OK, nice to hear so. Essentially I agree with you.
As shared in my above post #38, I will soon share my rather intensive investigations on the specific SACD, hopefully within a few weeks.
Last edited:
-
WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are many reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!
Does DSD recording benefit Japanese traditional instruments?
- Thread starter GXAlan
- Start date
OK, nice to hear so. Essentially I agree with you.
As shared in my above post #38, I will soon share my rather intensive investigations on the specific SACD, hopefully within a few weeks.
Hello @GXAlan and friends,
As wrote in my trailer post #38, I have almost completed my rather intensive but naive investigations on the specific SACD; even though my approaches here are not as sophisticated as @GXAlan’s thorough methods, I believe my naive objective and subjective methods are reliable accurate reproducible enough and they have high compatibility and affinity with my daily audio listening setup and environments.
First, the arrived hybrid SACD’s DSD-layer was extracted into 1 bit 2.8 MHz DSD64(1x) DSF files using the very early model of Sony PlayStation and “the” unofficial SACD ripper software. I dare not go into the details of this DSD-layer extraction method as you may kindly understand.
Then, I also ripped the CD-layer of the hybrid SACD into AIFF PCM of 16 bit 44.1 kHz as usual using dBpoweramp CD-Ripper ver.17.3 [64-bit] .
The extracted DSF tracks and AIFF tracks were taken into my music library of JRiver Media Center 30 (hereinafter abbreviated just as “JRiver”);
Since @GXAlan mainly focused on objective and subjective evaluations on the track-11 of this hybrid SACD, I did the same as follows.
I first analyzed the DSF and AIFF of track-11 by MusicScope 2.1.0;
As quite well thought/expected, as you can see easily, the DSF contains considerable amount of, but low-gain, ultrasonic noises in 34 kHz to through the upper limit of 88.2 kHz; quite common in SACD and downloadable DSD music tracks.
Just let’s take some comparative attentions on rather high frequency zone, as follows;
The music signals at 10 kHz is about -50 dB in DSF and -55 dB in AIFF; low gain and no significant difference in shape and gain.
The signals at 22.05 kHz which is upper limit in CD format is about -70 dB in DSF and -90 dB in AIFF; again low gain and no significant difference in shape and gain.
It would be interesting observing that “meaningful” music signal do exists up to 34 kHz in DSF, even though we usually cannot hear the sound beyond 22 kHz.
The typical “mountain” of ultrasonic noise in DSF reaches to -73 dB at 58 kHz; the entire noise mountain is still very low and of course inaudible even if we feed it into highly efficient super tweeters like my FOSTEX metal horn super-tweeter T925A in case if having no low-pass (high-cut) filter in the audio chain.
Let’s go back to the graphical and numeric reports given by MusicScope. I was/am very much impressed observing that the “audible music portions” of DSF and AIFF are quite and almost identical with each other in shapes, also in gain and other numeric representations. I believe that this hybrid SACD was produced with very much careful engineering and quality control on “consistency” of sound quality in DSD-layer and CD-layer.
At least judging from these reports given by MusicScope, I could assume there would be almost no audible difference between extracted DSF and ripped AIFF, if we would listen to them in exactly the same gain/volume using strictly the same audio setup.
Audio System Setup for Comparative Listening
Setup-1:
Native DSD-DSF feed vs. PCM AIFF feed into Single Stereo DAC to go into Single Amplifier and Passive LCR-Network multi-driver SP System
I believe my first comparative listening should be performed in 1xDSD native feed of the DSF track by JRiver into one stereo HiFi DAC capable of DSD decoding into analog line-level, then the line-level analog signal goes into one amplifier to drive SP system through passive LCR-network; L&R sub-woofer should be used as usual in my setup.
For comparison, of course 16 bit 44.1 kHz PCM feed into the same DAC by JRiver using exactly the same unchanged audio setup.
For this Setup-1, I can/could utilize my “single amplifier reference audio setup” which I have been using all the way through my multichannel audio project. In my current investigation in this post, I use OKTO DAC8PRO in 2-CH USB stereo DAC mode (ref. here), and ACCUPHASE E-460 integrated amplifier driving my 5-way 10-channel stereo multi-driver SP system including L&R sub-woofer as shown in this diagram.
Before going into actual listening comparison, I should objectively quantitatively measure the actual gain/volume difference between DSD feed and PCM feed in analog line-out signal given by the DAC DAC8PRO.
