HQ Player

[Miska]

What does HQPlayer do when fed 44.1kHz audio, with 96 kHZ convolution, and 192kHz output?

 

Convolution is performed at 44.1k before upsampling. Convolution is generally performed at source rate, except for DSD-to-PCM case where it is performed after conversion to PCM at 1/16th of the DSD rate (significant reduction in CPU load).

[cat6man]

If you don't specify whether you want PCM or SDM output, you'll get PCM output. So "need" depends on the particular case...

 

Then there's for example option to make volume setting fixed, so it'll ignore whatever the Control Point tells about volume.

 

sorry to bug you, but what about the comment to "configure name for the renderer"? i don't even see the word render in the readme file.

 

under sub-element 'network', i see

 

Attribute: address

Name of the remote network audio adapter to be used.

 

Attribute: device

Remote device name.

 

is 'address' the ip address of the micro-rendu?

what is 'device' then?

[Miska]

sorry to bug you, but what about the comment to "configure name for the renderer"? i don't even see the word render in the readme file.

 

It doesn't talk about renderer in the readme, because hqplayerd itself has nothing to do with UPnP spec. It can be used for local playback or for some other streaming system than UPnP. It is all generic.

 

Take a look at the example configuration file shipped with the daemon for a minimal configuration. The naming part is documented in the readme:

1.8 Element "title"
Human readable title for this player instance.

Attribute: value
Title string.

 

And also in the example configuration:

<title value="Test"/>

 

under sub-element 'network', i see

 

Attribute: address

Name of the remote network audio adapter to be used.

 

Attribute: device

Remote device name.

 

is 'address' the ip address of the micro-rendu?

what is 'device' then?

 

No, since typically NAA is using DHCP (like microRendu always) there is no static IP that could be configured. NAA is always addressed by it's name.

[Theobetley]

Trying to update from level 3.12 to latest. So I did the following:

 

 

1)Uninstalled 3.12

 

2)Went to Signalyst site downloaded 3.13 something (I downloaded the 64 bit version, is that correct for Windows 10...don't remember)

 

-- tried to open HQP but all I get is the Uninstall link

-- figured I need to agree to License

 

3)Don't see where I agree to Terms Conditions. Where do I do that?

 

Ok I d/l ed the 32 bit version now all is well.

[bogi]

Miska, one question related to streaming. I know almost nothing about streaming, because I didn't use it. I am always playing local files from my notebook (via RAM disk to get rid of HDD noise).

 

But I got the idea to try something like streaming to send what is actually played in Foobar2000 to HQPlayer Desktop. Only locally within my notebook. And interesting is, that I succeeded within some minutes.

 

I installed foo_upnp and configured it to expose what is playing as http://192.168.8.100:8000/content/psc.wav

I set output audio device in foobar2000 to 'Null Output'

 

I created .m3u8 playlist file with single line:

http://192.168.8.100:8000/content/psc.wav

I dropped the playlist in HQPlayer Desktop and pressed 'play' in foobar2000.

After delay about 5 seconds playback started.

Now everything I am 'playing' in foobar2000 plays through HQPlayer Desktop. I'm doing also PCM to DSD conversion in HQPlayer. In foobar2000 GUI all functions as if it would be played in foobar2000, only that few seconds delay exists (no problem).

Interesting is, that foo_upnp allows to perform DSP on PCM content before exposing data on the media server (or how it is called).

 

That's all fine, only sound quality is not quite on the same level as when I am playing the same content in HQPlayer directly. And that's the thing I want to ask you for opinion. Do you thing any technical reasons exist why playing directly a local 44.1k/16b FLAC file should sound better that playing it through http://...some/thing.waw on a single computer?

[R1200CL]

Not sure if you are clear

Let me explain better. I use HQPlayer with Roon and MicroRendu. So I was trying to use Alchemy as control point for the HQPlayer with setting as NAA.

I did not in this case use the latest beta HQPlayer.

 

It was not possible to setup Alchemy in my case. If you are saying the Alchemy can't control HQPlayer when combined with Roon, then your explaination make sence.

 

I have never used HQPlayer as a stand alone media player, and i probably never will. My only reason to use HQPlayer is to be able to upsample PCM Redbook and Tidal, and test some filters.

 

With set up, I mean "Filter Maps", it accepted the input settings, but it was not possible to create output settings. Also I did not find the setting for NAA vs Asio etc.

 

If you like we can create a new thread about setting up Alchemy. Not sure if usefull. This SW seems to be in its early development.

 

I found a minor error in that SW. I've notified support. Those filter settings is not so easy as they seems. You have to both select one of the three PCD/DSD/Auto and in addition chose between Auto Rate and Max rate. (Which is explained what is the difference in last page, last sentence in user manual)

 

If you have done any selections in the right most box (Dither selections), you can not change your filters.

 

I was thinking of using this SW to more easily test the different filters in "PCM*" selected input mode, it's to time consuming adding all the options. Then HQPlayer is better.

