HQ Player - a few questions

[tahsu]

Hello all!

I have recently utilised HQ Player trial and I have decided to purchase the full version (4). I have read the manual but I do have a few further questions…

I have set up Team Viewer (as suggested in my original thread) so I can navigate to my Roon Core which is running on a headless Windows 10 NUC, which is also running HQ Player, I then feed various endpoints around the house. Practically speaking though I will only want to run HQ player on two end points, in my office and lounge. I know I can’t run them both together and I understand there is no profile functionality in HQ Player which I knew before purchasing. However I would like to understand a bit better the settings for each endpoint so I can move between the two fairly quickly.

Just to be clear there is no multichannel, just stereo.

Office endpoint

Windows 10 NUC 7I7BNH (dual core) – Ethernet cable – Netgear switch – Patch-box – Ethernet cable – Netgear switch – Allo Digione streamer – coaxial cable into Arcam irDAC-II – Adam F5 active speakers.

Plan – to upscale all content to PCM, the limit of the connection using spdif

Office endpoint questions

• -3dBFS – I understood this was a good place to start, however I have noticed the dial turn red, so I assume I need to turn down. I have gone to -4 then -5, and I watching if it has any further effect.
• Buffer time and DAC bits – I assume I should leave on default, any other thoughts, in what scenario should I adjust?
• Default Output Mode – I intend to switch to PCM for this endpoint as I cannot get DSD to work outside of a direct Windows environment, or it better to leave on source?
• Filter 1x and Filter Nx – what is the difference, I could not see an explanation in the manual, how should I use the two?
• Poly-sinc-xtr – mp filters – I know the filters are subjective, however I mostly listen to guitar based or electronic music so I felt this was a good choice, any other perspectives?
• Other – NS9 – I am aiming to upsample to the limit of spdif connection, so 192kHz, so this seemed like a good setting, any other thoughts?
• Sample rate – I have set this to the theoretical limit of the DAC currently, so 384kHZ but practically I can’t get above 192kHZ, which is preferable?
• Vol Min and Vol Max – These have been suggested previously, do these still look acceptable?

Lounge endpoint

Windows 10 NUC 7I7BNH (dual core) – Ethernet cable – Netgear switch – Patch-box – Ethernet cable – Netgear switch – Allo USBridge streamer – USB cable into Mytek Libery DAC – Musical Fidelity M3i amplifier – Monitor Audio Bronze 5 speakers.

Plan – to upscale all content to DSD 256, the limit of the DAC.

Lounge endpoint questions

• -3dBFS – as above
• Buffer time and DAC bits as above
• 48k – I couldn’t see anything in the manual, when should I set this?
• SDM Pack – should this be set to DoP or none?
• Default Output Mode – Given the aim to upscale to DSD, should I just set like this?
• Oversampling 1x and Oversampling Nx – what is the difference, I could not see an explanation in the manual, how should I use the two?
• Poly-sinc-xtr – mp -2s filters – I know the filters are subjective, however I mostly listen to guitar based or electronic music so I felt this was a good choice, any other perspectives? I tried without the -2s and the music was breaking up, I assume it is too much for my server?
• Modulator – DSD7 256+fs Given my aim for DSD 256, this settings seemed appropriate, should I consider others?
• Bit rate – I have set to 44 x256 – this seemed reasonable, do I need to adjust?
• Vol Min and Vol Max – as above

A few extra questions

• Convolution – I have used REW recently and measured my lounge, and created a set of WAV filters that I have zipped up and loaded into Roon. I followed the guide created by Magnus on this forum. It looks like for the HQ Player’s convolution engine I can only load up a left or right filter say for 44.1 kHZ at 16 bit, the trouble is a good deal of the Qobuz content (and my own content) is above this rate e.g. 96kHz at 24 bit for example, is there a way to add multiple files?
• DSD source settings – I only have a few DSD albums, do I need to touch this setting?

Thank you for any input or advice.

