 
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
|
|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
72.69.58.103
In Reply to: RE: Upsampling everything to DSD SACD posted by srl1 on November 17, 2020 at 19:18:39
I'm streaming BBC radio 3 upsampled to DSD64. It is rather nice.
Something about the sound that is pleasant. I've gone through a couple of SACD players, they all wound up failing on the SACD playback after only a few years each. That could essentially be why SACD failed most places except Japan.
Follow Ups:
I played with SACD many years ago and the sound was pleasant on several recordings. However, I thought well recorded and mastered PCM CDs were right up there with SACD... perhaps just a little "different" sounding.Fast forward a couple decades and I played with DSD downloads vs PCM, as well as on the fly upsampling conversion of PCM - using HQPlayer or DSP that is built into Roon.
My conclusion:
To my ears, SACD/DSD will make otherwise harsh sounding CD/PCM sound smoother, better, more pleasant. BUT , take a CD or PCM recording that sounds excellent to begin with and SACD/DSD simply polishes it smooth..... almost unnaturally smooth in some cases. Much of the fine texture is gone. Some of the "edge" that the artist intended to be there is also polished smooth.
Duke Ellington: I found both to be outstanding but the DSD was not quite as dynamic as the PCM
Edits: 11/18/20 11/18/20
Abe-
nice selections, as always.
I recently got an Analogue Productions SACD of Bill Evans/Shelly Manne Empathy and initially thought I had completely wasted my money when I heard the CD layer (first). The piano had a nasty "ping" to it and the recording sounded flat - the drums lacked ambience in the decay.
When I flicked to the SACD layer, the difference could not be greater - the "ping" had gone on the piano and the drums sounded incredibly natural in the recorded acoustic.
These were both played through the same PCM DAC so the conversion stage is identical - I use an Oppo 103 set to output PCM and use an HDMI link to a NAD M51 which enables the SACD layer to be output to the DAC. Since the M51 uses a 108MHz video clock, the HDMI input has the same very low jitter performance of the other inputs.
So I decided to investigate further:
I took a CD track (The Washington Twist) and converted it to DSD64. Since the conversion involves reducing the bit depth to 1 bit, adding noise-shaped dither is essential so I decided to investigate what adding dither to the standard Redbook PCM would do. I did two versions, one with simple TPDF and the other with a lightly noise-shaped dither (iZotope MBit+) for comparison.
The results were very interesting:
The DSD64 conversion of the CD layer was virtually identical to the SACD layer.
The TPDF dither solved the annoying "ping" and removed the slight harshness and restored the imaging and ambience that the DSD64 version had although it wasn't quite as smooth.
So, this says 3 things to me.....one is that there is nothing wrong with Redbook PCM when properly mastered and dithered. Secondly, I suspect that "harsh" sounding CDs are likely not properly dithered and finally, conversion to DSD64 does sound noticeably smoother and subjectively "better" when directly comparing the original Redbook version (as presented on the disc). I am convinced that this is not because of the format so much (since the final conversion to analogue was PCM in all cases) and definitely not due to bandwidth (since it is identical in all cases).
I also repeated the experiment on the new MoFi SACD of Run DMC "Raising Hell". The CD layer on this was incredibly awful with a really harsh glare on the higher frequencies (and I am not crazy about the "yelling" now that I have become older....and hesitate to call this music), but repeating the same experiment as for the Bill Evans fixed the harshness.
I therefore conclude that dither and possibly noise-shaping is responsible for the (very clear) audible differences and I am convinced that the CD layer was likely not properly dithered when it was converted from the DSD file.
Regards Anthony
"Beauty is Truth, Truth Beauty.." Keats
how converting a master to a different format can truly *improve* it.
It can change it, but now you no longer have the original.
the reconstruction filter and the group delay; i.e. phase behavior of the DAC.
It isn't about the format, it's how the conversion to analog takes place. When a base PCM signal is transcoded to a very high-rate DSD signal, the out-of-band noise of the delta-sigma modulators is pushed far above the audio band and a very gentle reconstruction filter can be used with good phase response. It doesn't change the original recording, but it does change the way your ear perceives the reproduction of the signal. I'm currently transcoding everything to DSD 128 including low-bit rate internet streams and even broadcast FM and to my ears literally everything sounds better with the simple DSD FIR reconstruction filter in my dac versus just decoding the PCM directly.
also listen to everything in DSD 128, but the Software does the conversion, not the Dac.
You can't access the Dac without Software, can you?
I am using HQPlayer to transcode (upscale) the files to DSD128. HQPlayer has the best delta-sigma modulators on the planet and they far exceed any upsampling in any silicon chip. Most modern DAC chips do have resampler/upsampling modulators built in to them and can perform that task. However, none of them come close to the performance you can attain with a powerful desktop computer running HQPlayer, or provide all the choices for modulators.
