSD256?
As always, it depends on the professionalism of those who offer these high-resolution versions.
I've already encountered this problem several times, but without any explanation from the vendor.

At least in my case, after rather intensive comparative measurements and subjective comparative listening sessions, my present (and future?) stance is very simple as I wrote in my above post #22; "I will/shall never purchase music disks/tracks (except for rare experimental purposes liken in this thread) exceeding 96 kHz 24-bit, since now I have -48 dB/Oct low-pass (high-cut) filters at 25 kHz in my DSP configuration." (ref. here #532 on my project thread.)