Thanks for your reply! 
My monitors (DAC included) is too old to facilitate the DSD stream. Compared to todays very high bit depth and rates, my 24 - 96 is rudimentary, but still I would say that the singel largest step is taken from 16-44.1 to 24-96. But I wish I could have a pure direct line from my analogue inputs to the power amps, but they made this feature non-availble because most studio people want to have some room and EQ correction. So there is a mandatory AD converter on the analogue input which of course sounds crappier then the AES3 direkt input with any DSP shut off. Guess I have come to a point to decide to continue with these monitors, trade them for something better or re-build them to add the needed features.
Really enjoying poly-sinc-short-mp with headphones now 
Did anyone experience changes in sound when the basic Linux distribution in HQP 4 Embedded were updated some issues before this latest 4.22? I went from 4.19 to present so I skipped a few intermediate and there is a subtle change in sound.
Anything you know about, Jussi?
I went through every single iteration and personally did not notice any difference in sound, Lars you noticed the difference with DSD, PCM or both?
question: if iâm using AES/EBU to a Rockna Wavelight dac (AES/EBU 24bit 44.1-192k PCM, DSD64), is there any benefit if i set it at 24 bit in hqplayer settings? or should i just leave it at default?
I do not play DSD enough to be certain about changes, so this is my NAS library and Tidal mostly FLAC, wave and some master from Tidal. There is a tiny roll-off i high notes, leaving the especially cymbals almost half a dB held back compared to previous version. I mean the difference is subtle, really, but I know my reference playlist by heart and also my rig now and spot even the slightest change in my sleep.
Canât tell what it is. The software Jussi makes should be equal as there is no thinkable incentive for him to change anything but fixing bugs etc or launch new features, but this is exactly as is. I restored my setting from 4.19, just after booting up 4.22 and the change hit me right away and without me expecting any changes, which may lead to psykoacoustic expectations.My only thought is something in the kernel in the new OS?
I use every every additional sample and bitdepth I can get from HQP, but also let the software judge if it uses even multiples of base rate. I feel the bitdepth conversion allow the software to position the different sounds better in sound stage depth and width. Almost as grading when it comes to pictures, when there is a lack of values to paint every grade, there are more intermediate dynamic levels available. It is my personal experience and opinionâŠ
So you do set the appropriate bitrate in hqplayer, depending on the dac and protocol used?
No, I do not set bitrate, I let the software know the maximum bitrate the DACâs can handle. Then the software decide depending what outcome will be best as a result from the filter, dither, etc setting I have chosen in general. Since my monitor Dacâs support 24/96, I choose 24 bit output, I define upper limit bitrate to 96 kHz, and then depending on the source bitrate and depth, the software (i.e. HQP) manage the appropriate bitdepth and rate accordingly. So when I play CD source files from my NAS, HQP upsamples to 24 bits depth and 88.2 kHz as it is the even multiple of 44.1. If I play a Tidal Master 24/48 or 24/96 it outputs 24 bit depth and 96 kHz rate since it is both equal tp 96 in and double speed to 48 kHz source material. Giving HQP the role of deciding this, have in my case been beneficial for sound quality. Before I found out this feature/option, I forced 96 kHz rate out whatever source rate, and that did not sound as well as present setting. No harsh highs anymore.
1 Like
Edit: For this option to work, you need to set desired output rate to auto, upper limit to, in your case 192 kHz and check the box in Options called âauto rate familyâ
You are using HQPOS correct? in my case I am on Ubuntu Focal but using the JL kernels.
