HQ Player

[LoryWiv]

2 hours ago, LoryWiv said:

Good suggestion, Jussi.I have changed from balanced to high performance and cpu utilization during playback is a few percent less, This is still using HQP Desktop 4.30 as I'm reluctant reluctant to go back to 4.31 due to the cpu consumption at idle issue. I will report back it it makes any additional difference.

At Jussi's suggestion I changed Win 10 to "High Performance" power setting and tried HQ Player Desktop 4.31 (latest) again. Unfortunately, the problem persists whereby having client and server open together consumes ~15% cpu cycles even when idle with no playback. Went back to prior version 4.30 and all is well. I'll stay with 4.30 for now and hopefully whatever gremlin is causing this issue will resolve with next update. 😬 4.30 is thankfully still working and sounding great! 😁

[Miska]

1 hour ago, der_yeti said:

I run HQPlayer on a Windows 10 PC with i9-9900k. The DAC is connected by USB.

 

I assume you are using ASIO driver then? From HQPlayer, leave "Buffer time" set to "Default". Then from the driver control panel (usually in system tray) select safest mode / biggest possible buffer.

 

[evalon]

Hi all ...

 

I reckon this question may have been asked before but to me this would be quite attractive so I hope you will bear with me likely asking the question once more:

 

Since HQPlayer requires a quite powerful computer to run DSD at high rates "on the fly" do you know if it would be possible to use HQPlayer to convert a music file e.g. to DSD1024, save this file, and then play it back just like any other DSD file? I assume this would allow many more "normal" computers to play back HQPlayer converted DSD files at a high sample rate - even if it is not "live".

 

Personally, I often listen to music/sound in the semi-background - and here it is less important to have the best sound quality - whereas when I really take time off to listen I would prefer the best possible sound quality.

 

However, I don't have a very fast computer so it would be very attractive for me to use HQPlayer to convert the DSD files in the background and then be able to play them back at a convenient later time.

 

Thanks for any feedback on this 

 

Jesper

 

 

[craighartley]

30 minutes ago, evalon said:

Hi all ...

 

I reckon this question may have been asked before but to me this would be quite attractive so I hope you will bear with me likely asking the question once more:

 

Since HQPlayer requires a quite powerful computer to run DSD at high rates "on the fly" do you know if it would be possible to use HQPlayer to convert a music file e.g. to DSD1024, save this file, and then play it back just like any other DSD file? I assume this would allow many more "normal" computers to play back HQPlayer converted DSD files at a high sample rate - even if it is not "live".

 

Personally, I often listen to music/sound in the semi-background - and here it is less important to have the best sound quality - whereas when I really take time off to listen I would prefer the best possible sound quality.

 

However, I don't have a very fast computer so it would be very attractive for me to use HQPlayer to convert the DSD files in the background and then be able to play them back at a convenient later time.

 

Thanks for any feedback on this 

 

Jesper

 

 

HQPlayer Pro already offers this facility. 

[jimdukey]

I guess the ifi could also receive unprocessed DSD and PCM to get DSD 1024.

But then the Advantages of using HQ would be lost.

I don't know why Jussi doesn't seem to like the ifi, I know he likes the RME ADI2, much cheaper.

I have read 3 rave reviews of the ifi, which got my attention.

I don't have the Computer Knowledge to evaluate DACs, and their Promises...

The ifi is said to be full and rich, I've seen the RME described as lean and accurate.

Luckily both products can be returned for refund.

That's Huge, because in the past , at Brick and Mortar stores, Store Credit was the best you could hope for in a Return.

It's a bummer that the Firmware update broke 48K, I like that extra scosh of upsampling it delivers.

[perpetualapprentice]

6 minutes ago, jimdukey said:

It's a bummer that the Firmware update broke 48K, I like that extra scosh of upsampling it delivers.

 

The question becomes 'is 48K more important than MQA?' Reverting to older firmware is identical to the 'upgrade' process. Unless they eliminate that issue, my iDSD micros will forever remain with the v5.2 Limoncello firmware which was the last available before MQA support was added.

[jimdukey]

48K is definitely more important than MQA!

But at DSD 1024, does it make any difference?

[der_yeti]

7 hours ago, Miska said:

 

I assume you are using ASIO driver then? From HQPlayer, leave "Buffer time" set to "Default". Then from the driver control panel (usually in system tray) select safest mode / biggest possible buffer.

 

It's ASIO, right. Will check your advice - thanks!

[luisma]

I can't remember the answer and I think this was asked before.

HQPlayer digital volume, it is very good but not lossless correct? Using it with Roon the best setting is at -3db, anything above will clip.