[jussi_laako (Jussi Laako (Signalyst))]

-3dBFS – I understood this was a good place to start, however I have noticed the dial turn red, so I assume I need to turn down. I have gone to -4 then -5, and I watching if it has any further effect.

Vol Min and Vol Max – These have been suggested previously, do these still look acceptable?

When I’m using analog volume control, I set HQPlayer volume control to range from -6 to 0 dB. In this case, normal setting of -3 dB is volume control knob in the middle - point up. This allows easy adjustment of couple of dB and the setting is visually immediately noticeable.

Buffer time and DAC bits – I assume I should leave on default, any other thoughts, in what scenario should I adjust?

Only if you use WASAPI backend on Windows, you likely need to adjust “Buffer time” from default. Since most (all?) WASAPI drivers default to 10 ms buffer which is quite small.

Whenever using S/PDIF or AES/EBU output, like in this case, set DAC Bits to the actual resolution of DAC, but at most 24 (because that’s limit of the interface). This is because for performance reasons many S/PDIF / AES interface use 32-bit samples, but cut out the last 8 bits for sending it out. Setting the word length correctly avoids truncation from destroying the dither/shaper. Since S/PDIF and AES are unidirectional connections, HQPlayer cannot detect the DAC resolution.

Default Output Mode – I intend to switch to PCM for this endpoint as I cannot get DSD to work outside of a direct Windows environment, or it better to leave on source?

In most cases, you should select either PCM or SDM (DSD) as output mode, very rarely “[source]”.

Filter 1x and Filter Nx – what is the difference, I could not see an explanation in the manual, how should I use the two?

If you want different filter for RedBook/48k and hires, you can use this setting, otherwise you can set same filter for both.

Poly-sinc-xtr – mp filters – I know the filters are subjective, however I mostly listen to guitar based or electronic music so I felt this was a good choice, any other perspectives?

This is so much subjective choice… Use what you think sounds best.

Other – NS9 – I am aiming to upsample to the limit of spdif connection, so 192kHz, so this seemed like a good setting, any other thoughts?

That is fine, or you can use just TPDF or Gauss1 as well, since the DAC in question is SDM technology. You can subjectively select which one you find best.

Sample rate – I have set this to the theoretical limit of the DAC currently, so 384kHZ but practically I can’t get above 192kHZ, which is preferable?

It is anyway limited by the hardware to 192k, and with current “Adaptive output rate” setting it’ll fall back to 192k despite having 384k set.

Buffer time and DAC bits as above

Since this is USB connection and the DAC is not fooling, you can leave both at “Default”.

• 48k – I couldn’t see anything in the manual, when should I set this?

Leave it unset, at least Brooklyn DAC+ doesn’t support it, so it would be unlikely for Liberty either. There are only few DACs that support 48k DSD. (Mytek could likely fix this with firmware, but AFAIK they haven’t)

SDM Pack – should this be set to DoP or none?

None

Oversampling 1x and Oversampling Nx – what is the difference, I could not see an explanation in the manual, how should I use the two?

Same as above, if you’d like to have different filter for RedBook/48k content and hires, you have possibility to set different filters. Otherwise, you can set same filter for both cases.

Modulator – DSD7 256+fs Given my aim for DSD 256, this settings seemed appropriate, should I consider others?

ASDM7 and ASDM5 - I personally stick to either of these two, but it is also subjective…

Bit rate – I have set to 44 x256 – this seemed reasonable, do I need to adjust?

44.1k x256 is correct for Liberty (DSD256).

Convolution – I have used REW recently and measured my lounge, and created a set of WAV filters that I have zipped up and loaded into Roon. I followed the guide created by Magnus on this forum. It looks like for the HQ Player’s convolution engine I can only load up a left or right filter say for 44.1 kHZ at 16 bit, the trouble is a good deal of the Qobuz content (and my own content) is above this rate e.g. 96kHz at 24 bit for example, is there a way to add multiple files?

