Print

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

[The Computer Audiophile]

Simple, DSD is not a format that can be streamed while being converted.

Hi Bob, is there a reason you guys choose not to enable converting and streaming DSD?

[bob]

Hi Bob, is there a reason you guys choose not to enable converting and streaming DSD?

You can't generate a content-length in advance.

[stefano_mbp]

How do you think it is done in Audirvana and Roon? I don’t think they are some kind of wizard ...
As I could read elsewhere...
“ So the raw DSD audio signal output from JRiver's DSP Studio has to be packaged into:
either one of the (DSF or DFF) DSD audio file formats;
or DoP as raw 24-bit LPCM;
or DoP packaged in a lossless LPCM file format, so WAV, FLAC, etc.
 
It is the packaging part that is missing.”

Is it so difficult?

[Mike Rubin]

Please allow me to request that MC27 have the ability to use DSP to convert and serve files as DSD256.

I own Audirvana and am on a Roon trial.  Both applications convert files on the fly to DSD 256 for DLNA streaming.  To say that my two DACs love that processed DSD signal is an understatement. They sound marvelous.

At the same time, I despise the file management capabilities and interfaces of both Audirvana and Roon.  I also like their remotes a bit less than EOS, JRemote and JRemote2.    I find now that I am using JRiver as much as I ever did for metadata editing and artwork retrieval, but then using the other two applications for playback.

A MediaCenter 27 that can upsample to DSD256 for DLNA streaming would be the only player I'd need and use.  Please!

[Awesome Donkey]

Can you not do that right now by setting DSP Studio > Output Format > Output Encoding to 4xDSD then sending that?

[Mike Rubin]

Can you not do that right now by setting DSP Studio > Output Format > Output Encoding to 4xDSD then sending that?

No.  Not over DLNA.  I can and used to do it over USB to my DAC using an ASIO driver.

When I set up a new DLNA server that includes DSP, the upsampling to DSD option is not grayed out, so I select it.  Unfortunately, selecting that option does nothing actually to upsample PCM files and it prevents native DSD files from playing altogether.

Because JRiver doesn't upsample streamed files to DSD over DLNA, the DSP Studio setup options for DLNA servers should not include DSD upsampling options on the pick list.  They should be grayed out or suppressed for users setting up DLNA rather than ASIO connections.

[RoderickGI]

You can't generate a content-length in advance.

You had a couple of ideas to get around this for generally for Streaming, which I suspect would work for DLNA as well.

There are a couple of different ways to do the kludge and I think I'll do the one I didn't start down the path on.

I can see why the playlist items aren't getting transcoded, essentially the playlist itself would need to be re-written when sent.

While converting? They are converting something else to DSD and creating a fully seekable stream? To be able to do that you have to generate the content-length in advance while the file is in the process of being converted.

Maybe its time to implement a solution, even if it means that the playback isn't fully seekable?


Mike, is playback seekable when converting to DSD and streaming via DLNA in Roon and/or Audirvana? While Bob and JRiver might want to implement playback the correct way, maybe this is the compromise that is required. Particularly if the solution the other applications use has the same limitation. That is, playback is not seekable.

[Mike Rubin]

Roderick, thank you for your leadership on this issue.  Unfortunately, my understanding of these processes is limited enough that I don't know what "seekable" playback means.  Please let me know what I need to look for in Roon and Audirvana and I will let you know how they handle things.

[RoderickGI]

Seekable just means that you can move forward and backward in the playing Track using the Progress Bar in the application.

When an application doesn't know the content-length of the playing media, jumping forward and backward can either be very inaccurate, difficult, or impossible. It depends on the format of the file as well as the application and the DLNA Renderer, I believe. It isn't as simple as telling the Renderer to jump forward or backward 2 minutes. I think such jumps are done by percentage or similar. There has been some discussion on this in the past, but I don't recall the technical details.

Anyway, "Fully Seekable" means that you can seek (jump) to any part of the media file at any time, accurately. If seeking is inaccurate you may try to jump ahead 1 minute, and end up 15 minutes ahead, or just at the end of the currently buffered data. Or you may try to jump ahead to halfway through the file, but end up just a short time ahead of the current position.

Some other applications just set the content-length to the maximum value allowed, which can be 4 hours, for example. So if you have a 4 minute file which is still being converted for streaming via DLNA, and the content-length is set to 4 hours, and you jump from near the beginning to what you think will be 2 minutes into the file (halfway on the Progress Bar), the application may just move to the 2 minute position relative to the 4 hour content-length, so (2x60 seconds / 4x60x60 second) x 4x60 seconds = 2 seconds. You try to jump 2 minutes ahead and only jump 2 seconds ahead.

