Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

Extremely

Very

Not too important

Never understood what DSD was about or cared much

[scottm_dj]

Quite simply, DSD playback (SACD ISO) is by far the best feature of JRiver.   For someone who has carefully ripped all my years-cultivated discs (it can be done with the right player--no need for Playstation anymore), and having a HQ Exasound analog MultiChannel DAC--the time spent is totally worth it.  You can even categorize them under 2-channel, Quad and Multichannel under playlists.  You haven't lived til you've heard an expert Multichannel mix from the likes of a pro like Eliot Schneiner.

i really wish DVD-AUDIO playback would be added as well!  There is no way to play that format other than off disc or via the iSO file an older non-updated Oppo bluray player.    I have at least as many of those discs than i do SACD and so far there's been no way to stream them.

Forget the Bluray formats and its 87.1 channels, good 'old 5.1 channel SACD and DVD-Audio were the best inventions for musical reproduction.

[pschelbert]

i really wish DVD-AUDIO playback would be added as well!  There is no way to play that format other than off disc or via the iSO file an older non-updated Oppo bluray player. 

No problem. rip it with a DVD-rip software to flac.
Then play the multichannel flac.
Much easier to do than the encrypted SACD to rip.
MC can play multichannel flac as of today.

[RD James]

i really wish DVD-AUDIO playback would be added as well!  There is no way to play that format other than off disc or via the iSO file an older non-updated Oppo bluray player.    I have at least as many of those discs than i do SACD and so far there's been no way to stream them.

Forget the Bluray formats and its 87.1 channels, good 'old 5.1 channel SACD and DVD-Audio were the best inventions for musical reproduction.

There's a plug-in for foobar2000 which can play DVD-A tracks. You can use this to convert them to FLAC files which work in Media Center.
It would be nice if MC could handle this natively, but when a free solution exists and the conversion only needs to be done once, I don't mind things being as they are. It's easier than ripping SACDs.

[livingondogfood]

DSD is not important to me, but just like JRiver Media Center supports .mp3 I think it should support DSD especially since I suspect a lot of the people that would be inclined to buy JRiver Media Center would expect this and use it.

[Opawesome]

Just adding my two cents here after reading some comments.

The poll is not whether which is the best format between PCM and DSD but whether support of DSD by Media Center is important for its users.

IMHO, DSD support is paramount because SACD (which is a very common media for classical music) uses DSD and, as I understand it, conversion between DSD and PCM is not a bit-perfect process.

With regards,

Opawesome

[michael123]

Just adding my two cents here after reading some comments.

The poll is not whether which is the best format between PCM and DSD but whether support of DSD by Media Center is important for its users.

IMHO, DSD support is paramount because SACD (which is a very common media for classical music) uses DSD and, as I understand it, conversion between DSD and PCM is not a bit-perfect process.

With regards,

Opawesome

Not just about conversions - DACs are different, and some are better at handling DSD natively

[jacobacci]

The support for DSD in JRiver is extremely important to me.

JRiver is by far the best audio playback platform for me. The key capabilities that set it apart are:

• ability to select files with filters and expression and flexible and powerful functionalities to manipulate selected files (absolutely unique)
  
• ability to play DSD / SACD ISO (THANK YOU for ISO sidecar tagging)
• powerful DSP engine

The reason why DSD is important to me is not because it sounds better, but simply because certain music is only available in DSD format. I believe in as few format conversions as possible. Music recorded or mastered in PCM I prefer to listen to in PCM. Music recorded native in DSD I prefer to listen to in DSD.

There is a lot of confusion around DSD. Music recorded in DSD, then converted to PCM for mastering and then re-converted to DSD for publishing is in my view simply marketing snake oil (because DSD "sounds more analogue"?). If it sounded good after the PCM mastering, simply leave it in that format. If it needed the further conversion to DSD to sound good, the (PCM) mastering failed its purpose.

A few things I would like to see for DSD in JRiver:

• ability to downsample from DSD256 to DSD128 in DSP Studio (like the PCM conversions). Not every DAC supports DSD256 and it would be nice to not have to convert DSD256 to PCM and define several Zones for the same DAC
• Maximise the quality of DSD to PCM conversion (DSD Master http://dsdmaster.blogspot.com/ is the benchmark to my ears)
• more filter options for DSD to PCM conversion

Thank you and keep it up. JRiver is the best thing since sliced bread!