I did it rather naive but reliable and reproducible way as shown in the above diagram; I connected the amplifier E-460’s pre-out signal, which is “pass-through” line-out signal from DAC8PRO, into an audio interface TASCAM US-1x2HR (fully validated in my system, please refer to here and here) feeding the ADC-ed digital signal into my second PC running Adobe Audition 3.0.1 for recording. The recorded tracks were saved as 16 bit 44.1 kHz PCM AIFF track for gain/volume analysis by using MusicScope 2.1.0. Of course, the ADC and recording sequence were done with enough gain/level margin below the clipping level.
Although I already well knew that JRiver does not give 6 dB boost in DSD native digital output, I should objectively and quantitatively confirm it as shown in this result diagram.
As shown in above diagram, I could objectively confirm that JRiver’s DSD native digital feed is exactly 6 dB lower than the AIFF feed even in the case that MusicScope tells/reports both of the DSF and AIFF track have the same gain/loudness level; this is really important message of our interest here in this thread.
Consequently, in my present comparative listening with Setup-1, I need to boost the level/volume in 6 dB for DSD native feed, (or to suppress in 6 dB for PCM-AIFF feed), for perfectly level-matched listening comparison. I can/could do it very easily by JRiver’s internal digital volume controller with 0.1 dB granularity/accuracy.
Setup-2:
On-the-fly conversion of DSF and AIFF tracks into 24 bit 88.2 kHz (or 96 kHz) PCM to digitally feed into software DSP (XO/EQ/time-alignment) to give DSP-ed multi-channel digital signals into multi-channel DAC; my multichannel multi-SP-driver multi-amplifier 5-way 10-channel fully active (LCR-network and attenuators are fully eliminated) stereo audio system
This is my fully established multichannel stereo audio setup. Although DSD-DSF tracks are not natively fed into software DSP, I already proved the merit (total sound quality) of this setup well surpasses the Setup-1 configuration (i.e. native DSD feed into single DAC single amp), thanks to the direct dedicated drive of each of the SP drivers (full elimination of LCR-network and attenuator) by dedicated suitable amplifier (right person in right place) as wells as due to the establishment of 0.1 msec precision time alignment among all the SP drivers.
My present system diagram is like this;
Again, even in this Setup-2, before going into actual listening comparison, I should objectively quantitatively measure the actual gain/volume difference between DSD-DSF and PCM-AIFF, both on-the-fly converted into 88.2 kHz or 96 kHz PCM, in analog line-out signal given by the DAC DAC8PRO. How can/could I do it?
Very fortunately and nicely, I already installed my IEC 60268-17 compatible large glass-face DIY 12-VU-Meter Array in the setup (ref. here) as shown in this diagram;
and;
In this 12-VU-Meter Array configuration, I use one separate DAC, KORG DS-DAC-10, for gain/volume monitoring of the whole bunch of the analog line-out signal given by DAC8PRO; KORG DS-DAC-10 also has variable-volume stereo headphone-out which can be connected to the audio interface TASCAM TS-1x2HR for digital recording by separate second PC running Adobe Audition 3.0.1.
Just same as in Setup-1, I can/could hear and record the whole sum of the line-out signal from DAC8PRO under unchanged system parameters; for quantitative assessment of relative gain/volume difference between on-the-fly 88.2 kHz (or 96 kHz) conversion of the DSD-DSF track and AIFF track;
As shown in the above diagram, again JRiver’s on-the-fly PCM conversion of DSD-DSF track gives exactly 6 dB lower gain/volume compared to the PCM (44.1 kHz) to PCM (88.2 kHz) conversion even if MusicScope tells/reports both of the DSF and AIFF track have the same gain/loudness level; this is also really important message of our interest here in this thread.
Consequently, in my present comparative listening with this Setup-2, I need to boost the level/volume in 6 dB for DSD-DFS to PCM conversion, (or to suppress in 6 dB for PCM-AIFF to PCM conversion), for perfectly level-matched listening comparison. I can/could do it very easily by JRiver’s internal digital volume controller (or by DSP EKIO’s input level controller) with 0.1 dB granularity/accuracy.
Just for your possible interest, the total frequency response of this Setup-2 is like this (ref. here and here);
Setup-3:
Comparative listening to DSF into AIFF converted file and the CD-layer ripped AIFF file
In post #36 above on this thread, @voodooless kindly suggested as;
“So, next experiment: convert the DSD version to 44.1 kHz PCM, level match and ABX those.”
Nice suggestion!