I may be completely wrong in the details, but it is something like that.

[Mike Rubin]

Thanks for the explanation.

I listen mostly to instrumental jazz, so I may not be the very best tester for this, but, as far as I can tell, Roon is fully seekable and Audirvana is not.  With Roon, the progress bar stays put and plays from the selected spot when you advance  it or move it backwards.  With Audirvana, the progress bar can't be advanced beyond the apparent buffer and it won't stay to where you drag it when you back it up. I can do more testing if there's something specific I should try out.

[The Computer Audiophile]

Audirvana uses DLNA, Roon does not.

[Mike Rubin]

Chris I thought that Audirvana uses something proprietary, which is why, unlike JRiver, it won't show up as a server in a plain vanilla DLNA controller app.  Is that not the case?

[The Computer Audiophile]

It may change the way it is seen by DLNA controllers, but it must use DLNA if it's to send audio to DLNA renderers. Roon uses RAAT and can't send audio to DLNA devices.

[Mike Rubin]

Thanks.  Makes sense.

[bob]

DSD format option for DLNA in the next build of MC27.

[Matias]

Finally! Is DSD x2, x4 and x8 coming too?

[bob]

Finally! Is DSD x2, x4 and x8 coming too?

Perhaps. We'll see if this gets any use first.

[stefano_mbp]

Just trying now on 27.066 ... wonderful job bob! I’ll upgrade my MC26 soon.

[Mike Rubin]

DSD format option for DLNA in the next build of MC27.

I am not having luck getting this to stream files as dsd.

i added a new DLNA server and entitled it ”DSP.”

Mode is Specified Output Format.

Format is DSF.

Advanced is DSP Studio.

Studio Output Encoding is 4xDSD native.

Yet, DAC only shows file’s original PCM bit depth and sample rate. (Roon and Audirvana streamed DSD show as “DSD“ on same DAC.)

What am I doing incorrectly?

[Mike Rubin]

[Deleted.]

[stefano_mbp]

DLNA DSD upsampling is limited to DSD64 as described in the change log
.
27.0.61 (2/4/2021)
........
7. NEW: DLNA conversion to DSD (dsf 1x)
.........
and ... (maybe a silly question) have you checked OUTPUT FORMAT?

[Mike Rubin]

DLNA DSD upsampling is limited to DSD64 as described in the change log
.
27.0.61 (2/4/2021)
........
7. NEW: DLNA conversion to DSD (dsf 1x)
.........
and ... (maybe a silly question) have you checked OUTPUT FORMAT?

D'oh!  I am an idiot. Not only does the changelog say this, but so do the posts right above mine!

Thanks! I will try DSD64. (And, yes, I did change the format box.)

[bob]

This is native DSD, not DoPE.
I don't believe the DSP studio setting of output format in the DLNA server here will have any effect and if it does it's probably negative.
It works fine playing to a MC renderer.
If your DAC won't play it you may have a mimetype issue whic you may or may not be able to fix by messing with the fileassociations.xml.
Renderers aren't consistent about what they expect for dsf mimetype.
MC as a renderer will accept
audio/x-dsf
audio/x-dsd
audio/dsd
and
audio/dsf

[Mike Rubin]

This is native DSD, not DoPE.
I don't believe the DSP studio setting of output format in the DLNA server here will have any effect and if it does it's probably negative.
It works fine playing to a MC renderer.
If your DAC won't play it you may have a mimetype issue whic you may or may not be able to fix by messing with the fileassociations.xml.
Renderers aren't consistent about what they expect for dsf mimetype.
MC as a renderer will accept
audio/x-dsf
audio/x-dsd
audio/dsd
and
audio/dsf

Thanks, Bob.  After I set the output to 1x and remembered to link the streamer to this new DLNA server, I got this up and running.

I had the format box checked because that seems to be the default. Are you suggesting that I uncheck it?

[stefano_mbp]

Just tried deselecting Output Format in DSP Studio and it works too.
... but I have to admit that being able to select Output Format after selecting Format: DSF is a bit confusing ... isn't it?

[Mike Rubin]

I am enjoying the upsampling to DSD64, but have run into a consistent repeatable snag: I can play FLAC files with the DSP DLNA server but I can't play mp3's or m4a's now at all.  When I try to play them, the progress bar moves for about five seconds but then just stops.  No music is audible.  Going back to FLAC, I can hear music just fine, so the lossy files aren't crashing out the server.

