HQPlayer's Network Audio Adapter

[Miska]

4 hours ago, Jjazz said:

Thank you very much Miska for your help, I have done this with the overlay of hifiberry-dac which is the one that works with ian canada fifopi, however it does not work for NAA or Hqplayer embedded. The dac appears but music cannot be played. Any idea what the problem might be? I'm happy with Volumio but I'm sure I'd be much happier with room and hqplayer.

 

Check that you have the video output disabled, or at least it's audio side turned off.

Also check that you have the force_eeprom_read=0 just in case (see my example above).

 

If these don't help, then I don't know what is the problem. I have the HifiBerry boards working here, but I don't have anything else to test with.

 

[StreamFidelity]

What are the differences between NAA and UPnP and Roon RAAT?  What are the similarities?

[Miska]

1 minute ago, StreamFidelity said:

What are the differences between NAA and UPnP and Roon RAAT?  What are the similarities?

 

NAA, RAAT, AES67/RAVENNA and AVB are endpoint protocols. They reflect audio endpoint on the network and handle low level audio.

 

UPnP AV is arrangement of three entities. Control Point which controls what is played and where. Media Server that is holding the media content. And Renderer that actually plays the content from the Media Server. UPnP Renderer is the player, and then it has some kind of endpoint behind it that is not visible in the UPnP architecture in itself. As you know, HQPlayer Embedded can operate as UPnP Renderer. UPnP Renderer's task is to decode and play the content and provide position and other playback indication feedback to the Control Point. And limited amount of metadata. But main source of metadata is Media Server.

 

NAA and RAAT are most similar ones and designed for similar purposes, although with number of implementation differences.

 

AES67/RAVENNA and AVB are designed for different use cases than NAA and RAAT. These are mostly for studio and live sound. RAVENNA is extension to AES67 adding support for higher PCM rates (up to 384k) and DSD (up to DSD256). AES67 and AVB are limited to max 192k PCM but can do a lot of channels and are not limited like USB Audio Class in this respect. These don't work reliably over WiFi or non-dedicated networks since they have practically no buffers for minimal input - output delay. Which is important if you are playing instrument and need to listen to what you are playing through the entire ADC - DAC chain.

 

I have RAVENNA working properly over non-dedicated network, because I use the optional special unicast connection. But this needs some care and doesn't still go over WiFi.

[StreamFidelity]

Thank you for the detailed reply. The question I have is whether UPnP has disadvantages compared to NAA and RAAT?

 

In the Euphony thread there are complaints that via UPnP a song is sampled in a list and then stops. A familiar one uses an eRED-DOCK network audio interface (UPnP AV 2.0) in the DAC and has the same problem. Is UPnP more unstable than NAA?

 

4 hours ago, Miska said:

NAA, RAAT, AES67/RAVENNA and AVB are endpoint protocols. They reflect audio endpoint on the network and handle low level audio.

If NAA and RAAT process low level audio and UPnP possibly higher level, can UPnP have a worse SQ?

[stefano_mbp]

1 hour ago, StreamFidelity said:

In the Euphony thread there are complaints that via UPnP a song is sampled in a list and then stops

I used Euphony for a brief time and UPNP was not working reliably, only after a remote access session Zeliko was able to let it work, it was the V3 version.

It worked fine at first using RoPieeeXL/UPNP renderer but couldn’t work with SOtM sMS200/UPNP.

My opinion is that it is a matter of UPNP implementation, if it was done in a hasty way it is very likely that it will not work well and I believe this is the case.

[Miska]

1 hour ago, StreamFidelity said:

Thank you for the detailed reply. The question I have is whether UPnP has disadvantages compared to NAA and RAAT?

 

Well, it is apples to oranges comparison. UPnP as such doesn't guarantee you anything. You could have transcoding (as specified by the spec) going on the media server and the renderer (player) could be doing anything. If the renderer doesn't support FLAC for example, the server could be doing transcoding to MP3 for the playback. Or doing rate and format conversion of the renderer doesn't support a particular format.

 

1 hour ago, StreamFidelity said:

In the Euphony thread there are complaints that via UPnP a song is sampled in a list and then stops. A familiar one uses an eRED-DOCK network audio interface (UPnP AV 2.0) in the DAC and has the same problem. Is UPnP more unstable than NAA?