You only need one set of filters with HQPlayer. Recommendation is to create filters for as high rate as possible, 352.8k or 384k 64-bit filter files being recommendation. You can also use filter files created for lower rate and then select “HF Expand” option which will extend the frequency response flat above Nyquist (highest) frequency of the filter.

DSD source settings – I only have a few DSD albums, do I need to touch this setting?

Defaults are good unless you specifically want to tinker with it…

[andybob (Andrew Cox)]

Modulator – DSD7 256+fs Given my aim for DSD 256, this settings seemed appropriate, should I consider others?

I prefer this modulator when upsampling to DSD 512. I find that in my system it preserves more of the attack than other choices.

[tahsu]

Thank you @jussi_laako for taking the time to respond in detail, it is really very helpful to us novices!! This gives me a great jumping off point now to play around with filters. Top stuff!

[tahsu]

Thanks @andybob - I will try and experiment a bit now Jussi has helped me with the basics.

[tahsu]

Convolution – I have used REW recently and measured my lounge, and created a set of WAV filters that I have zipped up and loaded into Roon. I followed the guide created by Magnus on this forum. It looks like for the HQ Player’s convolution engine I can only load up a left or right filter say for 44.1 kHZ at 16 bit, the trouble is a good deal of the Qobuz content (and my own content) is above this rate e.g. 96kHz at 24 bit for example, is there a way to add multiple files?

You only need one set of filters with HQPlayer. Recommendation is to create filters for as high rate as possible, 352.8k or 384k 64-bit filter files being recommendation. You can also use filter files created for lower rate and then select “HF Expand” option which will extend the frequency response flat above Nyquist (highest) frequency of the filter.

I have had a quick play with this. I took some previous measurements (stored as *.mdat files) that I took in January. These originally had both the left and right speaker in one file. So I split them up, with one *.mdat for my left speaker and one for my right speaker in respect of my lounge. I then saved each one into a .wav file like so, using REW, based on Jussi’s recommendations. So I set it to mono, as only outputting in respect of one speaker at a time. 32bit, normalising samples to peak value and 384kHz, as shown below.

I have then uploaded to HQ Player. You will see in the shot below the automated IR Gain. When I played a few tracks I noticed the volume knob go red again, so I went back and added -1.0 dB gain (Update: now currently at -5.0 dB, after playing a number of tracks, getting red and then coming back and adjusting) compensation in the Convolution settings and that seems to have sorted it. Although i will keep an eye on it.

Perhaps @jussi_laako can let me know if any of this is incorrect. I know @dabassgoesboomboom and I discussed the Convolution engine in another thread, so I hope this is helpful.

The only other thing worth mentioning, I couldn’t see any visual confirmation that the convolution engine was running, which might be useful. The box is checked in the settings and you can hear a clear positive difference.

[jussi_laako (Jussi Laako (Signalyst))]

Perhaps @jussi_laako can let me know if any of this is incorrect.

Yes, looks correct. If you know (from REW) how much peak boost (if any) the Eq has, you can directly apply same amount of negative gain compensation. Eq’s with peak boosts will generally sound quieter this way, because there needs to be headroom left for the boosted frequencies when content happens to contain full level signal at that frequency. Which will not usually happen very quickly if the boosted frequency band is narrow, so hard to find experimentally with music (but easy to find with a sweep signal).

The only other thing worth mentioning, I couldn’t see any visual confirmation that the convolution engine was running, which might be useful.

There’s the button in tool bar, with green check mark. This can be used to toggle the convolution filter on/off on the fly.

[tahsu]

Hi @jussi_laako,

If you know (from REW) how much peak boost (if any) the Eq has, you can directly apply same amount of negative gain compensation. Eq’s with peak boosts will generally sound quieter this way, because there needs to be headroom left for the boosted frequencies when content happens to contain full level signal at that frequency. Which will not usually happen very quickly if the boosted frequency band is narrow, so hard to find experimentally with music (but easy to find with a sweep signal).

