Print

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

[ThierryNK]

Hi

As I have 2 JRiver licenses, one on Mac, one on Windows, I checked that what is described below happens on both Windows and OSX.

I only use JRiver as a UPNP server. I have a network player that accepts DFF, DSF and DoP formats for DSD.

DSD bitstream is enabled on JRiver.

With SACD ISO and DFF files everything works fine.

With DSF files, it works also, except a big "noise" at the end of a majority of tracks.

When disabling DSD bitstream (which cancels the Wave DoP process by JRiver), JRiver streams directly DSF files with no problems, and no "big noise" at the end of the tracks.

Maybe, the "tags" that are included in the DSF files are not "well" treated (I mean, NOT treated as audio) during the JRiver DoP conversion, as included tags are the only differences between DSF and DFF files.

Best and kind regards

[Matt]

To make sure I understand, what option are you toggling for DoP bitstreaming with DLNA?  The one in Options > Media Network > Add or configure servers with "DoPE" in the name?

If you play the files on a computer using MC, do you hear noises at the end?

[ThierryNK]

Hi Matt  

The option I use is:

Media Network > add or configure DLNA Servers > Advanced > Bit Stream DSD (requires DoPE Compliant Renderer)

If I play the DSF file on the computer, there is no noise, but in that case the file is converted to PCM, right? No DSD or DoP involved I think.

For information, this "DSF noise" appears on several albums, either directly bought online, or converted from DFF, or coming from SACD ISO.

The Network player is a Lumin.

No noise for this DSF files, if NO Bit Stream DSD option (but in that case, JRiver cannot stream from SACD ISO)

No noise with Bit Stream DSD option ON, with DFF files and SACD ISO

The only difference between DFF or ISO and DSF being the tags is why I suspect the DoP conversion of DSF files.

I already saw where the temporary JRiver wave DoP files are located, I will save them before they are cleared, and try to play them "directly". Maybe, we'll learn something.

But later, here it is 4 am  

Thanks for your help.

Kind regards

[Matt]

I already saw where the temporary JRiver wave DoP files are located, I will save them before they are cleared, and try to play them "directly". Maybe, we'll learn something.

Good idea.  Let us know what you find.

We'll test as well.

[john925]

Hi Matt  

I already saw where the temporary JRiver wave DoP files are located, I will save them before they are cleared, and try to play them "directly". Maybe, we'll learn something.

Interesting. Would you please share about where the files are located? Thanks!

[ThierryNK]

In the "temp" folder of JRiver. You can find the location in:
Options > Files Location > Temporary Files

[john925]

In the "temp" folder of JRiver. You can find the location in:
Options > Files Location > Temporary Files

Many thanks!!

[ThierryNK]

JRiver only keeps 2 tracks while playing. The current track and the next one.

You have to copy them before they are deleted. But you have all the length of the current track to copy, first step: the 2 first tracks, then, the "next next" track when you forward your playlist.

Do not expect to copy "all tracks at once 

[ThierryNK]

Hi

Here are the results of my investigations:

- I made an album with the DoP wave files built by JRiver while playing a DSF album:
Result: no noise when streaming directly those DoP wave files
Conclusion: the DoP process it self does not seem to be involved in the noise.

- As I am using JRemote, I tried using JRiver directly as control: the noise remains for DSF files.
Conclusion: JRemote not involved in the noise.

- I cannot say if it is at the end of the current track or at the beginning of "next" track that the noise happens. Maybe, it is more likely at the beginning of next track (not sure, I will try to monitor this more precisely).

- Question: From JRiver developers perspective, what "more" could be "sent" to the network player when making DoP of DSF files, that is not sent when making the same process for DFF files?
I also remind that direct streaming of DSF files (with no DoP process) works perfectly.

Just let me know what kind of monitoring I could do (network, UPNP,...) to find out what is happening.  I have "tools" to do this.

DSF files are so more convenient to have tags included inside the files. JRiver can handle the tags for DFF files, but they are lost if you simply move the files in a different folder.

I wish to keep "DSD Bit Stream" ON, as it is required for streaming SACD ISO.

If needed, I can share a DSF album with JRiver team. Just tell me.