[Awesome Donkey]

ability to downsample from DSD256 to DSD128 in DSP Studio (like the PCM conversions)

The thing is if this was added, it'd have to go through a PCM conversion in the middle as there's no direct DSD conversions so it'd be DSD256 > PCM > DSD128. Honestly you'd probably be better off converting to PCM in that case, IMO.

[jacobacci]

The thing is if this was added, it'd have to go through a PCM conversion in the middle as there's no direct DSD conversions so it'd be DSD256 > PCM > DSD128. Honestly you'd probably be better off converting to PCM in that case, IMO.

Totally agree. No need to go to PCM and back to DSD. I was hoping there would be a way to downsample DSD -> DSD directly, but the math does not seem to exist or to be extremely complicated.

[michael123]

The support for DSD in JRiver is extremely important to me.

JRiver is by far the best audio playback platform for me. The key capabilities that set it apart are:

• ability to select files with filters and expression and flexible and powerful functionalities to manipulate selected files (absolutely unique)
  
• ability to play DSD / SACD ISO (THANK YOU for ISO sidecar tagging)
• powerful DSP engine

The reason why DSD is important to me is not because it sounds better, but simply because certain music is only available in DSD format. I believe in as few format conversions as possible. Music recorded or mastered in PCM I prefer to listen to in PCM. Music recorded native in DSD I prefer to listen to in DSD.

There is a lot of confusion around DSD. Music recorded in DSD, then converted to PCM for mastering and then re-converted to DSD for publishing is in my view simply marketing snake oil (because DSD "sounds more analogue"?). If it sounded good after the PCM mastering, simply leave it in that format. If it needed the further conversion to DSD to sound good, the (PCM) mastering failed its purpose.

A few things I would like to see for DSD in JRiver:

• ability to downsample from DSD256 to DSD128 in DSP Studio (like the PCM conversions). Not every DAC supports DSD256 and it would be nice to not have to convert DSD256 to PCM and define several Zones for the same DAC
• Maximise the quality of DSD to PCM conversion (DSD Master http://dsdmaster.blogspot.com/ is the benchmark to my ears)
• more filter options for DSD to PCM conversion

Thank you and keep it up. JRiver is the best thing since sliced bread!

+1

[Alex M]

MC does not read the metadata from the .dff files. There are not many of them, only artist chunk and title chunk, but it's better than nothing.

[RD James]

MC does not read the metadata from the .dff files. There are not many of them, only artist chunk and title chunk, but it's better than nothing.

The DFF format does not support metadata.
You can pull that information from the filenames.

[Awesome Donkey]

And you can also use the sidecar feature with DFF files too.

[Alex M]

The DFF format does not support metadata.

The DFF format supports metadata:
https://dsd-guide.com/sites/default/files/white-papers/DSDIFF_1.5_Spec.pdf

[Cryptographer]

For me the most important difference between DSD and PCM is the precision of phase shifts.

Assume you have a 18000 Hz signal on the left speaker and a 18001 Hz signal on the right. With DSD you will hear (if you can it that high at all) an interference swinging between the speaker once a second. In PCM, at worst with 44100 Samples per second, there are not so many ways how to encode these signals. As a result, the interference will not swing, but hop between the speakers.

You could extend the test with 18002 and 18003 Hz on the rear speakers. So it should be obvious that the more channels you listen to the more important the exact phases become.

Though higher PCM sampling rates reduce the problem, DSD is still the only coding technology not having such a problem at all, resulting in a so much more vivid sound.

[pschelbert]

representation of timing precision and phase precision does not depend on samplerate.
This is a widespread misconception in digital signal processing.
Hence CD can very precisely reconstruct pahse at 20kHz.

[RD James]

The DFF format supports metadata:
https://dsd-guide.com/sites/default/files/white-papers/DSDIFF_1.5_Spec.pdf

I stand corrected. I guess most software just doesn't support tags in DFF files.

For me the most important difference between DSD and PCM is the precision of phase shifts.

Assume you have a 18000 Hz signal on the left speaker and a 18001 Hz signal on the right. With DSD you will hear (if you can it that high at all) an interference swinging between the speaker once a second. In PCM, at worst with 44100 Samples per second, there are not so many ways how to encode these signals. As a result, the interference will not swing, but hop between the speakers.

