Going by my own experience, I think that this will be your DAC switching from PCM to DSD. See if it happens when you’re only using PCM (if that’s your DAC’s default format). If this is indeed the cause, turning down the volume a bit on your DAC should lower the volume of the pops also.
1 Like
I think you’re right. I recognised similar biases/assumptions in myself when I first started using HQPlayer. Even though I didn’t have the Chord sound as a reference, somehow it had also been instilled in me that that was supposed to be the “industry standard” and going by one reviewer’s comments about Sinc-M being virtually identical to Chord MScaler (in terms of how it sounds, technically we know they are different and Sinc-M is apodising), I right off the bat used that filter as the one to compare everything else to. Until I heard the poly-sinc-gauss-long and experienced a “cognitive dissonance” (hehe) - music suddenly sounded so much more natural.
I must say, though, that it has been a bit of a journey for me too and when one’s ears are used to certain aspects of sound being in a certain way, it can take a while to fully appreciate a change. As is usually the case in this hobby, it’s a matter of living with the new setting for a while, then going back… only to discover that there is no going back because now it is obvious what we’d be missing.
At the moment, I am experimenting with Qobuz’s high-res files and am in awe of the poly-sinc-gauss-hires-lp filter. I remember having some reservations about it before, but that was with a different modulator. The 7EC-super puts everything how it should be.
Whilst there’s always a bit of a trade-off for me between the gauss-long and -xla filters with Redbook files (the former sounds even more natural than the -xla, the latter adds space which sometimes makes a significant difference), so far I think that this trade-off is almost non-existent when listening to high-res files using this filter. I’m hearing beautiful holography and space around each sound, they’re allowed to “breathe” (as opposed to being mere “pressure points”, as I like to call the effect that very long filters seem to create), everything is very natural and effortless (timing/transients), at the same time I don’t feel like I’m missing anything, there’s seemingly no price to pay for that (unless one prefers extra punch/“pressure” over natural delivery).
Just a few high-res tracks to check out:
The Doobie Brothers - Long Train Runnin’ (2016 remaster)
AC/DC - Girls Got Rhythm & Thunderstruck
Daft Punk - Lose Yourself to Dance
The White Stripes - Seven Nation Army
Eagles - Take it Easy
Run-D.M.C. - It’s Tricky & Walk This Way
Metallica - Enter Sandman, Whiskey in the Jar
Tool - Chocolate Chip Factory
And many, many more, of course. It’s just the ones I’ve listened to today, and switching between -xla and hires-lp makes a HUGE difference. A jaw-dropping difference, actually. I think that it’s a very interesting comparison as I understand both filters to be the same except for their length - so far it’s been the most immediately-apparent demonstration of what the filter’s length does to sound reproduction (if I understand correctly, with Redbook we’re playing a game of important trade-offs, but with high-res files we’re getting much closer to having the cake and eating it too).
3 Likes
Since the DAC doesn’t seem to mute properly all DSD transitions, you can try to minimize those by unchecking “Adaptive output rate” and setting output rate to “44.1 x128”.
Roon will still do a lot of unnecessary stop/start actions, but at least less DAC clock transitions.
Hah! My first reaction after listening to his full review was “this tells me more about you and the inherent limitations you are imposing on all your reviews than it tells me about HQ Player…”
If you have the same hearing and tastes as Rob Watts, and you have the same components and listening room as Rob does, then by all means just buy the Chord m-scaler and DAC. Most of us have none of the above. That leaves us with essentially three choices: 1) just go buy the stuff a reviwer tels you is best and tell yourself you love it and be happy (easiest choice); 2) go audition a lot DACs and hopefully listen to them attached to other equipment like yoursin a room like yours, or better yet borrow, or buy lots of DACs and try them at home and then pick the one you like best (more work and potentially more expense), or 3) buy the best NOS DAC you can afford, buy HQ Player and buy or have a fast computer and sit back and start listening to filter choices (lots of enjoyable work listening to all that music with different choices). What I love about choice 3 is that the more effort you put into it the more reward you get! Picking the best choices for you isn’t easy, but if you have patience you will discover why so many of us are diehard fans and supporters of Jussi!
3 Likes
I went back to ASDM7ECv3 recently. I had been using 7EC-super for a while now. IMO the biggest difference between these two is that -super offers slightly better depth and soundstage but v3 has better punch and ”snappiness”. It’s similar to how, on filter side, poly-sinc-gauss-long compares to -xla.
2 Likes
We must have very similar hearing as I remember thinking about the differences between those filters too when I was comparing the two modulators.