I quickly converted, therefore, the extracted DSF (1xDSD) file of track-11 into 16 bit 44.1 kHz PCM AIFF using dBpoweramp Music Converter 17.3 (64-bit), and compared so prepared DSF->AIFF file and the CD-layer ripped AIFF file using MusicScope 2.1.0;
We can clearly quantitatively observe that the DSF->AIFF file has exactly 6 dB lower gain/volume than the CD-layer ripped AIFF file.
The audio system setup in this Setup-3 should be exactly the same as that in above Setup-2; my fully established multichannel stereo audio setup.
Consequently, in my present comparative listening with this Setup-3, I need to boost the level/volume in 6 dB for DSD->AIFF file, (or to suppress in 6 dB for CD-layer ripped AIFF file), for perfectly level-matched listening comparison. I can/could do it very easily by JRiver’s internal digital volume controller (or by DSP EKIO’s input level controller) with 0.1 dB granularity/accuracy.
A note for Setup-1, Setup-2, Setup-3
Now I (we) understand that I need exact 6 dB gain/volume adjustment in all of the three setups for level-matched comparative listening session.
I know well that I can easily prepare 6 dB gain/volume suppressed new PCM AIFF from the CD-layer ripped PCM AIFF by using Adobe Audition or Audacity.
In this post, however, I did not take this approach to avoid any possible very faint/subtle change in sound quality (such as minimal change in distortion) given by the gain/volume suppression algorithm within Adobe Audition or Audacity; even though I believe the audibility of such possible subtle mal-effect would be negligible.
Results of Comparative Listening Sessions
Please note that our comparative listening sessions were not in strict blind ABX, but I can say they were quasi-ABX.
Actually four people joined in comparative listening; myself, my wife (usually joining my audio enjoyment), my son and daughter (fortunately stayed at my home two days during last week).
Comparative listening in Setup-1 (exactly level matched)
Essentially all of us had no audible difference between the two files.
Only I myself occasionally “felt” that I could hear very very faint/subtle differences between the two; by “occasionally” I mean in the morning after my good sleep and rest, with seriously intensive listening efforts/attentions. I do not know this would be real or placebo since only myself "did knew which was which” as I was giving the 6 dB gain/volume adjustment.
If it would be real, I assume the very subtle difference could be attributable to JRiver’s internal algorism on DSD native feed or AIFF PCM feed and/or DAC8PRO’s internal DAC procedure between DSD and PCM (by the ES9028 DAC chip).
In any way, I too had no audible difference at all in usual relaxed comparative listening circumstances.
Comparative listening in Setup-2 (exactly level matched)
All of us had no audible difference at all between the two, even in repeated session of different sound volume. We fully agreed the total sound quality was much better than that of Setup-1, thanks to my multichannel multi-amplifier time-aligned audio setup.
Comparative listening in Setup-3 (exactly level matched)
Again, all of us had no audible difference at all between the two, even in repeated session of different sound volume. We fully agreed the total sound quality was much better than that of Setup-1, thanks to my multichannel multi-amplifier time-aligned audio setup.
Some Simple Discussions
I could essentially re-confirm that we have almost no audible difference between DSD-layer and CD-layer of well-produced well-QCed hybrid SACD like the specific disk we analyzed and listened; of course in the same HiFi audio setup and in exact level matching.
We should be careful enough about, however, this finding would not be always true for other SACD disks; it may be dependent on the mastering engineering and QC of hybrid SACD production.
As for the UHF noises frequently contained in very high amount in poorly QC-ed SACD and HiRes downloadable music tracks, I (we) have already fully investigated and discussed as I summarized here and here. Consequently, I decided always using -48 dB/Oct low-pass (high-cut) filters at 25 kHz.
My post here would be also of your reference; there I wrote;
Such a high amount of "ultra-high frequency noises; UHF noises" would be possibly harmful for your tweeters and super-tweeters, and they are highly possibly harmful for our beloved pets, I mean e.g. dogs, cats, birds.... I know some actual cases that dogs and birds became much frustrated with the UHF noises which you cannot hear.
Nowadays, I seldom role-back my audio system into the reference setup of above Setup-1 for quite rare occasion of DSD-native feed listening using single-DAC single-amp LCR-network+attenuator passive configuration.
You would please find my rationales here on my project thread for on-the-fly conversion of all the music files (DSD-DSF and PCM) into 88.2 kHz or 96 kHz PCM with always having -48 dB/Oct low-pass (high-cut) filters at 25 kHz;
- Summary of rationales for "on-the-fly (real-time)" conversion of all music tracks (including 1 bit DSD tracks) into 88.2 kHz or 96 kHz PCM format for DSP (XO/EQ) processing: #532
I hope this rather long post could be some sort of your interest and reference.