The vast preponderance of my library is FLAC or DSD, but I also have lots of mp3 and m4a material that only has been available as digital downloads from emusic, Apple, Google or Amazon.  Obviously, I would like to be able to play that as well.

Please let me know if there's something I need to do to configure playback of lossy files with this version of the DLNA server.  (Per Bob's comment, I have deselected "Output Format" in DSP Studio.)

[bob]

I am enjoying the upsampling to DSD64, but have run into a consistent repeatable snag: I can play FLAC files with the DSP DLNA server but I can't play mp3's or m4a's now at all.  When I try to play them, the progress bar moves for about five seconds but then just stops.  No music is audible.  Going back to FLAC, I can hear music just fine, so the lossy files aren't crashing out the server.

The vast preponderance of my library is FLAC or DSD, but I also have lots of mp3 and m4a material that only has been available as digital downloads from emusic, Apple, Google or Amazon.  Obviously, I would like to be able to play that as well.

Please let me know if there's something I need to do to configure playback of lossy files with this version of the DLNA server.  (Per Bob's comment, I have deselected "Output Format" in DSP Studio.)

Just tested and this works fine for me with both mp3 and mp4 lossy or lossless playing to an Id, using Specified output format. It should also work find to do the Specified format as necessary with mp3 and mp4 unselected to play them without upsampling.

[Mike Rubin]

Just tested and this works fine for me with both mp3 and mp4 lossy or lossless playing to an Id, using Specified output format. It should also work find to do the Specified format as necessary with mp3 and mp4 unselected to play them without upsampling.

Hmmm.  I continue to have no luck playing the lossless files to either of two Sonore rendus.

When you say "specified output format," are you referring to the first box in DSP Studio or something else?  If you refer to that, I have confirmed that it makes no difference whether that is checked when I try to play lossy files.

 

[stefano_mbp]

In order to play mp3 files with DSF format selected I need to uncheck mp3 from Advanced/Audio formats to convert but I can play m4a (alac) files.
I don't have any aac file to test
 

[bob]

Hmmm.  I continue to have no luck playing the lossless files to either of two Sonore rendus.

When you say "specified output format," are you referring to the first box in DSP Studio or something else?  If you refer to that, I have confirmed that it makes no difference whether that is checked when I try to play lossy files.

 

Not in DSP studio, in the DLNA server configuration for Audio as in Stefano_mbp's post above.

[Mike Rubin]

Thanks, gentlemen.

I have tried every permutation of "specified output" and "specified output when necessary,"  mp3 and m4a both checked and unchecked (when using "specified output when necessary," as it's grayed out with "specified output"), and "output format" in DSP Studio both checked and unchecked.

The results are 100% consistent:

(1).  I cannot play lossy file types at all unless I exclude them from DSP processing.  More specifically, (a) when I have "specified output" checked or (b) when I have "specified output when necessary" checked with lossy file types not excluded, I cannot get them to play at all.  I get five seconds of silence on the progress bar and then nothing.

(2).  In contrast, when I have "specified output when necessary" checked with lossy file types excluded (unchecked), I can get them to play, but, obviously, not as DSD.

(3).  Lossless does play as DSD when the renderer is bound to the DSP DLNA server, regardless of whether format is "specified output" or "specified output when necessary."

(4).  Checking or unchecking "Output format" in DSP Studio doesn't alter any of the preceding results. 

This all tells me that JRiver actually hasn't figured out how to upsample mp3 and m4a to DSD or there's a bug showing up in my particular usage.

The only devices I have on the network that can handle native DSD are Sonore Signature Rendu SE/Wyred 4 Sound 10th Anniversary DAC and Sonore UltraRendu/Topping D90 combinations.  The results are consistent between both renderer/DAC combinations.  The two Sonores use basically the same board, so perhaps there is something about those components that makes JRiver unable to feed upsampled lossy.

On the other hand, I have just re-verified that, with either renderer/DAC set, mp3 and m4a do play as DSD when upsampled to DSD64 in Audirvana and Roon.  (They also upsample to DSD128 and DSD256 with Audirvana and Roon. I haven't yet tried DSD512, however, out of fear that I will cook my CPU.  DSD256 is a pretty heavy processing load.)