 

Maybe a defalt option like "Add All" in output mode could be an option. Or at least "Add All" pr. Dither. (After selected input).

 

 

Also, this shows us that if Roon would like to, they could add HQPlayer control / filter selections into their app.

They could even auto select filters based on tags

[cat6man]

It doesn't talk about renderer in the readme, because hqplayerd itself has nothing to do with UPnP spec. It can be used for local playback or for some other streaming system than UPnP. It is all generic.

 

Take a look at the example configuration file shipped with the daemon for a minimal configuration. The naming part is documented in the readme:

 

on your 'configuring hpqplayer....' web page, it says the readme has an option 'to configure name for the renderer'........that is what confuses me.

 

 

Configuring HQPlayer and Rygel

 

HQPlayer Embedded will utilize most of the configuration settings done by HQPlayer Desktop. There are some additional options specified in /usr/share/doc/hqplayerd/readme.txt.gz such as option to switch to SDM output mode and [b]to configure name for the renderer.[/b] These need to be manually changed to the configuration file.

[Hammer]

Hi, my current setup is a 6700k PC and it up samples fine to DSD256 (with convolution using a REW generated correction file) which is the max for my Vega DAC. The one area I would like faster response time is when switching between sample rates. When I play say a red book file and then a DSD64 fine, a blue bar comes on a it takes a few seconds before the new song plays. Will adding a CUDA card such as the new GTX 1070 speed that up? Thanks, Hammer.

[bogi]

Hi, my current setup is a 6700k PC and it up samples fine to DSD256 (with convolution using a REW generated correction file) which is the max for my Vega DAC. The one area I would like faster response time is when switching between sample rates. When I play say a red book file and then a DSD64 fine, a blue bar comes on a it takes a few seconds before the new song plays. Will adding a CUDA card such as the new GTX 1070 speed that up? Thanks, Hammer.

 

I mean HQPlayer could store and reuse values computed for different sample rates in the past. I cannot be sure but I expect HQPlayer computes the same thing again and again... Miska, could it be improved in some of the next HQPlayer updates?

[Outlaw]

Does the latest beta version support cuda offload ?

[dean70]

Not sure if you are clear

Let me explain better. I use HQPlayer with Roon and MicroRendu. So I was trying to use Alchemy as control point for the HQPlayer with setting as NAA.

I did not in this case use the latest beta HQPlayer.

 

It was not possible to setup Alchemy in my case. If you are saying the Alchemy can't control HQPlayer when combined with Roon, then your explaination make sence.

 

I have never used HQPlayer as a stand alone media player, and i probably never will. My only reason to use HQPlayer is to be able to upsample PCM Redbook and Tidal, and test some filters.

 

With set up, I mean "Filter Maps", it accepted the input settings, but it was not possible to create output settings. Also I did not find the setting for NAA vs Asio etc.

 

If you like we can create a new thread about setting up Alchemy. Not sure if usefull. This SW seems to be in its early development.

 

There is a thread here for Alchemy

[Miska]

Does the latest beta version support cuda offload ?

 

Only the Linux build which is built with the older CUDA 7.5 SDK. It doesn't work anymore on Windows and OS X, and the CUDA 8 SDK release is still pending (RC stage). I'm waiting until the the version 8 comes out before releasing HQPlayer 3.14 final.

 

For the same reason, I'm stuck on Ubuntu 14.04 while the final will be built on and for 16.04.

[Miska]

I mean HQPlayer could store and reuse values computed for different sample rates in the past. I cannot be sure but I expect HQPlayer computes the same thing again and again... Miska, could it be improved in some of the next HQPlayer updates?

 

No changes in that area in the plans. Moore's law will fix that too...

[Miska]

Hi, my current setup is a 6700k PC and it up samples fine to DSD256 (with convolution using a REW generated correction file) which is the max for my Vega DAC. The one area I would like faster response time is when switching between sample rates. When I play say a red book file and then a DSD64 fine, a blue bar comes on a it takes a few seconds before the new song plays. Will adding a CUDA card such as the new GTX 1070 speed that up? Thanks, Hammer.

 

CUDA can help convolution engine on both initialization and runtime. But depends a lot on the case if it is noticeable difference.

[Miska]

on your 'configuring hpqplayer....' web page, it says the readme has an option 'to configure name for the renderer'........that is what confuses me.

 

 

Configuring HQPlayer and Rygel

 

HQPlayer Embedded will utilize most of the configuration settings done by HQPlayer Desktop. There are some additional options specified in /usr/share/doc/hqplayerd/readme.txt.gz such as option to switch to SDM output mode and [b]to configure name for the renderer.[/b] These need to be manually changed to the configuration file.

 

The web page is quick dump of thoughts and talks about how to use HQPlayer Embedded as UPnP Renderer. While the readme.txt is focused purely on HQPlayer Embedded itself and on purpose leaves out everything that goes beyond HQPlayer.

 

That's why it may look confusing because scope is different...

[Miska]

Now everything I am 'playing' in foobar2000 plays through HQPlayer Desktop. I'm doing also PCM to DSD conversion in HQPlayer. In foobar2000 GUI all functions as if it would be played in foobar2000, only that few seconds delay exists (no problem).