As wrote in my trailer post #38, I have almost completed my rather intensive but naive investigations on the specific SACD; even though my approaches here are not as sophisticated as @GXAlan’s thorough methods, I believe my naive objective and subjective methods are reliable accurate reproducible enough and they have high compatibility and affinity with my daily audio listening setup and environments.
First, the arrived hybrid SACD’s DSD-layer was extracted into 1 bit 2.8 MHz DSD64(1x) DSF files using the very early model of Sony PlayStation and “the” unofficial SACD ripper software. I dare not go into the details of this DSD-layer extraction method as you may kindly understand.
Then, I also ripped the CD-layer of the hybrid SACD into AIFF PCM of 16 bit 44.1 kHz as usual using dBpoweramp CD-Ripper ver.17.3 [64-bit] .
The extracted DSF tracks and AIFF tracks were taken into my music library of JRiver Media Center 30 (hereinafter abbreviated just as “JRiver”);
Since @GXAlan mainly focused on objective and subjective evaluations on the track-11 of this hybrid SACD, I did the same as follows.
I first analyzed the DSF and AIFF of track-11 by MusicScope 2.1.0;
As quite well thought/expected, as you can see easily, the DSF contains considerable amount of, but low-gain, ultrasonic noises in 34 kHz to through the upper limit of 88.2 kHz; quite common in SACD and downloadable DSD music tracks.
Just let’s take some comparative attentions on rather high frequency zone, as follows;
The music signals at 10 kHz is about -50 dB in DSF and -55 dB in AIFF; low gain and no significant difference in shape and gain.
The signals at 22.05 kHz which is upper limit in CD format is about -70 dB in DSF and -90 dB in AIFF; again low gain and no significant difference in shape and gain.
It would be interesting observing that “meaningful” music signal do exists up to 34 kHz in DSF, even though we usually cannot hear the sound beyond 22 kHz.
The typical “mountain” of ultrasonic noise in DSF reaches to -73 dB at 58 kHz; the entire noise mountain is still very low and of course inaudible even if we feed it into highly efficient super tweeters like my FOSTEX metal horn super-tweeter T925A in case if having no low-pass (high-cut) filter in the audio chain.
Let’s go back to the graphical and numeric reports given by MusicScope. I was/am very much impressed observing that the “audible music portions” of DSF and AIFF are quite and almost identical with each other in shapes, also in gain and other numeric representations. I believe that this hybrid SACD was produced with very much careful engineering and quality control on “consistency” of sound quality in DSD-layer and CD-layer.
At least judging from these reports given by MusicScope, I could assume there would be almost no audible difference between extracted DSF and ripped AIFF, if we would listen to them in exactly the same gain/volume using strictly the same audio setup.
Audio System Setup for Comparative Listening
Setup-1:
Native DSD-DSF feed vs. PCM AIFF feed into Single Stereo DAC to go into Single Amplifier and Passive LCR-Network multi-driver SP System
I believe my first comparative listening should be performed in 1xDSD native feed of the DSF track by JRiver into one stereo HiFi DAC capable of DSD decoding into analog line-level, then the line-level analog signal goes into one amplifier to drive SP system through passive LCR-network; L&R sub-woofer should be used as usual in my setup.
For comparison, of course 16 bit 44.1 kHz PCM feed into the same DAC by JRiver using exactly the same unchanged audio setup.
For this Setup-1, I can/could utilize my “single amplifier reference audio setup” which I have been using all the way through my multichannel audio project. In my current investigation in this post, I use OKTO DAC8PRO in 2-CH USB stereo DAC mode (ref. here), and ACCUPHASE E-460 integrated amplifier driving my 5-way 10-channel stereo multi-driver SP system including L&R sub-woofer as shown in this diagram.
Before going into actual listening comparison, I should objectively quantitatively measure the actual gain/volume difference between DSD feed and PCM feed in analog line-out signal given by the DAC DAC8PRO.
I did it rather naive but reliable and reproducible way as shown in the above diagram; I connected the amplifier E-460’s pre-out signal, which is “pass-through” line-out signal from DAC8PRO, into an audio interface TASCAM US-1x2HR (fully validated in my system, please refer to here and here) feeding the ADC-ed digital signal into my second PC running Adobe Audition 3.0.1 for recording. The recorded tracks were saved as 16 bit 44.1 kHz PCM AIFF track for gain/volume analysis by using MusicScope 2.1.0. Of course, the ADC and recording sequence were done with enough gain/level margin below the clipping level.