I really would like to get JRiver DSD upsampling to work with my entire library, especially when you get up to DSD256, at which the Topping D90 really shines.  If I can make that happen, I'd abandon both Audirvana and Roon and would convert my NUC to a dedicated Linux JRiver server box rather than run JRiver on my desktop Dell as i do now.

[bob]

Thanks, gentlemen.

I have tried every permutation of "specified output" and "specified output when necessary,"  mp3 and m4a both checked and unchecked (when using "specified output when necessary," as it's grayed out with "specified output"), and "output format" in DSP Studio both checked and unchecked.

The results are 100% consistent:

(1).  I cannot play lossy file types at all unless I exclude them from DSP processing.  More specifically, when I have "specified output" checked or (b) when I have "specified output when necessary" checked with lossy file types not excluded, I cannot get them to play at all.  I get five seconds on the progress bar and then nothing.

(2).  In contrast, when I have "specified output when necessary" checked with lossy file types excluded (unchecked), I can get them to play, but, obviously, not as DSD.

(3).  Lossless does play as DSD when the renderer is bound to the DSP DLNA server, regardless of whether format is "specified output" or "specified output when necessary."

(4).  Checking or unchecking "Output format" in DSP Studio doesn't alter any of the preceding results. 

This all tells me that JRiver actually hasn't figured out how to upsample mp3 and m4a to DSD or there's a bug showing up in my particular usage.

The only devices I have on the network that can handle native DSD are Sonore Signature Rendu SE/Wyred 4 Sound 10th Anniversary DAC and Sonore UltraRendu/Topping D90 combinations.  The results are consistent between both renderer/DAC combinations.  The two Sonores use basically the same board, so perhaps there is something about those components that makes JRiver unable to feed upsampled lossy.

On the other hand, I have just re-verified that, with either renderer/DAC set, mp3 and m4a do play as DSD when upsampled to DSD64 in Audirvana and Roon.  (They also upsample to DSD128 and DSD256 with Audirvana and Roon. I haven't yet tried DSD512, however, out of fear that I will cook my CPU.  DSD256 is a pretty heavy processing load.)

I really would like to get JRiver DSD upsampling to work with my entire library, especially when you get up to DSD256, at which the Topping D90 really shines.  If I can make that happen, I'd abandon both Audirvana and Roon and would convert my NUC to a dedicated Linux JRiver server box rather than run JRiver on my desktop Dell as i do now.

Thanks for the detailed feedback.
Since they play for me upsampled with an Id I can only assume that there is a piece of information that's not in the content directory when playing mp3 upsampled that the Id doesn't need and your devices do.
Will give that a look when possible (not this week).

[Mike Rubin]

Thanks for the detailed feedback.
Since they play for me upsampled with an Id I can only assume that there is a piece of information that's not in the content directory when playing mp3 upsampled that the Id doesn't need and your devices do.
Will give that a look when possible (not this week).

Thank you, Bob. I look forward to hearing what you learn. Please let me know if I can furnish more information.

[Matias]

Is Convolution in DLNA supposed to be gapless by now? Because when I enable Convolution instead of PEQ, songs are no longer gapless with my ultraRendu.

[Mike Rubin]

Was hoping that MC28 would upconvert beyond DSD64 for streaming.  It doesn’t seem to.  Further, it still won’t upconvert mp3’s at all, at least in my system.

[bob]

Was hoping that MC28 would upconvert beyond DSD64 for streaming.  It doesn’t seem to.  Further, it still won’t upconvert mp3’s at all, at least in my system.

I don't seem to have any DSD capable DACs other than software ones like MC. Unfortunately my Oppo 103 which can play DSD via USB doesn't support it over DLNA.
If you you setup wireshark on your computer and do a trace of the interaction with one of your renderers and email that to me, I'll take a look and see if I can see something useful. Start wireshark with the IP address of your renderer as the filter (host ip 1.2.3.4  for example) just before you send the track to the renderer and stop wireshark as soon as it fails which will generate the smallest possible trace.
bob (at) jriver (dot) com

[Mike Rubin]

Thanks, Bob. I will try to figure out how to do that and will forward results.

[bob]

Thanks, Bob. I will try to figure out how to do that and will forward results.

I think we found something so you can wait a bit to do all of that...

[The Computer Audiophile]

I can send audio to a renderer and have it use the MC convolution engine. However, if I use a different control point, with MC as the server, audio doesn't route through the DSP settings.

Can this be changed?

