Thanks for the reply. I am glad it actually works.
I did download it twice, but maybe I used the same file twice.
I will give it a go again. But the one that I have working is working well right now so I will do some playing around with it first.
The downside of hi-res going mainstream (Apple Music, Amazon Music HD) is the increase in fake hi-res.
But one upside Iāve been noticing more and more each week from the big labels, is an increase in (real) 48kHz albums too, from big labels.
From what I analyse each week, the good news is the number of 48kHz albums is much higher than fake hi-res from big labels.
1 Like
sorry about distracting the discussion, but I am in need of support desperately.
Does anyone know why my HQPlayer embedded is not saving the selected .WAV files on Convolution page after clicking the āApplyā button?
I made some updates in this topic:
https://community.roonlabs.com/t/stuttering-during-dsd-playback-with-peq-filter/200500/10
Based on my recent tests, would you give some insights about the possible solutions? or atleast the possible sources of the problem?
I am running out of idea to carry on the tiral-&error.
It could be also related to the NIC in question, and how much offload features and such it has.
1 Like
I suspect that you are just running out of CPU power. What is the source format when you are getting drop-outs?
I have Small Green Computer sonicTransporter i7 with HQP 4.30.3 is this the most recent version? To update to a newer version it is done from the sonicTransporter web interface. I did the update to 4.30.3 in March but I donāt recall how. Anyone with a sonicTransporter know how?
Since upgrading to ubuntu 22.04 (upgrade went smoothly), I am noticing a new behaviour of HQPe.
On Ubunbu 20.04, my CPU (i9-10900k) was never powerful enough to 4xDSD, ASDM7EC, gauss-xla when playing 192khz content. The music would play, but stutter.
When testing this setup on Ubuntu 22.04, to check if the stuttering would occur, I get a message saying āroon lost control of the audio deviceā, and then my NAA (rasp4 running ubuntu) loses connection and has to be rebooted.
Is this expected behaviour?
Have you also updated the HQPlayer packages to the 22.04 (jammy) packages?
I installed v. 4.32.1avx2. Are there other packages to update?
Also remember to update the libgmpris. Everything else should get updated automatically.
If you have Nvidia drivers for CUDA, also remember to switch to the correct driver branch. But you are sort of forced to do this anyway.
Thanks. No Nvidia being used. I will update de libgmpris.
Interesting. Iāve not tried Audeze for many years, I forget the model.
I saw he shared FIR files for people to try, for AKG K371 so I tried them.
When I compare to my dead neutral Genelec speakers 1m from my ears, oratory is easily the most neutral sounding to me. Same tonality as my Genelecs which are my reference for neutral āout of the boxā. So Harman curve EQ for my AKG works nicely to achieve the goal of sounding like neutral speakers in a typical room.
oratory1990 shares some ways to adjust bass or treble to preference (the low and high shelf filters), so you donāt have to be stuck with just Harman (although Harmanās user preference findings were actually a plot range, not just a single plot).
But as you say, maybe different for different model headphones.
Check out the article I linked to, it does a better job of explaining it than I ever could. Its certainly different for different headphones as well. LCD-5 are more reference, where I doubt I would consider this route for an LCD-4 which has more of the Audeze house sound. Also as a way to EQ Convolution is superior (Mitch uses 32K points vs most PEQās Iāve tried have 10-11, or as little as 5. PEQ is just going to be an approximation at best). The nice thing about tuning these dead neutral is then you can throw a bass shelf on top of it to tune to preference or bring down the midrange or treble if you need to via PEQ. It requires much less fine tuning. Mitch is an audio engineer and does room correction as well so he knows his stuff (also his filters are based on multiple samples of headphones, Audeze always has variation between units so this becomes less of an issue)
If you havenāt heard Audeze in years, they are quite different now.
I know of him well - I bought his book to learn how to use Acurate for my DIY DSP active speakers 
Now I use Audiolense XO for DSP crossover design and speaker/room correction too for those speakers. And for room correction for my Genelecs (donāt need speaker correction there!)
Heās a really smart and nice dude.
I understand his points about 32k tap filter (I use 65k tap filter myself with my speakers). High taps is most useful for correction at bass frequencies - the main issue for speakers in rooms.
However (this is just a general comment, nothing to do with his headphone measurement) 32k taps applied to a āwrongā target will still sound crap to a simple 10-band āsimpleā IIR EQ with ācorrectā target.
When I say wrong and correct, I am only referring to personal preferences here. Not wrong/correct in technical sense.
My ears do like Harman curve EQ for headphones (my AKG K371, Focal Utopia, MrSpeakers Aeon2 , Airpods Max). Mitch says in that article heās not a fan of Harman curve EQ - thatās ok. Personal preferences. But that also automatically means I wonāt like EQ files he makes.
And again, I always compare to my Genelecs 1m from my ears (with sub under desk, everything time aligned) . If the EQ makes it deviate away from that, I canāt call it neautral.
I will give LCD-5 a demo !
1 Like
All very valid points, I certainly wouldnāt drop this kind of cash if was any old convolution filters but so far the response has been overwhelming positive at head-fi (which is a miracle in itself). Most common takeaway is tuned using his convolution filter the LCD-5 is the most detailed anyone has heard (in a headphone that is already ranked as one of the most detailed out there). Audeze is coming out with their Roon Preset soon (Reveal+ plugin - from Bob Katz / Gary Cole) and the FR is very flat, so not unlike the convolution filter Mitch has put together.
1 Like
Iāve also tried all kinds of approaches for my most used headphone sets:
- T+A Solitaire P
- Sennheiser HD800
- Shure SRH-1540
- AKG K601
Two first ones are my main HQPlayer development tools. And Shure is used the most on travel with iFi xDSD Gryphon.
And at the moment Iāve settled to using Oratory 10-band EQ typed in from the PDFs.
1 Like
And oratory measures group delay, distortions etc, which are all factored into his EQ settings. Not just frequency response.
And oratory uses the same ~ US$15k Gras 45CA that Dr Sean Olive (Harman) has in his lab and at home.
From oratoryās PDFs, I can adjust the low shelf filter to adjust bass.
I have 3 different Matrix Profiles in HQPlayer for each headphones and I can adjust on the fly which is cool. I use neutral, -2dB bass down, -4dB bass down. I rarely have to adjust from oratoryās neutral but itās nice to have those kinds of tone controls, on the fly.
2 Likes
Im having a small problem with playback. I am using HQP embedded (obviously!) and RopieeeXL (latest version) as the NAA. Occasionally (say twice per day), playback stops and I see in the HQP logs: ā2022/05/25 09:14:36 NAA output clNetEngine::Execute(): clSocket::Send(): send(): Connection reset by peerā. Its easy to recover from - I just apply the changes in the web ui for HQP and I can restart playback, but it is a bit annoying. Any ideas?
thanks in advance
I would try the custom NAA image for the pi4. See if your problem goes away, or not. Also ensure your network setup is good. I use a Cisco switch with HQPlayerEmbedded and my UltraRendu NAA each connected to it with Cat 6 ethernet cables, solid and problem free.