Although I already well knew that JRiver does not give 6 dB boost in DSD native digital output, I should objectively and quantitatively confirm it as shown in this result diagram.
As shown in above diagram, I could objectively confirm that JRiver’s DSD native digital feed is exactly 6 dB lower than the AIFF feed even in the case that MusicScope tells/reports both of the DSF and AIFF track have the same gain/loudness level; this is really important message of our interest here in this thread.
Consequently, in my present comparative listening with Setup-1, I need to boost the level/volume in 6 dB for DSD native feed, (or to suppress in 6 dB for PCM-AIFF feed), for perfectly level-matched listening comparison. I can/could do it very easily by JRiver’s internal digital volume controller with 0.1 dB granularity/accuracy.
Setup-2:
On-the-fly conversion of DSF and AIFF tracks into 24 bit 88.2 kHz (or 96 kHz) PCM to digitally feed into software DSP (XO/EQ/time-alignment) to give DSP-ed multi-channel digital signals into multi-channel DAC; my multichannel multi-SP-driver multi-amplifier 5-way 10-channel fully active (LCR-network and attenuators are fully eliminated) stereo audio system
This is my fully established multichannel stereo audio setup. Although DSD-DSF tracks are not natively fed into software DSP, I already proved the merit (total sound quality) of this setup well surpasses the Setup-1 configuration (i.e. native DSD feed into single DAC single amp), thanks to the direct dedicated drive of each of the SP drivers (full elimination of LCR-network and attenuator) by dedicated suitable amplifier (right person in right place) as wells as due to the establishment of 0.1 msec precision time alignment among all the SP drivers.
My present system diagram is like this;
Again, even in this Setup-2, before going into actual listening comparison, I should objectively quantitatively measure the actual gain/volume difference between DSD-DSF and PCM-AIFF, both on-the-fly converted into 88.2 kHz or 96 kHz PCM, in analog line-out signal given by the DAC DAC8PRO. How can/could I do it?
Very fortunately and nicely, I already installed my IEC 60268-17 compatible large glass-face DIY 12-VU-Meter Array in the setup (ref. here) as shown in this diagram;
and;
In this 12-VU-Meter Array configuration, I use one separate DAC, KORG DS-DAC-10, for gain/volume monitoring of the whole bunch of the analog line-out signal given by DAC8PRO; KORG DS-DAC-10 also has variable-volume stereo headphone-out which can be connected to the audio interface TASCAM TS-1x2HR for digital recording by separate second PC running Adobe Audition 3.0.1.
Just same as in Setup-1, I can/could hear and record the whole sum of the line-out signal from DAC8PRO under unchanged system parameters; for quantitative assessment of relative gain/volume difference between on-the-fly 88.2 kHz (or 96 kHz) conversion of the DSD-DSF track and AIFF track;
As shown in the above diagram, again JRiver’s on-the-fly PCM conversion of DSD-DSF track gives exactly 6 dB lower gain/volume compared to the PCM (44.1 kHz) to PCM (88.2 kHz) conversion even if MusicScope tells/reports both of the DSF and AIFF track have the same gain/loudness level; this is also really important message of our interest here in this thread.
Consequently, in my present comparative listening with this Setup-2, I need to boost the level/volume in 6 dB for DSD-DFS to PCM conversion, (or to suppress in 6 dB for PCM-AIFF to PCM conversion), for perfectly level-matched listening comparison. I can/could do it very easily by JRiver’s internal digital volume controller (or by DSP EKIO’s input level controller) with 0.1 dB granularity/accuracy.
Just for your possible interest, the total frequency response of this Setup-2 is like this (ref. here and here);
Setup-3:
Comparative listening to DSF into AIFF converted file and the CD-layer ripped AIFF file
In post #36 above on this thread, @voodooless kindly suggested as;
“So, next experiment: convert the DSD version to 44.1 kHz PCM, level match and ABX those.”
Nice suggestion!
I quickly converted, therefore, the extracted DSF (1xDSD) file of track-11 into 16 bit 44.1 kHz PCM AIFF using dBpoweramp Music Converter 17.3 (64-bit), and compared so prepared DSF->AIFF file and the CD-layer ripped AIFF file using MusicScope 2.1.0;
We can clearly quantitatively observe that the DSF->AIFF file has exactly 6 dB lower gain/volume than the CD-layer ripped AIFF file.