[The Computer Audiophile]

As a follow up, I want to use a new control point for iOS that has built-in support for Tidal and Qobuz, and use MC as the DLNA server for local files. If MC can route audio through its DSP engine, even if the audio is selected for playback with this new control point, it would be a fantastic solution. I can still use MC on my desktop for everything it does, and use it as the DLNA server elsewhere.

[bob]

As a follow up, I want to use a new control point for iOS that has built-in support for Tidal and Qobuz, and use MC as the DLNA server for local files. If MC can route audio through its DSP engine, even if the audio is selected for playback with this new control point, it would be a fantastic solution. I can still use MC on my desktop for everything it does, and use it as the DLNA server elsewhere.

That really should work because the DSP is server based not zone based.
Make sure your controller is using the correct MC server if you have more than one.

[The Computer Audiophile]

That really should work because the DSP is server based not zone based.
Make sure your controller is using the correct MC server if you have more than one.

I only have a single MC zone.

When I play audio the MC interface shows nothing and there is no way to even see what's playing because it's only a DLNA server, not a control point or renderer. It's as if the app is doing nothing, but audio is playing.

[bob]

I only have a single MC zone.

When I play audio the MC interface shows nothing and there is no way to even see what's playing because it's only a DLNA server, not a control point or renderer. It's as if the app is doing nothing, but audio is playing.

The audio information button is zone based so you won't see the information while processing but if you have the audio format set to "specified output format" the DSP will be applied.

[The Computer Audiophile]

The audio information button is zone based so you won't see the information while processing but if you have the audio format set to "specified output format" the DSP will be applied.

Hi Bob, cool I think I have it partially working. The only issue I have now is with DSD content.

In order to get MC to use DSP over DLNA I have to use "Specified Output Format" option. This limits content to either PCM or DSD, but not both. If I have it set to PCM 24-bit, it will convert the DSD content into 24-bit DSD and that is an invalid format.

My control point said playback was at 24 bit / 2.8 MHz rather than 1 bit / 2.8 MHz, or even converting it to a real PCM format for DSP may be reasonable.

[bob]

Hi Bob, cool I think I have it partially working. The only issue I have now is with DSD content.

In order to get MC to use DSP over DLNA I have to use "Specified Output Format" option. This limits content to either PCM or DSD, but not both. If I have it set to PCM 24-bit, it will convert the DSD content into 24-bit DSD and that is an invalid format.

My control point said playback was at 24 bit / 2.8 MHz rather than 1 bit / 2.8 MHz, or even converting it to a real PCM format for DSP may be reasonable.

Well do you want to apply DSP effects to the DSD material?

If so you can only do that by converting the DSD to 24 bit PCM as you have it set now, then in DSP studio change the sample rate for the DSD material to a sample rate supported by your renderer.

If you want to leave the DSD as is if you have a DSD capable renderer, do the "Specified output format when necessary" and under advanced unselect the DSD formats.

[The Computer Audiophile]

Well do you want to apply DSP effects to the DSD material?

If so you can only do that by converting the DSD to 24 bit PCM as you have it set now, then in DSP studio change the sample rate for the DSD material to a sample rate supported by your renderer.

If you want to leave the DSD as is if you have a DSD capable renderer, do the "Specified output format when necessary" and under advanced unselect the DSD formats.

Thanks. Will give it a shot now.

[adyc]

Hi,

I have a question how JRiver sends dsf file to streamers in DLNA mode? Is it in native form or DOP? In other words, how do I choose native or dop streaming for dsf files in DLNA mode?

[dtc]

You can choose which to send. Go to Tools - Options - Media Network - Add or Configure DLNA Servers - Advanced. Select the appropriate DLNA Server. There is an option to Bitstream DSD (Requires DoPE Compliant Renderer). If you check that box it sends DoPE. If you do not check the box it sends native DSD.

[adyc]

You can choose which to send. Go to Tools - Options - Media Network - Add or Configure DLNA Servers - Advanced. Select the appropriate DLNA Server. There is an option to Bitstream DSD (Requires DoPE Compliant Renderer). If you check that box it sends DoPE. If you do not check the box it sends native DSD.

Many thanks. This is exactly what I need. Another question is that whether JRiver can send DSD 512 dsf files natively in DLNA mode.

[Awesome Donkey]

I would imagine DSD512 over DLNA would only work with native DSD, not DoP, as it'd require the device to support a PCM sample rate of 1411.2 kHz and I'm not aware of any that do so native DSD is likely the only option there. Probably something you'd have to try.