You could extend the test with 18002 and 18003 Hz on the rear speakers. So it should be obvious that the more channels you listen to the more important the exact phases become.

Though higher PCM sampling rates reduce the problem, DSD is still the only coding technology not having such a problem at all, resulting in a so much more vivid sound.

This is a fundamental misunderstanding of how sampling and digital audio works.
I highly recommend that you watch the entire video, but here is a demonstration showing how timing precision is unaffected by sample rate: https://www.youtube.com/watch?v=cIQ9IXSUzuM#t=20m53s

[Alex M]

I guess most software just doesn't support tags in DFF files.

I do not care what other software products can or can not do. I just want MC to get a little better.
For DFF files, I regularly use AudioGate, MC and Tau. AudioGate can not only read, but also edit the "Artist" and "Title" tags. For Tau, due to its specifics, this is not a problem at all. But MC perfectly reads SACD images, which neither AudioGate nor Tau can do.

[Cryptographer]

I highly recommend that you watch the entire video, but here is a demonstration showing how timing precision is unaffected by sample rate: https://www.youtube.com/watch?v=cIQ9IXSUzuM#t=20m53s

What you see in the video is at 20:53 is a square wave with 44 samples/square, or round about 1 kHz at CD quality. Any yes, it is true you can place a single pulse with the nearly exact timing. But things become different the nearer you come to the Nyquist frequency. At 22.05 kHz only 0° and 180° phase are possible.

Even if the sampling theorem is fulfilled, aliasing may occur, an effect you can similarly be see on a computer screen when diagonal and curved lines are displayed as a series of zigzag horizontal and vertical lines. Checked jackets are mostly shown restlessly on a video.

Such effects can be reduced by oversampling (an option most video equipment does not have). And you need more or less intelligent algorithms to correctly e.g. put an 18 kHz sine wave into a 44.1 kHz raster, which will always remain a compromise. With the 1 Hz difference between the channels you will detect it.

[pschelbert]

no, you can have a resolution which is much higher, you can resolve depending on the accuracy of the sampling for example 0.0001degrees at 20kHz, not just 0 and 180 degrees.
Look it up in digital signal processing.
Oversampling does not help in accuracy, it just helps in easy filtering. Nyquist issues you will still have the same if you oversample or not.
The oversampling process in itself needs to deal with nyquist. No way around mathematics.
You will never jump over your shadow, no matter how fast you run...

[Cryptographer]

no, you can have a resolution which is much higher, you can resolve depending on the accuracy of the sampling for example 0.0001degrees at 20kHz, not just 0 and 180 degrees.

At 22.05 kHz (not at 20kHz) only 0° and 180° phase are possible. A higher precision at for instance 20 kHz can only be an average with the current phase continuously shifting around the desired value. Otherwise you could convert a range of 20 kHz between 44.1 kHz and 48 kHz as often as desired without a quality loss.

Compared to photography, oversampling can help to prevent moiré effects between pattern and resolution. This also applies to acoustics and does not contradict the Nyquist theorem as the final resolution is not affected.

[BillT]

Actually aliasing in a properly designed and implemented linear PCM system is not an issue. Audio signals above Fs/2 will be removed by the low pass filters.

Even in inadequately designed systems it's not much of an issue; in the real world of acoustic instruments audio signals above 20kHz are uncommon and at a very low level.

[pschelbert]

oversampling does not enhance precision and or quality.
No magic possible here

[dtc]

oversampling does not enhance precision and or quality.
No magic possible here

Difference issue, but oversampling moves the deleterious effects of the brick wall filters (22.05Khz for NOS DACs) out of the normal hearing range, which many believe increases the quality of the conversions and the sound.    The D to A conversions produces noise above the Nyquist frequency and how that is handled does effect the result. To avoid ringing from the brick wall filter, most PCM DACs oversample these days. Not magic, just straightforward physics.

[pschelbert]

yes, true.

In the oversampling process a digital brickwall filter needs to be implemented. No way around a brickwall filter be it analog or digital.
Yes Oversampling makes for easier DA conversion, though potential higher quality. I.e. the quality which the music-file has, can be converted to analog without further loss in quality. The quality of the music-file is not enhanced, in the best case preserved.

[michael123]

Even in inadequately designed systems it's not much of an issue; in the real world of acoustic instruments audio signals above 20kHz are uncommon and at a very low level.