 

[alec_eiffel]

On 2/9/2020 at 7:10 PM, alec_eiffel said:

Hey, happy HQPlayer customer here.`

My setup:

- Roon doing stereo --> stereo HAF Convolution for room correction at source rate

- Roon outputting to HQPlayer (same machine), for upsampling to 384kHz and active filtering, stereo --> 8 channels (4 channels per speaker, Bass1, Bass2, Mid, Tweeter)

- Network streaming to NAA endpoint plugged into Exasound E28  DAC

 

First, it seems that I cannot implement Room correction AND active filtering in HQPlayer, there's only one matrix, correct? If I want to perform all processing in HQPlayer, I guess I need to manually convolve my room correction files with the active filters file, right? Is there an easy command line tool for this ? 

 

 

Answering to myself...

I found a way to combine room correction and speaker filters using ConvolverCMD.

Here's the batch file:

convolverCMD.exe 0 1 -2 HLL.txt filterBASS1.wav B1HLL.wav
convolverCMD.exe 0 1 -2 HRL.txt filterBASS1.wav B1HRL.wav
convolverCMD.exe 0 1 -2 HLL.txt filterBASS2.wav B2HLL.wav
convolverCMD.exe 0 1 -2 HRL.txt filterBASS2.wav B2HRL.wav
convolverCMD.exe 0 1 -2 HLL.txt filterMEDIUM.wav MediumHLL.wav
convolverCMD.exe 0 1 -2 HRL.txt filterMEDIUM.wav MediumHRL.wav
convolverCMD.exe 0 1 -2 HLL.txt filterTWEETER.wav TweeterHLL.wav
convolverCMD.exe 0 1 -2 HRL.txt filterTWEETER.wav TweeterHRL.wav
convolverCMD.exe 0 1 -2 HLR.txt filterBASS1.wav B1HLR.wav
convolverCMD.exe 0 1 -2 HRR.txt filterBASS1.wav B1HRR.wav
convolverCMD.exe 0 1 -2 HLR.txt filterBASS2.wav B2HLR.wav
convolverCMD.exe 0 1 -2 HRR.txt filterBASS2.wav B2HRR.wav
convolverCMD.exe 0 1 -2 HLR.txt filterMEDIUM.wav MediumHLR.wav
convolverCMD.exe 0 1 -2 HRR.txt filterMEDIUM.wav MediumHRR.wav
convolverCMD.exe 0 1 -2 HLR.txt filterTWEETER.wav TweeterHLR.wav
convolverCMD.exe 0 1 -2 HRR.txt filterTWEETER.wav TweeterHRR.wav

HLL.txt HRL.txt HLR.txt HRR.txt are "convolver" files that are meant to extract the relevant mono filter from the stereo HAF_HL and HAF_HR correction files, for example HLL.txt is :

384000 1 1 0
0
0
HAF_384_HL.wav
0
0.0
0.0

Last step is to import all 16 filters in HQPLayer Matrix, filter path is blanked for privacy.

As expected, processing is heavier as "long" room correction filters have to be applied 16 times instead of 4 previously, CPU usage increases from 30% to 50% on all 4 cores. SQ has slighly improved compared to room correction in Roon at 44.1kHz. I'll need more time to A/B, but I am happy I did it

 

[evalon]

@craighartley: 

 

Quote

HQPlayer Pro already offers this facility. 

 

Thanks for the feedback, however, HQPlayer Pro is a little bit outside of my budget ... So I assume this also means that it is not in the Desktop version, right?

 

Cheers - Jesper 

[craighartley]

17 minutes ago, evalon said:

@craighartley: 

 

 

Thanks for the feedback, however, HQPlayer Pro is a little bit outside of my budget ... So I assume this also means that it is not in the Desktop version, right?

 

Cheers - Jesper 

No. Only in Pro

[Miska]

3 hours ago, luisma said:

I can't remember the answer and I think this was asked before.

HQPlayer digital volume, it is very good but not lossless correct? Using it with Roon the best setting is at -3db, anything above will clip.

 

How do you define "lossless"? In most cases it is lossless. And certainly in most cases less lossy than anything analog...

 

[StreamFidelity]

4 hours ago, Miska said:

In most cases it is lossless. And certainly in most cases less lossy than anything analog...

 

I use the volume control of the HQPlayer with great enthusiasm.

 

With a sound engineer I compared the volume control with the Amp T + A PA 3000 HV. Result: Since then I have been controlling my mono amplifiers directly without a preamplifier. I can regulate the volume well via Roon, which reliably sends the signals to the HQPlayer. 👍

 

Nevertheless, I'm interested in the technical background. I think I have read that every attenuation with -6dB means a bit loss of 1. If the HQPlayer scales with 64 bits and the DAC gets a maximum of 32 bits, I am not worried. 32 bits means -192 dB headroom. Have I understood that correctly?