I think I’d disagree in relation to “punch”. Bigger-sounding modulators and filters tend to sound punchier to my ears (sometimes overly so) and the -super modulator is no exception here. To me -super sounds like a great trade-off between v3 (the wispiest but also the most direct/intimate) and the 512+ (sorry, can’t remember the exact name just now - the most distant and punchiest sounding, to the point that it’s a bit bloated; although I’ve only tried it with DSD512).
I do, however, agree that V3’s snappiness/directness is the best, akin to what shorter filters seem to excel at. At the moment it’s the -super modulator that’s caught my attention and stopped me wanting to switch back and forth.
2 Likes
I’m finding it fascinating to compare high-res vs 44.1kHz versions of the same albums (such as Radiohead’s 2017 release of “OK computer” or Coldplay’s “Parachutes”) and applying gauss-hires-lp to the former and either gauss-long or -xla to the latter.
It kind of only confirms what the theory would suggest but it still surprises me how obvious the differences are during longer listening sessions.
Gauss-hires-lp with hi-res - it has it all. Both intimacy and space. Both body and snappiness. Both overall soundstage size and lots of air around each sound. Both micro and macro impact. Can’t get enough of it.
Gauss-long on 44.1kHz - it has some of the above qualities, it gets quite close, I must say, but it doesn’t have the same space. Space aside, it achieves similar outcomes, the DNA is there. It just can’t fully spread its wings, I suppose due to needing to be a longer filter to fully attenuate at Nyquist.
Gauss-xla - as much as I love this filter, it gets further away from the -hires one than Gauss-long does. It brings the missing space back but at the cost of less focus being given to each individual sound, macro now trumps micro. The focus is shifted to the overall performance as opposed to each sound being an event in its own right.
Switching between -long and -xla to me is clearly choosing between which trade-off works better for me (sometimes track-dependent). Such is the nature of Redbook, perhaps.
However, going high-res and applying the -hires filter elevates the experience to completely new levels. To the point where I now get sad if a record I love is not available in 88.2kHz or higher.
1 Like
Based on this, have your tried gauss or gauss-short instead for RedBook? Maybe worth checking!
Thanks for the suggestion, I’ve briefly tried them before and they seemed to have even less space than Gauss-long, which would make sense, and that was a trade-off I wasn’t prepared to make as a sense of space is needed for realism.
I actually often enjoy switching between -long and -xla and experiencing the same song from two different perspectives. But from now on, if a high-res version is available, it’ll be hires-lp for me all the way. It’s like another (big) step towards tricking my mind that no digital to analogue conversion is taking place at all.
I’m wondering what might be the best filter combination for my Holo May KTE DAC.
I have music files ranging from cd thru 256dsd. Mostly Jazzy type music and some rock. I just got a Sonictransporter i9 version with imbedded HQ4
Hi. i have nuc i710fnh. on it roon with parametric equalizer and hqplayer. I have endpoint naa on raspberry and further to dac on ak4490eq.
tell me what sdm settings to use to optimize nuc work? when upsampling x256 on ext3, you can hear the artifact of denser tracks. you can see that the processor threads go to the max, even though the total efficiency is, for example, 40%.
Thx
DSD256 or DSD512 output using ASDM7ECv2.
Filter of your choice, but the default filters are a good starting point, 1x=poly-sinc-gauss-long Nx=poly-sinc-gauss-hires-lp. But there are some genre based suggestions on the Help page (see Help-link near top right corner of /config page).
1 Like
Is it with HQPlayer Embedded or Desktop? It is recommended to move the parametric EQ to HQPlayer’s matrix processor and keep Roon as bit-perfect.
Use ASDM7EC-light modulator and try for example ext2 and poly-sinc-gauss-long filters. ext3 is roughly as heavy as poly-sinc-gauss-xl(a).
1 Like
Thank you, i try it.
Thank you for the starting suggestions
@jussi_laako Can you please advise if I’m right in thinking that the gauss-long filter, when fed 48kHz content, would reconstruct the frequency range up to 20kHz, just like with 44.1kHz content, and then roll off slowly until full attenuation at 24kHz, thus effectively making it a shorter filter (more room for attenuation)?
It’s very rare that I find both 44.1 and 48 versions of the same album, but if the above is correct, then in those rare cases it would in fact make sense to opt for the 48 version.
For 48k content the corner is roughly 2 kHz higher.
For 44.1k vs 48k version of the same content, the critical question is what is the original recording format. You’d want to have minimal amount of conversions at the source side.
Would it be a mismatch when running 96 or 192 files in Roon? When running all files fixed to 44.1 - 128dsd?
No, that is totally fine and with most DSD capable DACs on the market you need to use 44.1k DSD rate multiples.
1 Like