This is debatable

[BillT]

Of course!

Those with a vested interested in wasteful systems will argue that there is a meaningful level of supersonic signal in music.

[michael123]

Of course!

Those with a vested interested in wasteful systems will argue that there is a meaningful level of supersonic signal in music.

For these with wasteful interest in audio quality,
That's what DSD subject is about and nothing else

[kr4]

Apparently, it is debatable!   

[JimH]

Everything is debatable.  Sometimes it's like having teenagers.

[Fitzcaraldo215]

Whether DSD is wasteful or not, whether it sounds better or not, whether humans can hear ultrasonic or not aren't the issues here.  The point is there is a lot of great sounding music a fair number people want to collect and play that happens to have been released in DSD, mainly on SACD. 

There is no definitive answer to the question does the CD layer of a hybrid SACD sound just as good as the DSD Stereo layer and should we therefore insist that people preferring the DSD layer should be cut off, ostracized and ignored? Some might say yes, some no. Ok, but what about the Mch DSD layer?  Should that be ignored too?

Fortunately, the powers that be at JRiver want to continue to support the format in stereo and in Mch.  They do not wish to be engulfed in a petty, meaningless debate about the relative sonic merits or the wastefulness of the format. If a segment of the user base wants to use it, let 'em have it.  Why not?  And, what are there useful improvements that can be made to the support of it?

But, if for whatever reason you don't like DSD, just don't use it. 

[Daydream]

there is a lot of great sounding music

I actually got curious (I'm not an audiophile), and was planning to buy a DAC that does DSD256 just for kicks, to hear things for myself. My problem is with the idea highlighted above. Maybe I don't know enough about this market but how and what music is available in this format? Take out the symphonic, the dead artists, and niche ones that only 100 people heard about them. What's left? SACD music that one needs to buy the discs from Mars, a 1000 player to rip them, and even after that there's the problem (aka not a quality guarantee) of "how was this mastered, from what source?"

This is just too niche to even warrant a debate. JRiver supports it, just as they support 1000 other things; why does this poll even exist?

[Alex M]

What's left? SACD music that one needs to buy the discs from Mars

If you think that Japan is on Mars, you should teach geography 

[bmoura]

what music is available in this format?

Check the NativeDSD Database at https://www.nativedsd.com/database
Lists over 30 music download sites with DSD Downloads available.  Most in Stereo, some sites offer Stereo, Multichannel and Binaural DSD.

[pschelbert]

there are 1430 albums on nativedsd.

How many which are not classical music see yourself.

[michael123]

Everything is debatable.  Sometimes it's like having teenagers.

Yes but that's extreme. My both kids (14, 18) first say 'No' to everything

[Fitzcaraldo215]

I actually got curious (I'm not an audiophile), and was planning to buy a DAC that does DSD256 just for kicks, to hear things for myself. My problem is with the idea highlighted above. Maybe I don't know enough about this market but how and what music is available in this format? Take out the symphonic, the dead artists, and niche ones that only 100 people heard about them. What's left? SACD music that one needs to buy the discs from Mars, a 1000 player to rip them, and even after that there's the problem (aka not a quality guarantee) of "how was this mastered, from what source?"

This is just too niche to even warrant a debate. JRiver supports it, just as they support 1000 other things; why does this poll even exist?

Www.hraudio.net lists all SACD and BD-A releases in all genres, stereo/Mch, etc.  Many may, however, be OOP and no longer available new.  There are over 10,000.  I have well over 4,000 albums on my NAS, mostly classical, mostly Mch.  New releases keep showing up, increasingly via download.

Yes, it is a niche, but growing in interest now that more people can rip SACD via certain Oppo, Pioneer, Sony, etc. players. But, yes, it is predominantly classical, so it will never be more than a niche, one that many have never heard of.  It is of special interest to those who prefer music in Mch, like me, since SACD/DSD are overwhelmingly the main sources of Mch music.

I believe the poll exists for JRiver to determine what needs or issues DSD users have and to gauge overall level of interest and what resources to allocate to DSD support beyond just continuing current support, which is very satisfactory to me, by the way.

[tij]

Www.hraudio.net lists all SACD and BD-A releases in all genres, stereo/Mch, etc.  Many may, however, be OOP and no longer available new.  There are over 10,000.  I have well over 4,000 albums on my NAS, mostly classical, mostly Mch.  New releases keep showing up, increasingly via download.