If your DAC has the capability, you can set it up to upscale everything to DSD without software, usually with some sort of button on the front of the DAC. Again, most modern chips and DAC implementation allow for this. But there are compromises because of compute power on the DAC chip and software scaling is always preferable.
It isn't about the format, it's how the conversion to analog takes place.
Perhaps, but a recording mastered at Redbook will never have the extension of 24 bit recordings sampled at higher rates no matter how you diddle with it.
If you enjoy the artificial "softening", so be it. Perhaps it's because the Revel has a rising top end response which I don't find desirable.
Courtesy of Stereophile:
in that it seeks to invalidate what I posted AND it insults my choice of speaker as defective.So here's the lecture you begged for with that, I will do my best Abe here and be a know-it-all ass:
My background is EE and I took a year of filter theory, a semester of analog filter design and a semester of discrete time processing. I designed and built an exemplary baseband PCM system for use in our university's communications lab as my senior design project and put that knowledge directly to use as the project manager and system architect. So, I have a background in this stuff, have also done all the math and actually have some experience with it. I'm not just another armchair expert.
Early cd players were criticized fo their terrible sound quality. Why? The Sony 101 didn't use any noise-shaping upsampling it was a 16 bit ladder dac PCM converter with a steep analog filter in the output stage. Those reconstrucion filters messed with phase and sounded bad. Philips first implementation used oversampling with a 14-bit converter and a less steep reconstruction filter at a higher frequency and it did sound better. The loonies diy'ers of the time thought the early 16 bit Philips dac chips of the second generation sounded best with no oversampling at all and a similar more gentle filter. There was eventually a convergence of thought on why all of this was. It was the way that the digital imformation was converted to analog, not the digital data that was making a difference and one of the most important aspects was keeping the output stage reconstruction filter far away from the frequency of the musical content, moving it much higher in frequency so it impacted the musical signal much less. I listened to a Sony CDP-101 in 1982 and is really was terrible.
How important are frequencies above 22kHz? It depends, I suppose because of harmonics. Sampling theory tells us that in a band-limited system like one from 0-22kHz all of the frequency information is preserved when sampling it using known methods. Transient information might be an exception because of harmonics which is what an impulse response is. You can't get it back, no. It is gone and conventional upsampling doesn't interpolate new information it just changes the data format. Why bother? Because we also know that changing the data to a 1 or multi-bit delta or pulse width like DSD allows the use of a simple low-pass filter for reconstruction. There is no discrete DAC needed, just clock the data in and send it through a filter, the best case a really good digital FIR with a gain stage behind it. That isn't smoothing anything. It is preserving the information you have in your original file while moving the filter corner frequency out to such a high frequency and gentle slope that it doesn't sound terrible.
Regarding my speakers, I have run sweeps and built very high quality FIR room correction filters in acourate that provide a frequency response within +-2db from 30Hz-24Khz at my listening position, which is extraordinary real-world performance. Impulse response and interaural coherence coefficient are also quite exceptional with my latest builds.
Edits: 11/20/20
I'm not arguing your precious PC based delta-sigma modulator. I simply called you out for being flat out wrong that "CDs are .wav". 100% undeniably incorrect, Mr EE.As for preferring quality PCM over DSD, that's just my subjective preference. Nothing to argue about here.
Edits: 11/20/20
You invited this sort of snark years ago with your insufferable and seemingly uncontrollable pedantry and one-upsmanship. I think you have illustrated a great many times why you have earned my derision. It wasn't about the .wav error, it was about the 1000 times previously that you jumped into a thread in order to be right about something/anything. Point to Abe. Hurry, go put that on the bedpost of your online life. Its pretty apparent that this is your only form a joy these days.
Speaking for myself, I'm happy when someone who is right (about anything) jumps into a thread! ;-)
So you bring my name into a thread that I have nothing to do with..... as you rant on about your world dominating PC based delta-sigma modulators, supposed EE, and extensive background in filter theory, etc. I don't care. I wasn't part of that thread.Please leave me out of your shit storm. I want no part of it.
Edits: 11/21/20 11/21/20
You're welcome to your own preferences.
I've been dabbling with upsampling to DSD on the fly using Audirvana on my iMac. It can do up to DSD256 without issue, I've tried 512 but it couldn't keep up.
It does make any lower res recordings sound better to my ears as my DAC has native DSD conversion. Much smoother on CD and even on 24/44 or 24/48. I've even left it on for 24/96 but at that rate I didn't think it was better. It wasn't objectionable either. The only issue is that it won't switch the conversion on the fly based on the bit rate of the source. I'd have to get up from the couch...
| FAQ |
Post a Message! |
Forgot Password? |
|
||||||||||||||
|
||||||||||||||
This post is made possible by the generous support of people like you and our sponsors:
Top of Page ]
[ Contact Us ]
[ Support/Wish List ]
[ Copyright © 2024, Audio Asylum, all rights reserved ]
