Something I can't hear is making my ears ring (not good)

[R_C2 (R C)]

I’m halfway through the HQPlayer trial and enthusiastic about the listening improvements. However I’ve noticed that, with the listening volume quite modest (I listen to classical attentively, not loudly) from time to time my ears are telling me “it’s too loud!” Yet the volume is already set to the lowest level I would want to go. When I wake up in the morning I’m aware that my ears are SLIGHTLY ringing (as if had been to a loud concert). That’s not good. I think that whatever is going on is at a frequency outside my hearing range, otherwise surely I’d perceive the music as being too loud.

Did notice the bit in the manual about inter-sample overloads and have limited software volume to -3db. But I’m wondering if maybe there’s something inter-sample going on that’s far more extreme? Or another explanation? What am I doing wrong?!

Please find SDM settings below (everything from Roon is being converted to DSD 256. My DAC is a Merging Anubis SPS Premium, which I believe uses a 32bit ES9026PRO ( https://eu.mouser.com/datasheet/2/1082/ES9026PRO_Datasheet_v3_5-3074412.pdf ) and, since I’m MacBook Pro M1, everything is DoP.

By comparison when I run a similar conversion to DSD using the Roon MUSE do not have the ears ringing problem (settings also below).

[Chunhao_Lee]

@R_C2

But I’m wondering if maybe there’s something inter-sample going on that’s far more extreme? Or another explanation? What am I doing wrong?!

Have you tried other modulators like ASDM5EC-light, ASDM5EC-ul or ASDM5EC? In my experience ASDM5EC-light is the best modulator for Anubis Premium / DA8P / ADA8P / NADAC. ASDM7EC-super has some strange result showed on my PicoScope (maybe my setup was incorrect) so I personally won’t use it for Merging DACs.

[R_C2 (R C)]

There was a 5 series modulator as default and so that’s where I started. It seemed to beat Roon’ MUSE series 5. Then went up to series 7, picking what I took to be the most processor hungry one since my computer seemed to be able to handle it. I thought it sounded better than the Series 5. It’s just this weird invisible-noise problem… I’ll wait and see if other suggestions come in and then go back to Series 5 if necessary. Happy to meet another Merging user. Thanks!

[Chunhao_Lee]

@R_C2

It’s just this weird invisible-noise problem…

Higher order modulator doesn’t mean ‘better’. Still need to see DAC chip’s characteristics. For Merging DACs with ESS chip, better go with 5th order modulator. Here’s the screenshot from my PicoScope, the filter was poly-sinc-gauss-long and the test signal was REW 24/48 two-tone (19KHz+20KHz) for checking the IMD:

As you can see the noise floor of 5EC-light is lower than 7EC-super, and also the signal strength of the IMD’s side band is much less than 7EC-super.

In fact, if you want the lowest noise floor and the minimal IMD on Merging’s DACs, better go with DXD.

–Edited–
I forgot to mention if you use AES67 sends 24/48 audio stream to Anubis directly, same two-tone test 19&20KHz you’ll see fairly strong IMD:

So upsampling to DXD thru HQPlayer is still very helpful for sound qualities.

[R_C2 (R C)]

That’s interesting. You are implying that I might have been victim of some high frequency noise that I couldn’t hear? Perhaps in the region of 20kHz? Makes sense, though looking at the graphs the difference in attenuation between the 7 and the 5 is hardly enormous. Was thinking that perhaps it had more to do with noise creation by the 7.

In fact I preferred the 7 super over the 5 (except for the problem). And the 7th order seems to work fine on the Roon (see Roon config above) - just not as well as HQPlayer

Perhaps it’s a bug that was never fixed because Signalyst didn’t have a Merging to test it on?

BTW what’s the difference between a “light” and a “super”? I prefer the 5-super to the 5-light: was thinking that “super” means more computationally intensive and therefore more thorough, a better job. Not a problem on 2 channels, but can imagine that on 8 it gets a bit much.