The audio system setup in this Setup-3 should be exactly the same as that in above Setup-2; my fully established multichannel stereo audio setup.
Consequently, in my present comparative listening with this Setup-3, I need to boost the level/volume in 6 dB for DSD->AIFF file, (or to suppress in 6 dB for CD-layer ripped AIFF file), for perfectly level-matched listening comparison. I can/could do it very easily by JRiver’s internal digital volume controller (or by DSP EKIO’s input level controller) with 0.1 dB granularity/accuracy.
A note for Setup-1, Setup-2, Setup-3
Now I (we) understand that I need exact 6 dB gain/volume adjustment in all of the three setups for level-matched comparative listening session.
I know well that I can easily prepare 6 dB gain/volume suppressed new PCM AIFF from the CD-layer ripped PCM AIFF by using Adobe Audition or Audacity.
In this post, however, I did not take this approach to avoid any possible very faint/subtle change in sound quality (such as minimal change in distortion) given by the gain/volume suppression algorithm within Adobe Audition or Audacity; even though I believe the audibility of such possible subtle mal-effect would be negligible.
Results of Comparative Listening Sessions
Please note that our comparative listening sessions were not in strict blind ABX, but I can say they were quasi-ABX.
Actually four people joined in comparative listening; myself, my wife (usually joining my audio enjoyment), my son and daughter (fortunately stayed at my home two days during last week).
Comparative listening in Setup-1 (exactly level matched)
Essentially all of us had no audible difference between the two files.
Only I myself occasionally “felt” that I could hear very very faint/subtle differences between the two; by “occasionally” I mean in the morning after my good sleep and rest, with seriously intensive listening efforts/attentions. I do not know this would be real or placebo since only myself "did knew which was which” as I was giving the 6 dB gain/volume adjustment.
If it would be real, I assume the very subtle difference could be attributable to JRiver’s internal algorism on DSD native feed or AIFF PCM feed and/or DAC8PRO’s internal DAC procedure between DSD and PCM (by the ES9028 DAC chip).
In any way, I too had no audible difference at all in usual relaxed comparative listening circumstances.
Comparative listening in Setup-2 (exactly level matched)
All of us had no audible difference at all between the two, even in repeated session of different sound volume. We fully agreed the total sound quality was much better than that of Setup-1, thanks to my multichannel multi-amplifier time-aligned audio setup.
Comparative listening in Setup-3 (exactly level matched)
Again, all of us had no audible difference at all between the two, even in repeated session of different sound volume. We fully agreed the total sound quality was much better than that of Setup-1, thanks to my multichannel multi-amplifier time-aligned audio setup.
Some Simple Discussions
I could essentially re-confirm that we have almost no audible difference between DSD-layer and CD-layer of well-produced well-QCed hybrid SACD like the specific disk we analyzed and listened; of course in the same HiFi audio setup and in exact level matching.
We should be careful enough about, however, this finding would not be always true for other SACD disks; it may be dependent on the mastering engineering and QC of hybrid SACD production.
As for the UHF noises frequently contained in very high amount in poorly QC-ed SACD and HiRes downloadable music tracks, I (we) have already fully investigated and discussed as I summarized here and here. Consequently, I decided always using -48 dB/Oct low-pass (high-cut) filters at 25 kHz.
My post here would be also of your reference; there I wrote;
Such a high amount of "ultra-high frequency noises; UHF noises" would be possibly harmful for your tweeters and super-tweeters, and they are highly possibly harmful for our beloved pets, I mean e.g. dogs, cats, birds.... I know some actual cases that dogs and birds became much frustrated with the UHF noises which you cannot hear.
Nowadays, I seldom role-back my audio system into the reference setup of above Setup-1 for quite rare occasion of DSD-native feed listening using single-DAC single-amp LCR-network+attenuator passive configuration.
You would please find my rationales here on my project thread for on-the-fly conversion of all the music files (DSD-DSF and PCM) into 88.2 kHz or 96 kHz PCM with always having -48 dB/Oct low-pass (high-cut) filters at 25 kHz;
- Summary of rationales for "on-the-fly (real-time)" conversion of all music tracks (including 1 bit DSD tracks) into 88.2 kHz or 96 kHz PCM format for DSP (XO/EQ) processing: #532
I hope this rather long post could be some sort of your interest and reference.
Last edited:
maty
Major Contributor
The sound difference has been known for a long time. It is very evident if you record an orchestral mass in DSD 1 bit, directly, without further PCM processing. The problem is that there are very few recordings available.