Yes, it is a niche, but growing in interest now that more people can rip SACD via certain Oppo, Pioneer, Sony, etc. players. But, yes, it is predominantly classical, so it will never be more than a niche, one that many have never heard of.  It is of special interest to those who prefer music in Mch, like me, since SACD/DSD are overwhelmingly the main sources of Mch music.

I believe the poll exists for JRiver to determine what needs or issues DSD users have and to gauge overall level of interest and what resources to allocate to DSD support beyond just continuing current support, which is very satisfactory to me, by the way.

JRiver is niche ... whether its music (dsd, hires, ect) or video (3d, madvr, etc)

masses do streaming (netfix, imusic, etc) these days

[stanzani]

For me the key disadvantage is that DSP could Not be Used.

+1
This is the reason why I always convert DSD to PCM @192KHz

[kr4]

+1
This is the reason why I always convert DSD to PCM @192KHz

This is the reason why I almost always convert DSD to PCM @176.4KHz on the fly.  Someday, I may not need to.

[Jack McCue]

I use DSD heavily for multichannel classical music (exaSound e38). I do not like any type of conversion from the original files for any format, if it is under my control. I have been an audiophile since high school (graduated 1961) and I can barely count the number of times file conversion has been erroneously said to make no difference in sound...(and now MQA!).

That said, I think it primarily matters for classical, and in particular, orchestral/choral/opera. It requires a good system and careful listening, but there is a difference between the original and flac and pcm and etc., etc. All that really matters is whether it matters to you. Some people think mp3 sounds good!

[dtc]

+1
This is the reason why I always convert DSD to PCM @192KHz

MC converts DSD by first converting to 1/8th the sample rate of the original file. That means you get 352 KHz or 705KHz or 1410Khz PCM, all of which are integer multiples of 176 KHz. Downsampling to 192KHz means you are doing a non-integer conversion.  You might try converting to 176 KHz instead. if you can play that, just because the conversion is faster and consumes less CPU. 

[RD James]

MC converts DSD by first converting to 1/8th the sample rate of the original file. That means you get 352 KHz or 705KHz or 1410Khz PCM, all of which are integer multiples of 176 KHz. Downsampling to 192KHz means you are doing a non-integer conversion.  You might try converting to 176 KHz instead. if you can play that, just because the conversion is faster and consumes less CPU.

Using SoX I see an 0.1% difference in CPU usage between resampling to 192kHz or 176.4kHz on my system (3.05% vs 3.15%).
Without SoX the difference is slightly larger at roughly 0.5%, but CPU usage for both is higher than using SoX.
And that's with three other VST plug-ins in the DSP chain too.

[Hendrik]

Integer or non-Integer resampling is not very relevant with modern resampling algorithms. The quality and resource usage differences are minimal to non-existant. If anything a difference of load might be the result of the different amounts of data that need to be processed with different sample rates.

[dtc]

OK - I know this used to be an issue. Sounds like it is not anymore. I have not tested in a while. My bad.

EDIT : Actually, I just tested it, and did see a big difference. A 264 MB 4xDSD dsf file took 67 seconds to convert to 96 KHz flac and only 34 seconds to convert to 88 KHz, using SoX.

[Hendrik]

Speaking of DSD, this is finally coming in the next build:
NEW: Added a DSD Bitstreaming option to select the maximum DSD sample rate to bitstream, and otherwise fallback to decoding.

You can find it in the Custom Bitstreaming dialog.

[dtc]

Speaking of DSD, this is finally coming in the next build:
NEW: Added a DSD Bitstreaming option to select the maximum DSD sample rate to bitstream, and otherwise fallback to decoding.

You can find it in the Custom Bitstreaming dialog.

Awesome! Thanks.

[michael123]

Speaking of DSD, this is finally coming in the next build:
NEW: Added a DSD Bitstreaming option to select the maximum DSD sample rate to bitstream, and otherwise fallback to decoding.

You can find it in the Custom Bitstreaming dialog.

Finally
Many thanks

[nsxelent]

DSD is a must have for me.  I have a lot of SACDs that I've ripped which have mastered versions that so far have only been released on SACD.  Even the CD layer on the same disc has a different mastering.  Love that JRiver has supported bit-perfect DSD for so long!

[thomaspf]

Fantastic, thank you so much!!