To those concerned.
I've done enormous amounts of research and listening tests after David
(BlueSaint) showed me the significance of DSD encoding on PCM files.
Playing PCM directly may be slightly more airy sounding and overall blended together, but in terms of sonic accuracy, it is levels behind DSD512.
DSD over DoP on JRiver Media Center is
15% better
-
Note: DSD over DoP even on a MacBook will still sound better (more articulate) than standard PCM, however what you want is Native DSD Encoding.
DSD512 encoding (standard algorithm) on JRiver Media Center is
30% better.
DSD512 encoding (advanced algorithm) such as "poly-sinc-xtr-2s" from HQPlayer is
40% better.
DSD512 encoding (similar to CORD DAVE algorithm) such as "poly-sinc-xtr" from HQPlayer is
46-48% better.
Jussi Laako (Signalyst)
Developer of HQPlayer's comment on the poly-sinc-xtr filter.
https://community.roonlabs.com/t/xtr-filters-why-do-they-sound-so-good/25687
The
poly-sinc-xtr filter was created by Jussi to mimic that of what was done in the Chord Dave when it comes to filtering.
UNFORTUNATELY to run this filter you will need a super computer that has
10 CPU CORES running at
90% workload in order to properly encode and play music simultaneously.
FORTUNATELY you can run the 2s version of this filter which is less computer intensive and only requires
4 CPU CORES at around
70% workload (year 2013 or higher).
The 2s version (poly-sinc-xtr-2s) is a two stage filter that jeopardizes some filtering quality in order to lower CPU workload.
The non-2s version (poly-sinc-xtr) is a single stage filter that offers better a algorithm which results in less ringing and provides better frequency response and signal attenuation.
LONGER EXPLANATION:
The highest modulation available right now is DSD512.
Both HQPlayer and JRiver Media Center offers this level of processing.
HQPlayer has multiple "FILTERING" options whereas the JRiver only has a stock profile.
The problem with stock filters is that you cannot improve sound quality by other filters that are more intensive when it comes to processing of the DSD encoded file.
A higher end filter will use more CPU processing power which results in more cores being used.
In HQPlayer there is a filter option known as poly-sinc which is one that is better than the stock filter available on JRiver Media Center.
These filters offer better linearity of the music that's being encoded/produced as well as cut off pre and post ringing you hear in music that makes it sound un-natural.
To take things even higher, poly-sinc has an extreme mode available (called poly-sinc-xtr) that takes 5 times more processing power compared to standard poly-sinc.
In order to run the poly-sinc-xtr filter, you must have very fast computer with a CPU that has 10 cores working at 90% in order to convert PCM to DSD512 using that filter. Unfortunately I don't have a computer that fast thus why I have to go with the second best option and that is to separate the conversion stream into 2 stages (2s) for oversampling instead of one. A single stage offers a better algorithm of converting the file which results in less ringing and provides better frequency-response and signal attenuation compared to a 2 stage (which jeopardizes those listed things a little in order to lower CPU workload).
END RESULT:
The setup I currently have consists of a 16 CPU CORES (AMD THREADRIPPER) in order to run
POLY-SINC-XTR-MP.
Note: mp = minimum phase (adds more post echo to help with transient reproduction)
- sounds better for most modern electronic music.
- I use this very often since I feel we need some subtle post echo's in order for music to sound a little more natural.
Cheers,
Thanh