 

UPnP is among to most non-standard standards. It is relatively flexible and complex, but doesn't go much into details of media formats and such. Then DLNA was created slightly later to define some common subset of formats that all compliant devices would support, so that they can actually interoperate in first place. Because you could very well have Media Server by company X that can only output WAV and WMA. And Media Renderer by company Y that would only support AIFF and AAC. These two wouldn't be able to play anything together. So DLNA goes on to define that for example 128 kbps CBR MP3 support is mandatory (WMA is also included). And that server must be able to transcode any other format it supports to that MP3. None of the standards say a word about DSD.

 

I created NAA because I needed something that fits HQPlayer needs. UPnP is not suitable for HQPlayer output, because it doesn't do what HQPlayer needs. RAAT was created for similar reasons.

 

1 hour ago, StreamFidelity said:

If NAA and RAAT process low level audio and UPnP possibly higher level, can UPnP have a worse SQ?

 

Yes, because it is very hard to know for sure what is going on with UPnP unless manufacturer of both ends tells you what their implementation does.

 

[The Computer Audiophile]

I would love to see more high end manufacturers adopt a truly high end solution like Ravenna or even NAA. 

[robi20064]

1 hour ago, The Computer Audiophile said:

I would love to see more high end manufacturers adopt a truly high end solution like Ravenna or even NAA. 

 

I couldn't agree more! However it seems we are just a minority in the grand scheme of things, and even tho implementing NAA on the endpoint would be relatively simple (according to Miska) there are limited resources everywhere for firmware development and not enough demand to justify the need. Still, one can hope.

[AudioDoctor]

32 minutes ago, robi20064 said:

 

I couldn't agree more! However it seems we are just a minority in the grand scheme of things, and even tho implementing NAA on the endpoint would be relatively simple (according to Miska) there are limited resources everywhere for firmware development and not enough demand to justify the need. Still, one can hope.

 

My Pacific DAC is currently running NAA software versus the Roon Bridge that came as standard in the ethernet input.

[robi20064]

That is some nice support from the Polish folks (and a nice DAC in general)! I would be more interested in one of the top streamer brands (Lumin, Auralic, etc.) adding NAA functionality, so I don't need to keep yet another streamer dedicated for only this purpose. Hell, even Gustard is providing NAA to their R26 DAC these days.

 

I recall Xuanqian of Auralic stating at some point in the past that they won't consider adding it unless the protocol become open source. I believe that had to do more about HQPlayer being a competitor to their Sirius upsampler rather than being afraid of any risk related to a compiled NAA library.

[Miska]

1 hour ago, robi20064 said:

I recall Xuanqian of Auralic stating at some point in the past that they won't consider adding it unless the protocol become open source. I believe that had to do more about HQPlayer being a competitor to their Sirius upsampler rather than being afraid of any risk related to a compiled NAA library.

 

I guess they have Roon stuff though, and AFAIK that is not open source either...

 

[robi20064]

Sounded like an excuse pretty much. Spotify Connect, Tidal Connect, Roon Ready and similar SDKs are barely open source these days. Or maybe they trust them more as there are companies behind those tech and you are just you, a pretty powerful one man army.

[Miska]

36 minutes ago, robi20064 said:

Or maybe they trust them more as there are companies behind those tech and you are just you, a pretty powerful one man army.

 

I would say it is not that, but some other reason. My company has existed since 2008 and the software got started in 1998, so quite a bit older than many of those others.

 

But what ever the reason, it is their choice. Some other companies choose differently. I just few days ago updated the NAA support list on my web page. It is not long, but it is certainly growing.

 

[AudioDoctor]

I think it can't be difficult to take a Raspberry Pi compute module, create your own ethernet input with the same effort some manufacturers put into their usb inputs, a user accessible SD card for the software of the users choice, and you have a DAC with a high performance network input that's not locked down to any specific software.

 

It can be done, it's just a matter of willingness.

 

edit 1:

essentially this, but integrated into the DAC itself, not an external box.

https://www.kitsunehifi.com/product/holoaudio-red-ddc-network-streamer/

 

edit 2:

Or this, with something like a JCAT USB XE card in the one PCI slot for those that like to tinker.

https://www.raspberrypi.com/products/compute-module-4-io-board/

[Miska]

9 hours ago, AudioDoctor said:

I think it can't be difficult to take a Raspberry Pi compute module, create your own ethernet input with the same effort some manufacturers put into their usb inputs, a user accessible SD card for the software of the users choice, and you have a DAC with a high performance network input that's not locked down to any specific software.