[luisma]

On 2/11/2020 at 6:31 AM, Miska said:

 

How do you define "lossless"? In most cases it is lossless. And certainly in most cases less lossy than anything analog...

 

let me see if my answer lives to your expectations probably not but I would appreciate your straight feedback to my simplistic reasoning

 

Lossless by definition (at least to me) means unchanged and without degradation during the playback process, I understand from the moment we are taking a source file or content and upsample such content in HQPlayer of course the content changes completely and in any case is not degraded but improved so of course since you designed the entire code and you apply filters, dithering etc. you have your own algorithm to increase gain (volume) on such content.

What is is not very clear to me (and I should have been more explicit) is how it works with Roon / HQPlayer, I would think when you increase volume in Roon what Roon does is simply pass that setting transparently to HQPlayer without (Roon) doing anything else. I remember back like 2 years ago you advised to use the volume control in HQPlayer freely as it was really good instead of any other gain controls.

You have also explained several times to limit HQPlayers volume to no further than -3db as beyond that clipping could occur.

 

Then 2 days ago a friend of mine who I have been trying to sell HQPe sent me this

"With the pending commercial release of Leedh's 'super' algorithm for truly lossless digital volume, plus advanced up-sampling—already incorporated by Soulution"

and  I wanted to provide an exact answer for him, maybe we should ask why they call it "truly lossless" instead.

 

The intention behind all this is to have a simpler (more cost effective) path. Connect our DACs to our AMPS (without a pre) and use just the digital volume control of HQPe (at least that's what I'm proposing) no DAC with integrated volume or anything like that.

 

Last but no least does HQPlayer volume quality is the same at -30db that at -3db? or you would recommend to have it always if possible at -3db?

 

Thank you

 

 

 

[Miska]

1 hour ago, luisma said:

What is is not very clear to me (and I should have been more explicit) is how it works with Roon / HQPlayer, I would think when you increase volume in Roon what Roon does is simply pass that setting transparently to HQPlayer without (Roon) doing anything else. I remember back like 2 years ago you advised to use the volume control in HQPlayer freely as it was really good instead of any other gain controls.

You have also explained several times to limit HQPlayers volume to no further than -3db as beyond that clipping could occur.

 

Yes, that's the case, you can see when you adjust volume in Roon, the volume setting in HQPlayer changes.

 

1 hour ago, luisma said:

"With the pending commercial release of Leedh's 'super' algorithm for truly lossless digital volume, plus advanced up-sampling—already incorporated by Soulution"

 

To me, that looks like marketing jargon.

 

1 hour ago, luisma said:

The intention behind all this is to have a simpler (more cost effective) path. Connect our DACs to our AMPS (without a pre) and use just the digital volume control of HQPe (at least that's what I'm proposing) no DAC with integrated volume or anything like that.

 

In this case it is recommended to have some safety precautions, such as low enough gain in the power amp or some other mean that limits the maximum possible volume to a safe value. To void accidents due to configuration mistake or similar. For example Windows sometimes decides to change the default audio output device and then some system notification sound could cause a nasty surprise.

 

1 hour ago, luisma said:

Last but no least does HQPlayer volume quality is the same at -30db that at -3db? or you would recommend to have it always if possible at -3db?

 

If we assume for example that source is RedBook and output is 32-bit PCM and you use just TPDF dither, you can attenuate up to -96 while the volume being "truly lossless". If you use noise shaping with upsampling, you gain at least some ~40 dB more.

 

With DSD, it is similar, exact figures depending on the output rate and modulator.

 

If you then take into account that analog noise floor anyway sits somewhere around -120 dB level equivalent of about 20-bit PCM resolution (with TPDF dither), you don't have much worries. Because that is far above the digital one.

 

Only if you output at 16-bit PCM, then you need to carefully utilize both upsampling and noise shaping and then you can still have digital noise floor below analog one.

 

[Yviena]

32 minutes ago, Miska said:

 

Yes, that's the case, you can see when you adjust volume in Roon, the volume setting in HQPlayer changes.

 

 

To me, that looks like marketing jargon.

 

 

In this case it is recommended to have some safety precautions, such as low enough gain in the power amp or some other mean that limits the maximum possible volume to a safe value. To void accidents due to configuration mistake or similar. For example Windows sometimes decides to change the default audio output device and then some system notification sound could cause a nasty surprise.

 

 

If we assume for example that source is RedBook and output is 32-bit PCM and you use just TPDF dither, you can attenuate up to -96 while the volume being "truly lossless". If you use noise shaping with upsampling, you gain at least some ~40 dB more.