BTW, we need more than 16 bits (CD) to record and play the full dynamic range of some orchestral instruments.
Verified both by myself again and again in my cheap second audio system.
BTW, we need more than 16 bits (CD) to record and play the full dynamic range of some orchestral instruments.
Verified both by myself again and again in my cheap second audio system.
Last edited:
Yes, I agree!
BTW, my observation in above post is just only for "the" specific hybrid SACD.
I wrote;
We should be careful enough about, however, this finding would not be always true for other SACD disks; it may be dependent on the mastering engineering and QC of hybrid SACD production.
Last edited:
So, you checked that the CD layer mix is identical to the DSD layer in the audible spectrum? You've level matched the playback, did an ABX test?
I don't. What track actually needs a dynamic range of 96 dB? To actually hear that, you'd need the lowest sounds to be above the room noise floor. Let's say it's a generous 30 dB? You'd need the loudest peaks to be 126 dB at your listening spot. That is ear damagingly loud! An actual Orchestra concert would realistically have peaks up to 110 dB or so (there are probably exceptions), and from what I've read it's usually much quieter ( < 95 dB).
And then let's not forget that the actual dynamic range of 16-bit content can be higher when you add jitter.
And otherwise: pick a 24-bit recording that you think needs more than 16-bits, let somebody make a proper dithered 16-bit downmix, and ABX it.
maty
Major Contributor
I have some great albums in different formats: DSD, SACD, CD, Vinyl... and I learned years ago that having information about how modern recording was made is key. DSD-1 still tolerates a move to PCM in the listener's system, but if the original recording has been corrected via PCM then the "magic" of the direct sound is lost. Off course, very good recordings with physical instruments and natural voices (without Autotune).
Last night I listened again to an old album, mono from 1956, pressed on vinyl in 2018 and the sound was spectacular, better now with the new Magnat MA 900 hybrid integrated (untouched, under warranty).
The Charlie Mingus Jazz Workshop - Pithecanthropus Erectus (1956) Vinyl, Speakers Corner Records 2018, Germany
Last night I listened again to an old album, mono from 1956, pressed on vinyl in 2018 and the sound was spectacular, better now with the new Magnat MA 900 hybrid integrated (untouched, under warranty).
The Charlie Mingus Jazz Workshop - Pithecanthropus Erectus (1956) Vinyl, Speakers Corner Records 2018, Germany
So, knowing how it's recorded makes it sound better?
maty
Major Contributor
If we know that they did not take great care, it is foolish to spend money to buy it in a better format.
How do you determine whether or not they "took great care?"
And how much classical music was originally recorded on media with a higher than 16 bit dynamic range?
Keith_W
Major Contributor
That's the problem. In a live performance, this article claims that classical music can have a dynamic range of up to 90dB. I don't think I have ever been to a performance where I have subjectively heard a DR that wide, but a live performance of Tchaikovsky's 1812 Overture (they used mortars inside a concert hall!) and a live pipe organ performance which I have both heard would come very close. 16 bit audio maxes out at 90dB. This means that if someone used up all 16 bits to accurately re-create the dynamics of a live performance, you would have to wind your volume knob all the way to the max to hear the quieter passages and you would be awfully close to the noise floor. Typical classical music recordings have about 30dB of DR.
As Toole says in his book, the reason you go to a live performance is because of the sense of envelopment. I would also add, because the dynamics can not be recreated convincingly on any home audio system I am aware of, not even mine (horns).
maty
Major Contributor
If someone loves orquestral music: "big" room + open baffles + big two subwoofers is the way.
computer-audiophile
Major Contributor
The dynamics and attack of many live instruments cannot really be reproduced by a home hi-fi system. Not even a concert grand piano.
For example, we once gave a house concert at our home with large drums. Xenakis Rebonds a and b were played. I had to warn our neighbours about earthquake-like vibrations of the massive building. Fortunately they were tolerant. I can't find the photo in question. But it's the same drum set as in the following picture, where our dear friend Keiko Kida is also playing Rebonds from Xenakis.
I also have horn loudspeakers because they come closest to the live-like sound.
Last edited:
Well, I understand what you are suggesting, but I do not understand your reasons and/or rationales for it.
Why do you suggest that "You could ignore the CD layer"??
Majority of home audio users have only CD player, and therefore
Some home audio users have SACD player, but they seldom have capability/method of DSF extraction from the SACD as well as they seldom have DSF to PCM conversion capability, except for the cases of who are really computer enthusiast.