 

It is not that hard. If I would have time to do hardware, I could make discrete multichannel DSD DAC supporting up to 8 channel DSD512 through NAA. And I know exactly how.

 

[Phil261575]

Hi Miska, my actual and stable setup is ROCK on NUC and HQPE OS on a dedicated i9 12900T server. Everything is stable and simple.

 

However, the 8 Eproc of the i9 are of very limited use. If I remember well, it is because AWX-512 instructions are not available or hidden on efficiency cores. Are you working on an alternative that would make use of those limited but available Eproc?

[Miska]

2 hours ago, Phil261575 said:

Hi Miska, my actual and stable setup is ROCK on NUC and HQPE OS on a dedicated i9 12900T server. Everything is stable and simple.

 

However, the 8 Eproc of the i9 are of very limited use. If I remember well, it is because AWX-512 instructions are not available or hidden on efficiency cores. Are you working on an alternative that would make use of those limited but available Eproc?

 

Unless necessary, those are left to be used by the OS itself and other things, such as Roon.

[Phil261575]

Does the SSE version of HQPE change something with the use E core?

[Miska]

9 hours ago, Phil261575 said:

Does the SSE version of HQPE change something with the use E core?

 

No, all versions use the cores same way. Efficiency cores are used when determined that they would be useful. For other cases they are left for running other tasks, so that for example OS itself and something like Roon has place where to run without disturbing HQPlayer.

 

[madman73]

Hi @Miska what are the differences between linux version 4.1.1 and the latest one? In the last two years I haven't been able to follow all the developments and updates, these days I'm putting my system back into operation, and I can't use the most recent versions.

My NAA is based on an old AMD E-350 mini ITX motherboard powered by a custom made LPSU. I've always used it with gentooplayer (for a full driver support) and it works perfectly with 4.1.1, but new versions crash.

I use HQPlayer Embedded on I7-3770K and I don't play DSD, just PCM @768. For this I don't need very powerful and particularly recent hardware.

 

 

 

 

[Miska]

20 minutes ago, madman73 said:

Hi @Miska what are the differences between linux version 4.1.1 and the latest one? In the last two years I haven't been able to follow all the developments and updates, these days I'm putting my system back into operation, and I can't use the most recent versions.

 

There are some improvements especially on the input functionality. And also fixes for audio driver behaviour with latest kernel versions.

 

20 minutes ago, madman73 said:

My NAA is based on an old AMD E-350 mini ITX motherboard powered by a custom made LPSU. I've always used it with gentooplayer (for a full driver support) and it works perfectly with 4.1.1, but new versions crash.

 

Have you tried booting up the latest NAA OS image? But if it doesn't have support for SSE4.2 instructions, then that could be the reason.

 

[madman73]

4 minutes ago, Miska said:

 

There are some improvements especially on the input functionality. And also fixes for audio driver behaviour with latest kernel versions.

 

 

Have you tried booting up the latest NAA OS image? But if it doesn't have support for SSE4.2 instructions, then that could be the reason.

 

No, as i remember you didn't support my motherboard, so i had to use gentooplayer. if you included the drivers i will try.

Buy why did you removed support for older model?

[Miska]

5 minutes ago, madman73 said:

No, as i remember you didn't support my motherboard, so i had to use gentooplayer. if you included the drivers i will try.

 

Please check if the CPU has SSE4.2 support. If it doesn't the image is not going to work...

 

If you have some Linux running, you can use something like this:

cat /proc/cpuinfo | grep sse4_2

 

Or 

lscpu | grep sse4_2

 

If there's no output, then there's no support for SSE4.2...

 

[madman73]

5 minutes ago, Miska said:

 

Please check if the CPU has SSE4.2 support. If it doesn't the image is not going to work...

 

If you have some Linux running, you can use something like this:

cat /proc/cpuinfo | grep sse4_2

 

Or 

lscpu | grep sse4_2

 

If there's no output, then there's no support for SSE4.2...

 

Now I can't try, but from what I read here I think not

https://www.cpu-world.com/CPUs/Bobcat/AMD-E Series E-350 - EME350GBB22GT.html

[Miska]

23 minutes ago, madman73 said:

Now I can't try, but from what I read here I think not

https://www.cpu-world.com/CPUs/Bobcat/AMD-E Series E-350 - EME350GBB22GT.html

 

Yeah, back then AMD was a bit messy about instruction set support. I have one A-series mini-laptop that has SSE4.2 and AVX. While I think Phenom and E-series didn't have such...