Interesting is, that foo_upnp allows to perform DSP on PCM content before exposing data on the media server (or how it is called).

 

Now the question is what happens if the source format on foobar playlist changes for example from 44.1/16 to 96/24. When that happens, the correct thing would be to add new stream to HQPlayer playlist and HQPlayer will then auto-proceed to that one once the current ends (connection closed). If foobar just drops the connection and expects re-connect to the same stream on format change, you can help the behavior by clicking to "repeat one" button down from the toolbar. That will play the current item in a loop.

 

That's all fine, only sound quality is not quite on the same level as when I am playing the same content in HQPlayer directly. And that's the thing I want to ask you for opinion. Do you thing any technical reasons exist why playing directly a local 44.1k/16b FLAC file should sound better that playing it through http://...some/thing.waw on a single computer?

 

One possibility is that foobar is performing transcoding to something and that would affect the sound quality too. UPnP Media Servers are expected to perform transcoding as necessary. MinimServer doesn't do it by default though.

 

Another possibility is due to how HQPlayer deals with the files vs how foobar does it.

 

Check what HQPlayer reports as "Format" in the main window - it'll tell at least something.

[vladimirb0b]

Convolution is performed at 44.1k before upsampling. Convolution is generally performed at source rate, except for DSD-to-PCM case where it is performed after conversion to PCM at 1/16th of the DSD rate (significant reduction in CPU load).

 

 

Is it OK to use 96kHz convolution files even though it convolves at 44.1kHz? Thanks!!

[vladimirb0b]

Also, can HQPlayer process UPnP input? I would like to try it running Plex.

[bibo01]

Also, can HQPlayer process UPnP input? I would like to try it running Plex.

 

Only HQPlayer Embedded (Linux only) can work with UPnP. You need to buy a Linux license and then upgrade the license to Embedded requesting at support.

 

Is it OK to use 96kHz convolution files even though it convolves at 44.1kHz? Thanks!!

 

Yes, normally Miska even recommends to use a 192kHz filter.

[vladimirb0b]

Yes, normally Miska even recommends to use a 192kHz filter.

 

Great.

 

A suggestion: I would like to see a way to upsample, apply filter, and then upsample again. This could be done in one step within the program, or by adding an "input" that would allow a virtual cable to be the output of player instance 1 cabled to the input of player instance 2. Adding an input would also give the ability to play various sources through HQPlayer.

 

Maybe I am missing a way to do this?

[bogi]

Also, can HQPlayer process UPnP input? I would like to try it running Plex.

 

As I wrote previously, I'm not experienced in streaming, but I remember the same what Bibo answered: no upnp with HQPlayer Desktop. But HQPlayer Desktop can access http streams for example from radio stations. Create .m3u8 playlist file containing single line

http://radio.cesnet.cz:8000/cro-d-dur.flac

and drop it to HQPlayer Desktop. Czech Radio will start to play after some seconds.

I knew about that and therefore I was searching, if Foobar2000 is able to broadcast currently played content in similar way as internet radio stations do. So that's the main principle what I tried and described yesterday.

 

Despite of the foobar component is called foo_upnp (see foobar2000: Components Repository - UPnP/DLNA Renderer, Server, Control Point), it does 3 things: Media Server, Media Renderer and Control Point. I am running Media Server part and I have Media Renderer part stopped. It is interesting that it is really possible to use HQPlayer sound engine from Foobar2000 using simple config and noone pointed to it before. But I have to find limitations. If I will find it well usable, I will describe it in a separate topic.

[Miska]

A suggestion: I would like to see a way to upsample, apply filter, and then upsample again.

 

What would be the point of doing so? I can only think of negative effects...

[Keith_W]

Miska, I too would like to see an input to HQPlayer to allow a virtual cable to be inserted. This is the reason why: in other posts, I have said that I am using HQP to generate my crossovers. With a virtual cable, I will be able to send measurement sweeps from REW or Acourate through HQP to check if the crossovers are working as designed.

[vladimirb0b]

What would be the point of doing so? I can only think of negative effects...

 

I was stuck in my old thinking that the Convolver needs the sample rate of the file to be the sample rate of the filter... So all that is not necessary. But I would still want to upsample before applying the convolution, not after. Would it make any difference? I'm not sure.

 

A side note: My DRC can generate 96 kHz filters at 24 bit, not sure about 32 bit. Would it be a good idea to upsample to 192kHz/32 bit? The only software I know that does 192 is Dirac Live, and I can't afford it.

[vladimirb0b]

As I wrote previously, I'm not experienced in streaming, but I remember the same what Bibo answered: no upnp with HQPlayer Desktop. But HQPlayer Desktop can access http streams for example from radio stations.

 

I saw you use foobar2000 a few posts up to create a kind of wav stream. My question is: is there a way to set a HQPlayer output that creates a wav stream that can then be loaded by another instance of HQPlayer? Thanks!