Ross:

This makes sense, I dug out my screen shots when I originally measured - see below

I am not where I should be looking in respect of peak boosts - perhaps ‘overall max boost’ which is set to 11dB, or am I way off??

[jussi_laako (Jussi Laako (Signalyst))]

I am not where I should be looking in respect of peak boosts - perhaps ‘overall max boost’ which is set to 11dB, or am I way off??

That is the maximum boost the filter design is allowed to use. So setting -11 dB as gain compensation is at least safe. But possibly it didn’t end up using all the allowed amount - this could be seen from the EQ dialog where it shows parameters of each parametric EQ.

[andybob (Andrew Cox)]

I use -8dB when convolving and upsampling in HQP, but other room correction files will have different boosts.

[tahsu]

Hi @jussi_laako

I am using the pipeline setup now after get new filters from Thierry at Home Audio Fidelity. When I was using the Convolution Setup previously there was a visual tick to say it was working. Is there an equivalent when using Pipeline setup? Thanks.

[jussi_laako (Jussi Laako (Signalyst))]

I am using the pipeline setup now after get new filters from Thierry at Home Audio Fidelity. When I was using the Convolution Setup previously there was a visual tick to say it was working. Is there an equivalent when using Pipeline setup?

With convolution engine you can switch it on/off on the fly to compare the results, that’s why you have the visual button too.

But with pipeline this is not feasible because of channel routing and mixing. So when pipeline processing is enabled, it is always active. When pipeline is used for digital cross-overs and such, it is not good idea to switch it on/off on the fly. With the upcoming matrix profiles feature, you will be able to switch it when not playing.

You can check the log file if you want to see details about your filters.

[tahsu]

Thanks for the reply @jussi_laako

I understand (I think). I enabled the log file and had a look at the details. I have attached a screenshot. So I think its working as it should be!

[jussi_laako (Jussi Laako (Signalyst))]

I understand (I think). I enabled the log file and had a look at the details. I have attached a screenshot. So I think its working as it should be!

That’s the configuration file and looks correct. Log file has .log suffix…

[tahsu]

That’s the configuration file and looks correct. Log file has .log suffix…

Whoops my mistake! - I have found the log now…Thanks

[dabassgoesboomboom]

I am using the pipeline setup now after get new filters from Thierry at Home Audio Fidelity. When I was using the Convolution Setup previously there was a visual tick to say it was working. Is there an equivalent when using Pipeline setup? Thanks.

Hi @tahsu

Do you still have your room measurements that you sent to HAF.

And do you have the room measurements with HAF’s pipeline matrix files enabled? Both frequency domain and time domain measurements.

Kind of a ‘before and after’, that HAF’s files made in your room?

[tahsu]

Yes, I think so, @dabassgoesboomboom.

I asked for these from Henry, see screen shots.The first are the original and the second are the adjusted. I also have files with xtalk and without.

[dabassgoesboomboom]

I asked for these from Henry, see screen shots.

Thanks @tahsu

Regarding the files you asked from Henry.

Once you have his digital room EQ files, did you just re-run the same initial REW measurements yourself , but with his EQ files enabled?

Otherwise I imagine he’s just sending you the “predicted” results.

Even expensive software like Dirac and even Genelecs own AutoCal software (and others) give you only the “predicted” response but I find that useless on it’s own.

I’d rather do a proper verification because sometimes “predicted” and “actual” can be way off.

So are those graphs the predicted that Henry sent you or you’ve rerun REW measurements yourself, to verify?

[Audiogeek]

It took me a while to work out that DSP/room correction should be the last step
Looking at those measurements, there are big issues in your set-up. You might be better off working on speaker placement and room treatment before applying dsp.

[tahsu]

No I haven’t re-run REW, I think that is something I will do based on your comments which do seem sensible. I will need to think about how I do it, I might go back to Henry and see what he has to say about it as well.