Which HQP Filter are you using? [2015-2023]

Well there you go. Thank you for your help

20 is better than ā€œdefaultā€?

1 Like

Very good question, I was wondering that myself, I thought the ā€œdefaultā€ option queries the DAC and the DAC provides the bit depth, now if HQP further internally reduces that to a ā€œsafe better linearity valueā€ or if it uses the full depth only Jussi can answer. Good one Christopher

In this case likely yes. Default detects number of bits the digital device / driver interface can accept. Depending on the backend it may be more or less accurate, but it is always at least number of bits driver accepts.

You cannot exceed the detected number of bits with the manual setting, but you can go below it when you know better.

2 Likes

Sweet, thanks.

Iā€™ve really cooled my jets on the HQPlayer settings. Meaning, Iā€™m not trying to do too much. Iā€™ve settled on the ext2 filter, and Iā€™m sticking with PCM at a max rate of 192khz. I donā€™t really need to stretch things to 768khz or convert to DSD or whatever. (Besides SACDā€™s, I donā€™t have any DSD files at this point.)

It sounds absolutely fabulous ā€“ no dropouts or other buffering weirdness ā€“ and the processor (mac mini) seems to prefer a less-ambitious configuration.

1 Like

I have now set to 20 rather than 24ā€¦but what is default setting. Is that ā€œ0ā€?

When I say 21 bits, I mean what DAC supports a dynamic range of more than 21 bits? Heck, most DACs today, even R2R DACs, do 24 bit math. So, 21 bits of resolution means 21 bits of linearity which means 21 bits of actual dynamic range.

After being away from HQPlayer for a couple of years, a series of unfortunate (or perhaps fortunate) events caused me to give HQP a second try. I lost my Chord HMS due to a busted power supply. One of my friends encouraged me to look at HQP for PCM up-sampling to Chord DAVE. I am using Sinc-M with max rate of 768kHz with auto rate family.

I have a Audio Linux i7 NUC server running Roon Core and HQP embedded (which I bought after just couple of days of trial, so thanks @jussi_laako as that is how long it took me to realize the potential). My current end point is a Pentium NUC also running AL.

PCM upsampling to 705/768kHz got me very close to what I am used to hearing from HMS + DAVE. PCM upsampling with HQP seems like a feather weight for the CPU. Both my NUCs are idling at 97% or more, not using CUDA or anything fancy.

Initially the infamous white noise issue reared itā€™s head on my right channel with 96kHz or higher source material, it was not very predictable but always happened. At that point I was running NAA on Euphony. Then I remembered reading Jussi suggesting it could be the end point not being able to keep up with the high sample rate especially with Chord DACs. I moved to AL NAA and things have been steady since.

To provide a subjective point of view (hopefully useful to someone) to the discussion, this is what I seem to prefer sonically after almost a week of extensive testing with Metrum Onyx R2R DAC over USB and coax (4-8x PCM upsampling, 20-32 DAC bits settings) in a headphone audio chainā€¦

Filters:

  1. poly-sinc-ext2
  2. closed-form
  3. poly-sinc-short-lp

Dithers:

  1. LNS15
  2. NS5
  3. No dithering

Please note that poly-sinc-ext2 is a clear favourite of mine and somehow seems to synergize well with LNS15. That might be the primary reason why LNS15 pulled ahead. I would also point out that the majority of filters and even dithers included in HQPlayer were subjectively not to my liking, wouldnt use them on a long-term basis.

2 Likes

The dropdown should have ā€œdefaultā€ as the default, first selection.

1 Like

No, linearity and dynamic range are not the same thing! Specifically on R2R. You can improve linearity and thus low level distortion significantly while improving SINAD. By using less bits in a correct way.

Hey all! So i noticed when using hqplayer 4 desktop, if i check mark ā€˜multi-core dspā€™ i stutter like crazy. When i switch it off or to grayed out, i dont have any stuttering. In my task manager, with ā€˜multi-core dspā€™ on, my processor is working at around 32% usage, where as with ā€˜multi-core dspā€™ off, It averages 20% usage. I have an i9-9900k. with 2070 super. Anyone else experience this and is there a reason for it? Im trying to max my performance of my i9 for hqplayer so any help is appreciated <3

Also, I wanted to ask if anyone had a good bit resolution for the Holo Audio Spring 2 R2R DAC. Currently set at deafult, but willing to hear what people have liked with other specific resolutons

Coolā€¦I probably did see it when I first got it but just forgot it was there lolā€¦cheers

Seems like 20 is the best rate to add according to the developer, so who am I to argue

I sold my mscaler. HQPlayer provided the same sonics (to my ears) and spent the money on a server with low noise usb and HQPlayer embedded. Just better and far more flexible.

2 Likes

I see it as being system dependent especially given both HQP and HMS are both up-scaling using 1M Sinc filters. So it will come down to quality of USB connection, quality of power, server etc. All things being equal, then it will be a matter of taste WTA filter vs HQP Sinc filter.

Not having my HMS, I am unable to compare and if one likes the WTA filters, then there is an advantage of the simplicity HMS offers as it is easy for me to route multiple sources through it (like listening to music and also watching movies).

From a strict music listening perspective, what I see with HQP and PCM upsampling is potential to possibly go beyond Sinc-M, then perhaps HQP can clearly jump ahead of HMS if @jussi_laako can provide such options. Though I do not know if there is any true advantage of filters longer than 1M and where the next big step lies.

Yes, I agree. I also found the mscaler a more stable platform. I felt that with the amount of rfi purported to be induced by the mscaler that very expensive cables were probably required to obtain the sound improvement that was promised. I did several A/B comparisons between the mscaler and HQPlayer. These tests were not totally blind but I found there to be no discernible difference in my system. I donā€™t feel one is better than the other but I do find that either the WTA or HQP filters add something that approaches the analog sound of vinyl without all itā€™s inherent problems.

1 Like

Soooooā€¦ interestingly, listening to a Qobuz stream of a Paul Simon album. FLAC 96/24. I have bits set to 20 as suggested, and Roon canā€™t play it. Error ā€œRoon lost control of the audio deviceā€ keeps popping up. So I go in and change the bits back to default.

And it plays no problem.

What to make of it?

Hmmmā€¦I donā€™t have that problem. Iā€™ve never had it on default