 

With DSD, it is similar, exact figures depending on the output rate and modulator.

 

If you then take into account that analog noise floor anyway sits somewhere around -120 dB level equivalent of about 20-bit PCM resolution (with TPDF dither), you don't have much worries. Because that is far above the digital one.

 

Only if you output at 16-bit PCM, then you need to carefully utilize both upsampling and noise shaping and then you can still have digital noise floor below analog one.

 

Wouldn't -26dB+ digital attenuation if the analog noise floor is at -120dB reach the 96dB SNR of 16 bit CD content, and what about 24 bit content wouldn't any digital attenuation there reduce possible information?

[Miska]

58 minutes ago, Yviena said:

Wouldn't -26dB+ digital attenuation if the analog noise floor is at -120dB reach the 96dB SNR of 16 bit CD content, and what about 24 bit content wouldn't any digital attenuation there reduce possible information?

 

Would be the same with "perfect" analog volume control as well, because that loss is inherent to analog domain. With analog volume control you have some added noise and distortion components. With digital volume control you just have constant analog noise floor and what ever performance the DAC is capable of. If the DAC is good, analog volume control can only make it worse.

 

So the method of volume control doesn't remove the fact that analog domain is the limiting factor. Not the digital one.

 

Solution is to have right amount of gain in the power amp, so that you need only minimal amount of attenuation before it. Because it is counter productive to attenuate a lot just because next you amplify it a lot. Something like Benchmark AHB2 has low gain mode for this reason.

[Lbameule]

In my understanding, Lossless is any (digital) file format that maintains the "sound wave information" of at least 44.1/16bit, ie CD/RedBook, equal to at least CD quality. This can happen both compressing (like FLAC files and other formats) or not compressing (WAV, AIFF, please correct me if I'm wrong) the information in the file, the thing is that the "output" must be at least equal to 44/16.

 

Cheers,

 

 

L
 

[Yviena]

45 minutes ago, Miska said:

 

Would be the same with "perfect" analog volume control as well, because that loss is inherent to analog domain. With analog volume control you have some added noise and distortion components. With digital volume control you just have constant analog noise floor and what ever performance the DAC is capable of. If the DAC is good, analog volume control can only make it worse.

 

So the method of volume control doesn't remove the fact that analog domain is the limiting factor. Not the digital one.

 

Solution is to have right amount of gain in the power amp, so that you need only minimal amount of attenuation before it. Because it is counter productive to attenuate a lot just because next you amplify it a lot. Something like Benchmark AHB2 has low gain mode for this reason.

So would for example maxing out the volume on a headphone amp to effectively bypass the volume pot, and attenuate digitally be okay, or would you recommend setting the pot to 12 o'clock?

[luisma]

2 hours ago, Miska said:

Only if you output at 16-bit PCM, then you need to carefully utilize both upsampling and noise shaping

The output we are talking about the DAC capabilities correct, usually most DACs supports 24 or 32 bits and that's the value we should configure on HQP for PCM and DSD.

[ericuco]

8 minutes ago, luisma said:

... and DSD.

 

DSD is 1-bit.

[Miska]

16 minutes ago, luisma said:

The output we are talking about the DAC capabilities correct, usually most DACs supports 24 or 32 bits and that's the value we should configure on HQP for PCM and DSD.

 

It is usually auto-detected when set to "Default", especially on Linux after I improved that piece of kernel. You have that if you use my kernel. Or if you use for example 5.4 upstream kernel. When going over unidirectional connection like S/PDIF it needs to be set manually because there's no way to detect what is at the other side.

 

Especially on R2R DACs it is important to set it to linear range of DAC instead of what is "supported". For example on Holo Audio DACs to 20-bit. And then you can use noise shaper to pull dynamic range beyond. This linearizes the DAC.

 

And yeah, for DSD it doesn't apply at all.

 

[Miska]

29 minutes ago, Yviena said:

So would for example maxing out the volume on a headphone amp to effectively bypass the volume pot, and attenuate digitally be okay, or would you recommend setting the pot to 12 o'clock?

 

Depending on gain of the amp, for safety I would do the latter and then use digital volume for the rest.

 

Good compromise is to first set digital volume to -3 dBFS and then analog volume control to as high as you would ever want to listen. And then use digital volume control for the rest. This gives safety against accidents, while improving THD+N and channel balance. (pots have worst channel balance at lowest volume settings)

 

[Hensema]

I like to oversample 1x together with LNS15 noise shaping. However, when Nx is selected without filtering (NOS), is the LNS15 in that case also automatically disabled?