Well, yeah… add the old trick on top → switch off “auto rate family” and set the rate limit to “45158400” - than indeed upto 96/24 soirces one can get stable “Light 512+fs”@1024
It’s just these new “512+fs” for 5 and 7 are so heavier than their “simple” versions in 1024 domain…
also with convolution but, nope - can’t do “super 512+fs” with out drops.
“light 512+fs” works for 44/16, all right…
7EC-super, 7EC-light, 7ECv3 all are very fine with @1024x48 with all sources up to 192/24, with Sinc-L (that takes rediculus amount of RAM @1024) or new Sinc-Long…
Was surprised - can’t do new IIR or IIR2 @1024 - very demanding filter that is.
Running 13900KS, but for this build I have left it pretty much at stock setting and did not go “overclocking”. As the resul the build is so stable, like I have never seen before, current run time is 66 days (no restart), and I don’t really want to mingle for now… just keep it running. @jussi_laako - so excellent job with version 5 !!!
Yes, 4090. It can do just all @512x48 for me, with sources up to 192/24 (i dont’t have the “higher” sources). But my sandbox is @1024x48 I don’t do much 512.
Does anyone know where the closed-form-16M filter went? it’s in the manual but in the selection screen of either the client or the server I can only see closed-form,normal, fast and M…
EDIT: Apparently 16M is only available for DSD output, not sure why though.
Because DSD output rates are higher and thus can also accommodate much higher filter lengths.
16 million taps linear phase causes 8 million samples delay. Sample rate is naturally number of samples per second. So you can calculate amount of delay it would have at PCM rates…
nay - “nblocks” is somewhat an enigmatic hqplayer parameter that can be tweaked via hqplayerd.xml, but legends hold a secret to what the right values could be as there’s no right or wrong.
in my earlier quest for having Sinc-L performing @1024x48 with sources up to 192/24 that setting did play a role - i can enjoy Sinc-L with all EC modulators but 512+fs group, but that is already huge!
@jussi_laako : objectively, is there any advantage to using the 7EClight + 512fs filter instead of the 7EClight for DSD256 output on an A26 ?
There’s a listening advantage ? also the processing advantage (assuming the same load) ?.
I find it hard to evaluate : not worse, better ? a cognitive bias ?
I am sensitive to issues in dynamics caused by poly-sinc filters converting 48k to 44.1k in HQPlayer, but I don’t perceive any issues if I just let my DAC handle 48k inputs. This is why I leave 48k and adaptive rate on.
If the source is a multiple of 44.1k, I enable DSD Direct on my DAC for a slight improvement in transparency. If the source is a multiple of 48k, I disable DSD Direct on my DAC, otherwise I hear audible crackles.
For some reason, many of the HQPlayer filters create a sensation of the sound coming from behind my ears, which is unpleasant. BUT, the halfband filters, especially poly-sinc-gauss-halfband, do NOT cause this issue. This may have something to do with (quote) “halfband filters pass the original data through as-is and only add new samples between.” And, this filter sounds much more transparent than just outputing PCM. I have not had issues with the lack of apodizing as most of my sources are high quality. So all-in-all, I am very satisfied with HQPlayer now that I’ve found these parameters!
You have to email Gustard directly to get a copy of it. It is different than the Gustard Renderer driver, which is available for download. I can email you what they sent me as well, if you want since it is a tiny file.
You can test to see if you need it by turning 48k DSD on, turning adaptive rate off, setting the bit rate limit to 48k x256, and upsampling a 48kHz (or 96kHz) PCM source file to DSD256 (12.288M 1-bit output) in HQPlayer. Enable DSD Direct on your Gustard A26. If you hear crackles and pops, that means you are up to date. If you don’t hear crackles and pops, that means DSD Direct is not working, and all your audio is being filtered.
No, crackling can be heard throughout the playthrough if HQPlayer is converting a 48kHz family file to DSD output with DSD Direct enabled. For some reason, the DAC chip doesn’t like that input directly, but it processes it correctly when pre-filtered. But that also makes it easy to tell if your DSD Direct functionality is working or not.
There’s no simple answer. If you do a lot of EQ, or multichannel-to-stereo mixdown, or if you use HQPlayer’s volume control, the 512+fs may be better since it’s goal is to provide extra headroom in <100 kHz band at the cost of bandwidth (which you have plenty anyway at 512+fs rates). If you run it at DSD256 we are talking about <50 kHz band.
But in the end, if nothing else, laws of physics in analog domain will be the limiting factor.
I have also HQPlayer → A26 → Oor + Hypsos, but different headphones. HQPlayer is also doing headphone EQ as defined by Oratory (PDF).
But I run with default filters 1x=poly-sinc-gauss-long Nx=poly-sinc-gauss-hires-lp and ASDM7EC-light modulator.
Most of my content is RedBook and most if it will trigger Apod counter incrementing, so I stay with apodizing filters. Especially since there’s no harm running apodizing filters for something that doesn’t need such.
Nice! I have been recommending the A26 + Oor stack to everyone interested in your software. It is a great value for people who don’t have the money for Holo Audio, but still want significantly benefit from HQPlayer, especially your PCM to DSD schemes. The stack has enough transparency and staging to differentiate your different filters. Also, the voltage setting on the Hypsos can balance out filters that are good, but feel too harsh. The A26 and Oor both pop up regularly on used sites, too, which is where I got mine for a good price. I also tested HQPlayer on my more budget iFi Zen stack, but it was more difficult to notice concrete differences between filters and modulators.
I use EQ in the pipeline, just a simple 5 band filter that I ripped off of Resolve to fix the Arya Organic’s boosted treble. I was wondering if there an advantage to using EQ in the pipeline as apposed to EQing in Roon, besides the quality of the EQ? Do your filters or modulators interact with or account for the EQ in any way?
hi,
I am currently using Convolution filters made with Focus Fidelity in Roon (Zip of Wav files).
If I remove them from Roon, how do I integrate them into HQPlayer?
I can have Wav files L/R separated.
I have PCM files from 44.1/48 to 705.6/768 K (multipe 44.1 or 48) and DSD 64 to 512 (so multiple 44.1), in this case what frequency choice should I put in HQPlayer?
I don’t do much 512.