In my post #42 above, therefore, I objectively and subjectively compared DSD-layer extracted DSF with CD-ripped PCM-AIFF in Setup-1 where the DSF was natively fed into the DSD-capable DAC. This comparison would be valuable and meaningful for ordinary home audio system users having only CD player, or having SACD player.
BTW, in my post #42 above, I actually prepared the 16/44 converted PCM-AIFF file from the extracted DSF (1xDSD) in my Setup-3, and objectively and subjectively compared the DSF->AIFF 16/44 with CD-layer ripped AIFF 16/44 in my DSP-based multichannel audio system; MusicScope reported no difference except for the 6 dB gain/volume mismatch, and we heard no audible difference between the two in exact-level-matched quasi-ABX.
This means that there should/would be no audible difference in your suggested comparison of the extracted DSF and its PCM converted AIFF (16/44) in my Setup-1 in exact level matching where the DSF to be natively fed into the DSD-capable DAC.
Of course, I am talking these only for the specific SACD we are analyzing and listening.
Again, therefore, I still do not understand your reasons and/or rationales for you suggestion "You could ignore the CD layer and compare the DSF file with a 16/44 conversion of that same DSF file."
Nevertheless,,,
Today, we quickly did your suggested comparative (quasi-ABX) listening to the extracted DSF(1xDSD) vs. its DSF->PCM-AIFF(16/
Please note this time only myself and my wife participated in this quasi-ABX listening session.
The result was just the same as I wrote in my post #42 above at Result Section Setup-1;
Essentially we had no audible difference between the two files.
Only I myself occasionally “felt” that I could hear very very faint/subtle differences between the two; by “occasionally” I mean in the morning after my good sleep and rest, with seriously intensive listening efforts/attentions. I do not know this would be real or placebo since only myself "did knew which was which” as I was giving the 6 dB gain/volume adjustment.
If it would be real, I assume the very subtle difference could be attributable to JRiver’s internal algorism on DSD native feed or AIFF PCM feed and/or DAC8PRO’s internal DAC procedure between DSD and PCM (by the ES9028 DAC chip).
In any way, I too had no audible difference at all in usual relaxed comparative listening circumstances.
Note:
Just for your interest, again (as always!) we (myself and my wife) fully agreed that the "total sound quality" is considerably better in our Setup-2 audio system thanks to that "Setup-2" is my nicely established DSP-based multichannel multi-SP-driver multi-amplifier fully-active time-aligned (0.1 msec precision) 5-way 10-channel stereo system with complete elimination of LCR-network and attenuators.
Last edited:
Newman
Major Contributor
- Joined
- Jan 6, 2017
- Messages
- 3,507
- Likes
- 4,344
Thanks, appreciated.
To eliminate any difference in mastering. I thought one of your key questions was whether 16/44 was audibly inferior to DSD as a format. It seems that it is not.
- Thread Starter
- #58
Thanks @dualazmak for your analysis.
This suggests that fully volume matched in a digital environment, the differences are small. The way a DAC processes DSD vs. PCM may result in differences that can be audible.
One other example would be a 1 kHz test tone may measure differently from a DAC at 192 kHz vs 384 kHz even though 1 kHz should be fully captured by a 2 kHz sampling rate because the DAC performs differently at 192 vs 384
This suggests that fully volume matched in a digital environment, the differences are small. The way a DAC processes DSD vs. PCM may result in differences that can be audible.
One other example would be a 1 kHz test tone may measure differently from a DAC at 192 kHz vs 384 kHz even though 1 kHz should be fully captured by a 2 kHz sampling rate because the DAC performs differently at 192 vs 384
My (and I assume the OP's) interest here in this thread is the objective and subjective comparison of DSD-layer vs. CD-layer of the specific hybrid SACD! (as the title of this thread tells.)
Please let me write again.
We should be careful enough about, however, this finding would not be always true for other SACD disks; it may be dependent on the mastering engineering and QC of hybrid SACD production.
I hope OP @GXAlan will further "coordinate" this thread soon...
Because you have no idea what might or might not been 'done' to the CD layer. You don't know if it has been processed/mastered in some way that might conceivably produce an audible difference unrelated to sample rate or bit depth. That makes it another variable you can't control.
DSD and CD layers that have purposely been mastered differently are known to exist.
If you work only with the DSD layer, you control how the comparable 'CD' version is made.
tl;dr If you do find an audible difference between the DSD and CD layers of an SACD....so what? What can it tell you and what can it mean for SACD vs CD?
Last edited:
Similar threads
