HQ Player

[Yviena]

Just wondering but is it maybe possible to add samplerates to DSD/PCM that hqplayer detects is not supported, as i can actually play upsampled native DSD256 in Jriver. (driver actually reports 11.2mhz and i get sound)

But in HQplayer i can only use DoP and max DSD128, native in HQplayer only produces white noise.

 

Could be a firmware bug with the dac but who knows.

[Yviena]

nvm i fixed it by flashing XMOS firmware again.

[Yviena]

Is it really possible to create a filter that has the advantages of both linear and minimum phase and no disadvantages?

I found a old article https://mrapodizer.wordpress.com/2011/10/28/introducing-a-new-up-sample-filter/

he says that he has actually created a filter like that or is it bs?

[Yviena]

I just tested some windows builds and it seems that any build newer than 14931 crashes the 64 bit version on startup 32-bit is fine.

 

 

 

BTW just wondering does anyone else have a Geek pulse Finity? i'm curious what filters you guys are using on it.

[Yviena]

Has something changed between the 3.14.4 update and the 3.14.3? the closed form filter is now using 85% with interrupts 13% while on previous version it only used around 50-52% totally. nothing has changed in settings.

[Yviena]

Only the things listed on the web page. So I optimized initialization phase of the convolution engine. And fixed one timing problem in one CUDA algorithm initialization which sometimes resulted in silent output. Nothing else.

 

Of course OS, compilers, etc get periodic updates, at different times on different platforms with more or less unknown content.

Kinda weird because its only the closed form that behaves this way on 3.14.4, if it actually was because of compilers shouldn't all filters have increased cpu usage?

If i disable 1 core in the affinity on hqplayer process the CPU usage lowers SQ stays the same with no stutters, enabling it again increases the usage.

 

EDIT: Ok so using the older exe file gives same cpu usage when used with 3.14.4 files, reverting to the older files but newer 3.14.4 binary reduces CPU usage.

[Yviena]

Just wondering but is it possible to somehow use the hqplayer pcm upsampling filters and use a VST plugin by maybe redirecting the stream to jriver or foobar.

[Yviena]

Hmm looks like it is possible to output the pcm stream of HQplayer to ASIO:Jriver unfortunately it crashes HQplayer with anything higher than 192khz. So it is actually possible to use pcm filters and VST plugins. though it is very buggy and most of the time just crackles/distorts or plainly just refuses to work.

[Yviena]

For anyone who has a ryzen cpu, you can decrease Hqplayer CPU usage  by 2-3% by putting  processor affinity to 0 2 4 6 8 10 12 14 in process lasso.

[Yviena]

I don't really know why but the closed form filter reacts weirdly with my 290x while trying out some games.

 

The CPU usage of hqplayer with closed form, is only 14% but I loose like 20-30 fps with frequent micro stuttering, I'm also noticing it when I'm browsing. But running the xtr filter is around 23% usage but with no impact on fps or performance anywhere. I actually thought xtr would worsen the issue but it looks like it only happens in the closed form filter.

 

I don't really have a Nvidia card to try but could it be a AMD GPUs specific issue?

 

@Miska

[Yviena]

Just wondering but is it possible to have HQplayer have a fallback selection to another filter if the currently selected filter cannot play that samplerate? For example mixed playlists with both 44.1 & 48khz files running closed form when it encounters a 48khz song it just stops and resets and i lose track of what song played last.

[Yviena]

@Miska Have you thought about maybe adding a MP3 decoder now that mp3 licensing is free?

It would be great to save space without re encoding 320kbps to flac just to use HQplayer.

[Yviena]

3 hours ago, MarcinGD said:

Thanks for the info. I'm using 6700k. Could someone report that it works for him on 7700k ?

Depending on your 6700k overclock 7700k will have minimal if any improvement. Your best bet is to either get a CPU with more cores or use CUDA if you have a Nvidia GPU.

[Yviena]

Hmm is the closed form filter a good choice for a oversampling  R2R DAC?

i noticed that running it at 352-384khz bypasses the internal filter/oversampling.

The DAC in question is soekris dac 1541.

 

[Yviena]

@Miska is it possible to change how hqplayer works with invalid files in playlists? 

Currently if there is something like cover.jpg etc the whole playlist becomes empty, would it not be better if hqplayer just ignores the invalid file.

[Yviena]

@Miska Just wondering but is the Hqplayer license for 1 unique computer or can i like also install in a USB drive as i was thinking about maybe getting a laptop. Will it still work if i for example change my CPU or motherboard etc on my desktop.

[Yviena]

@Miska I like how my soekris DAC1541 sounds with the closed form filter with NS5 but just wondering is it possible to make the closed form filter even more computationally expensive to improve it even further?

[Yviena]

Actually wouldn't higher DSD rates also require higher frequency clocks and as a result also have higher phase noise.

[Yviena]

Hmm just switched to the poly-sinc-shrt filter for my R2R dac1541 I kinda feel that the imaging/depth is a little bit better than the closed form filter, could be because it's apodizing.and replacing ADC ringing. 

 

Just wondering but does NOS mean no oversampling anywhere or only for FIR1 as my dac is still oversampling the FIR2 filter.

[Yviena]

@Miska Just wondering but what is the actual stop band attenuation of the poly sinc-shrt filter as i noticed that my dac likes higher than -175db attenuation.

[Yviena]

Hmm I see that most DAC measurement's show  less distortion when ran at  -8 or -10dbfs signal level would setting hqplayer volume to -8 or -10db do the same thing?

 

 

[Yviena]

@Miska

what do you think about this  https://archimago.blogspot.com/2018/01/audiophile-myth-260-detestable-digital.html

 

especially this part

Quote

The key here is to remember that within a properly bandwidth limited signal where all the frequencies are below Nyquist, a linear phase FIR filter actually does not create ringing regardless of the impulse response appearance. As I have said in the previous weeks, any decent recording will follow this rule. And if it does, then the ideal filter to use is clearly a linear phase, sharp filter that can reconstruct all the frequencies in the audio data with essentially ideal temporal resolution.

And that folks is the "myth" we need to say goodbye to in 2018! Linear phase, steep "brick wall" type antialiasing/anti-imaging digital filters performed with high precision, and with no intersample overloading, do not "ring" with good recordings that only contain "legal" frequencies below Nyquist.

 

 

[Yviena]

@miska just wondering but have you considered creating a Mega/xtr variant of poly-sinc-short with more taps etc or is this basically the xtr version?

i kinda want a time domain xtr filter as i find the xtr filter a little bit fuzzy sounding compared to say poly-sinc-short.

[Yviena]

2 minutes ago, Le Concombre Masqué said:

Our tastes might differ, our systems too ; I wish that when discussing the merits of this or that filter we could ALL HERE agree on a few reference tracks, resolution and mastering defined, ie RB Glenn Gould's Goldberg + a few other tracks. I have also published my measured response curves. 

Otherwise, I prefer this or that is difficult to appreciate as a guideline

btw, @Miska, is ext2 apodizing ?

It's also DAC dependent too i have found that my R2R dac either likes closed-form-M or poly-sinc-shrt.

Closed-form for decently  mastered music, poly sinc-shrt for everything else.

[Yviena]

@Miska Hmm do you think HQplayer could take advantage of the new tensor cores in the 2000 series?

 

Hopefully it will be able to DSD512 on my other DAC.