Which HQP Filter are you using? [2024]

I suspect you like it precisely because it is apodizing…

But there’s the poly-sinc-ext2-xl if you want…

Thanks, I’ll experiment with those. From a quick listening session, the apodising version of the ext2 sounds quite different from the Sinc-Mx - leaner, more analytical, possibly a bit airier. The Mx sounds thicker, weightier, meatier in my system to my ears. I presume that it’s an even longer version of the ext2-xla.

I think a non-apodising version of the Sinc-Mx would be a winner for me - the weight and tactility of this filter with that added airiness and a bit more transparency that I perceive the non-apodising filters to exhibit. The fact that it’s apodising no doubt makes it a very pleasant and easy listen on some tracks that otherwise can sound a little harsh, but it’d just be good to have the option available.

Yes, for example recordings made at 44.1k using any of the modern ADC chips will be pretty bad without apodizing filter… Apart from some easy signals (from digital point of view) like solo piano, but then apodizing/non-apodizing also doesn’t matter.

The older recordings from early CD era are generally better. If the recording doesn’t have issues, then the apodizing filter just behaves like a non-apodizing one.

So from technical perspective, I don’t see much point in non-apodizing filters, as they leave all the erroneous mess there. And the more interesting focus is in time/frequency-domain balance.

1 Like

Using HQPlayer through NAA now on my NUC, I was wondering if anyone perhaps had some favorite settings to share for their speakers (if anyone uses it that way).

Formerly only used it with my desktop setup so would like to try some new things here

@jussi_laako

In HQP manual you have marked all the filters with Y/N if they’re apodizing or not. But there are few filters with marking “1/2” in that column. What does it mean?

It is something I consider somewhere in between. Apodizing, but not to the level I would consider complete.

It is result of my re-evaluation of all the filter categorizations. These span almost 30 years of development…

3 Likes

Def enjoying this apodizing Sinc-S filter combo this weekend! Sounds great!

D

@jussi_laako I have a dgx spark and was curious if HQPlayer 5 would work on it since it is ubuntu. My question is - Is there an aarch64 Linux build of HQPlayer 5 Desktop or Embedded with CUDA offload working on Grace Blackwell GB10 / sm_121? I am building a workstation with a threadripper and have an RTX5060 but that is a few weeks out waiting on components. Will my current install on MAC Studio transfer ?

I asked since after a long time using the default poly-sinc-gauss-long, I did some filter rolling and ended up selecting poly-sinc-xtr-mp-2s. It has that 1/2 on the apodizing column. I’ve usually stayed with linear phase filters, but now I find this minimum phase filter impressively accurate, punchy, deep and powerful sounding. Not sure if there’s some synergy with my new DSP filter, which is done with Focus Fidelity software and is minphase also.

1 Like

Did you try by chance good old “fat” poly-sinc-short-mp? If you did - what you think?

After few years of pushing @1024x48 on my Holo May, it was @Moy who inspired me to “test” @256 - now i am totally addicted 7EC-Fast + short-mp, no mater what i try i come back to this combo @256. “defaults” are very nice too. surprisingly, sinc-long-h is equally awesome in some cases. The reason i am curious - i’ve tried so many times the “ext”’s and “xtr”’s , especially xtr’s looking for “snappy” transients, but something does not work for me…

3 Likes

I think I did but I’m not 100% sure, I can check it again later today. I liked poly-sinc-long-mp also, but xtr-mp was the punchiest/most dynamic and somehow deepest sounding of the mp-variants in my setup. It’s also easy to listen to. Ext-family is a whole another thing. Surgical precision and extreme resolution, if that’s your thing. I’ve never played with the -ext variants for longer than testing them.

I upsample to DSD256x48 with DAC200. I’ve never been much into DSD1024. DSD512 is nice with slightly bigger and smoother sound, but DSD256 is the most forward and cleanest sounding.

I also went back to DSD256 awhile ago and never looked back. I can’t pull myself away from sinc-MGa / ASDM7EC-fast

1 Like

The choice of filter etc sure dépends on drc. i reworked my drc, finally adding 4 eq points above transition , admitting that the side walls, too close, actually mess hf. And that leads me to like most the recommended settings 256 7ec fast psg long. Or hires with hires. Another conséquence is that linear drc is a too dangerous game. It seemed that linear drc and linear filters made sense . I think it s more source dependant (analog source, mp adc…) to pick or not a mp filter. anyway it s free and easy with Client. The one rule i think is that filtering etc should not compensate for inadequate drc. At least drc should be judged with best objective settings

I tried the poly-sinc-short-mp. I understand what you mean now, it’s thick and full sounding filter with great punch. Slightly different character to the xtr-mp though and less space in the sound. There’s also the xtr-short-mp available but somehow I like the balance of depth, space and punch with the xtr-mp. Anyway, it was a good decision to dive into the mp-filters, great options among them.

I only do room correction for ~100hz window between ~25-125hz, works best in my space. Sounds best to not go over room’s schroeder frequency.

EDIT: Fixed 125-125hz to 25-125hz :slight_smile:

2 Likes

Very narrow :grinning_face:

Now as I read it, I realised I was only largerly experimenting with xtr-short-mp for comparison, indeed there is “just” xtr-mp !

2 Likes

Anybody using Aes instead of i2s or usb ?

but isn’t it limited to 24/192 and DSD/DoP64?

Not at the moment. I have the build environment and can build such. But I don’t have hardware to test on, so I’ve been reluctant to make such release.

As long as you stay with HQPlayer Desktop, it is rather easy to transfer the license to another machine.

It seems you re right ! Which makes it pretty useless for us. And Stereophile s praise of a 35k streamer tested via aes pretty useless too