No, I use the embedded on Ravenna OS, 4.22. I went from same but 4.19 to 4.22 in one step. What really puzzels me is that on 4.19 I did not continue using Sinc-S, because I thought it was to âgreyâ, it did not display the texture of the musical fabric. Almost as I photographed an old man in black & white, pressed both exposure and development to get the crisp contrast in the lines and wrinkles. Then with Sinc-S it was like an airbrusch artist painted everything smooth and shiny, everything my music is not. But now, the Sinc-S is better in comparison to poly-sinc-EXT2. Still the overall feel is a slight 0.5-1 db roll of from about 2 kHz and up with a Q value in the area of 0.4-0.5 with this new 4.22. Some cymbal clicks is now hidden in the normal listening volume and I have to turn up to hear. I have no idea why. I am not hacker material on the linux domain so I donât know s**t about it, hence the Embedded that suits my knowledge fine. But I have very sensitive ears to subtle changes in those tiny parameters that forms the musical fabric. People have asked how on earth I can spot cable changes with same cable but another length etc in blind tests, but canât really say how, I just feel the difference, hence my profession as audio cable designer and mains cleaning equipment.
This mystery box, the computer and HQP, is going to become what different tubes are for them enthusiasts âŠ
If your DAC is not one of the Merging devices (Anubis, Hapi, Horus or NADAC), you certainly shouldnât use the RAVENNA image version, but instead the regular one without RAVENNA support.
Hi Jussi, if you around have a quick question, I am sending a friend a small PC with HQPe, if he likes it he will go ahead and purchase the license
His DAC only does PCM up to 96 khz
He sent me this âDAC is R2R (using the NOS philips TDA-1543 chipâ
Since he will be using PCM to 96 khz max
Should he go with NS5? instead any of the other ditherings?
Should he set the DAC bits to 19, since the DAC is 24 bits but R2R?
Usually Iâm not that certain to use with these DACs
Thanks
Jussi, this is the link to what I am using; https://www.signalyst.eu/bins/hqplayerd/images/hqplayer-embedded-ravenna-4.22.0-x64.7z
Should I use this instead: https://www.signalyst.eu/bins/hqplayerd/images/hqplayer-embedded-4.22.0-x64.7z
I use it from an USB stick in a normal PC, just shift boot order.
The outbut by USB is lead to a USB â AES3 converter also containing a XMOS chip. The AES3 is daisy chained between the monitors and end point is the sub. Every monitor split their part from the signal respectively depending on a dip switch choice made at the time of setup.
The monitors are JBL LSR 4328P and JBL LSR 4312SP sub. I do not use any room correction or EQ inside, there is actually no need, the room is behaving very well, with the small acoustic fixes and placement I done. The only downstream DSP is the bass management, a x-over 4TH order linkwitz-Riley for the sub at 50 Hz.
I am all ears to receiving your recommendation on which embedded I shuld use, please
Hi:
I am new to HQP. Currently I am using poly-sinc-long-mp and I think this suits me best. I am using HQP on my lenovo i7 core and connect it with my Chord Mojo. I am using PCM 387kz limit. Any one trying this filter? Let me know what you think!
Hi. Good filter with Chord.
Also try sinc-m and sinc-s filters ! And sinc-l !
sinc-m takes some seconds to start (initialise) but itâs great.
Use LNS15 for dither
Yes, I already tried sinc-M (great filter for great treble and nice cut-off make the piano sounds great), sinc-L is my go to for the Scorpion âHolidayâ song with the guitar.
I am trying and get clean sound from NS5 dither. I have used LNS15 before, but let me try it again to compare with NS5. Thank you for your suggestion.
1 Like
No worries. Why donât you try to upsample to PCM768kHz?
Is it because you get the classic Mojo white noise burst 
This is only for Merging network converters that use RAVENNA protocol, do not use it for other cases.
Yes, this is what should be used in all other cases.
Just remember to set âDAC Bitsâ to 24 in HQPlayer. And in your case use either TPDF or Gauss1 dither.
This is especially important for such XMOS based converters (AES or S/PDIF), because XMOS uses 32-bit data format over USB while AES and S/PDIF can transfer only 24 bits. If there is data incorrectly sent as 32-bit, the lowest 8 bits get just truncated (thrown away) causing distortion. When DAC Bits is set to 24 in HQPlayer, the lowest 8 bits of 32-bit word are empty and throwing those away doesnât have effect.
2 Likes