Best and kind regards

[Matt]

- Question: From JRiver developers perspective, what "more" could be "sent" to the network player when making DoP of DSF files, that is not sent when making the same process for DFF files?

DoPE is just a 44-byte WAV header followed by the raw DSD data.

Is it possible the DLNA player is playing the next track's 44-byte header as DSD data?

We use another copy of MC as our reference DLNA renderer when testing.  You could push to another MC with DoPE enabled and see if that MC plays it correctly to a DSD-capable DAC like a Mytek, etc.

[ThierryNK]

Thanks for your effort Matt, I appreciate.

But, as I said, the temporary wave DoPE files that I have copied from the the JRiver temp directory are streamed smoothly without any "noise".
The noise appears only when the DSF album is played.

To make it clearer:
DSF album > DoPE > NOISE (when ending or starting the next track)
Wave DoP album made from the previous process > GOOD
Same album in DFF > DoPE > GOOD
Same album in DSF, NO DoPE > GOOD
Same album in DFF, NO DoPE > GOOD
Same album, DoP files made by me > GOOD

Something is happening whith "real time" DoPE on DSF files. Where? Renderer or JRiver? I do not know!! I do not have a single clue about what could be happening.

My renderer is OK with the Wave DoPE files made from DSF when they are already created, but not during the real time process of making DoP files from DSF files and stream the files.

A question Matt: how do you test DLNA stream with another MC renderer that can not decode DSD? You can send DoPE files to the other MC, but then, what does the receiving MC do with DoP files? The second MC have to be connected to a DSD DAC by USB? That is the way to do it?

As I will start the review of Ayon Stratos DSD DAC next week for Audiophile Magazine, I will try to make this test.

If you need one of the DSF albums that produces this kind of behavior, just tell me.

Best and kind regards

[Matt]

Something is happening whith "real time" DoPE on DSF files. Where? Renderer or JRiver? I do not know!! I do not have a single clue about what could be happening.

The server just serves those files you're copying.  So I don't know either.

You could visit the playback URL in a browser and pull the files that way.  Then, you're actually reading from the server.  Look in the 'Media Network' view to see the URLs.

A question Matt: how do you test DLNA stream with another MC renderer that can not decode DSD? You can send DoPE files to the other MC, but then, what does the receiving MC do with DoP files? The second MC have to be connected to a DSD DAC by USB? That is the way to do it?

Yes, exactly.

If you need one of the DSF albums that produces this kind of behavior, just tell me.

The way you describe it makes me think it's only going to happen to some devices.

Let me test streaming to MC in the next couple days and I'll follow-up with my results.


Sorry to not have a more absolute answer on this one.

[ThierryNK]

As the noise appears at the track changes, I have disabled the SetNext Support. This breaks the gapless (i have just checked) and I know, for sure that Lumin have implemented it as I have directly suggested them to do it after some discussions here with JRiver team, and they have announced it for a firmware release some months ago.

And, TADAAAAA!!!

No more noise.

Does this give you a hint of what could be happening?
I recall (sorry, it is just for clarity) that this happens only with the "real time" DoPe and not with the wave files made by JRiver, or with any other DSD file, DoPE or no DoPE.

[Matt]

It's hard for me to explain why using SetNext would cause noise to be heard unless the renderer is doing something strange in this case.

Internally for us, SetNext just tells the device about the next file ahead of time.  The mechanism to get the file data from our server is exactly the same.

And since you've looked at the files we're serving and think they're correct, I can't think what we could be doing to cause it.

Is there any chance Lumin might take a look?

[ThierryNK]

I have used UPNP Inspector to try to see what was happening...

- One DSF track
- The Wave DoPE file that has been generated by JRiver for the same DSF track.


The UPNP announced file size is not the same for the two cases (size= on the right of the screen copies):

Real time DSF converted to DoPE:269,340,716


DSF




Saved DoPE WAVE from DSF; 269,328,974

WAVE


They should be strictly identical, no?. The real time Wave-DSF is bigger than the saved one from the temp directory.

Could there be some "modification" between the file that is generated in the temp directory and the one who is served by the JRiver http server?

Best and kind regards

[ThierryNK]

Everything runs fine in version 18/212
But the issue remains on Mac 18 and 19. I will open a thread for Mac 19
Kind regards