[Outlaw (Jason P)]

When using HQPlayer turn off upsampling in Roon

[R_C2 (R C)]

Hey, you’re right. I wasn’t clear. I used either Roon-HQplayer OR Roon/MUSE-upsampling direct to DAC. Not both at the same time!

Thanks

[Chunhao_Lee]

@R_C2

I prefer the 5-super to the 5-light: was thinking that “super” means more computationally intensive and therefore more thorough, a better job.

Same thought: more computation doesn’t mean ‘better’. 5EC-super’s CPU utilization rate is just a little bit more than 5EC-light on my system (13ch DSD) but you’ll get stronger side band of IMD test:

I’ve done the regression test for different HQP modulators several times and so far the best modulator is 5EC-light on my Merging’s multichannel system by IMD’s test result perspective. You still can use your favorite modulator but just keep in mind that you can’t avoid ESS chip’s internal DSP even you feed DSD. Less computation / process in front of the ESS DAC the SQ would be better.

[IgorSki]

@R_C2

I listen to classical attentively, not loudly

few non-scientific thoughts around your qustion… (master @Chunhao_Lee , does not stop impressing )

I’ve noticed in your Roon screen you are set with - “precise, mimimum phase” roon filter. comparing to your HQPlayer screen - I’m guessing that “poly-sinc-gauss-long” is actually a linear phase. This nuance may be introducing your perceptional difference…

For the classical music (beside your personal taste) many things would depend on source material, recoring label, type of music as such… live, studio, orchestral, strings, solos, quartets, quintets, etc… Also worth to note, that PCM and DSD sources would use slighly different path (PCM: filter-modulator, DSD: integrator - converter - modulator)

To my personal taste orchestral pieces can get somewhat “busy” or “dense” with EC-Super modulator. So much more often I would prefer “EC-Light” for these. For the solo works or small collectives I would go for ECv3 (ASDM5ECv3 / ASDM7ECv3) modulators.

For DSD sources I have FIR2 / XFi, and actually never changing this…

For PCM sources here are some of my filter preferences (for 1x / Nx):
poly-sinc-short-mp - solo piano works, where I would like to accenuate stoccatos for example (more of the jazzy stuff, actually)
poly-sinc-long-mp (-2s) - also great for solo piano works, but brings you more of a volume, like with open pedal, for example for the likes of Lubomyr Melnyk

poly-sinc-ext2
poly-sinc-ext3
poly-sinc-gauss-xla
These can give more timbre or “weight”, but i rarely use ext3 lately, may be because of the same “ringing” effect as you describe, actually

sinc-Ll - I used to be amazed for strings, like Vivaldi

sinc-L - quite powerful and impressive, would never call it an “all rounder”, but at times it can have an impressive match to small or big ansambles.
sinc-long - same as above

sinc-MGa - this one I almost only use with “EC-Light”, more as “all rounder”

poly-sinc-gauss-long / poly-sinc-gauss-hires-lp is a good “all rounder” really, but sometime you can do slightly better.

members are also referreing to closed-form, this I actually never used… I wonder if I have ever tried…

PS: and yeah 5EC’s are better for ESS, may be with rare exceptions in my own experience.

[R_C2 (R C)]

Thanks. Your synthesis of personal experience is quite eye-opening and incredibly useful to an HQP newbie. I’m now checking the interface carefully to see exactly what is being applied during playback. Have not yet understood why sometimes it’s “Oversampling 1x” and other times “Oversampling Nx” that is applied during PCM conversion. Certain filters when associated with “Oversampling Nx” seem to prevent tracks from playing at all. For example I can have “sinc-LI” on 1x no problems, but on Nx playback (in Roon) hangs.

My headphones are electrostatic (Stax SR-L700 Mk2) and I’ll bet they are quite good (frequency response 7Hz – 41kHz) at some of these inaudible frequencies that seem to have been giving me discomfort. I wish there were a warning it the manual about this: I would have stopped each time my instincts told me something was wrong.

Browsing relentlessly found the following post: Which HQP Filter are you using? [2024] - #249 by jussi_laako

In which it is said that “Both, too leaky filters, and also the long filter ringing gives me “digitus”, which is listening fatigue after a while. This is usually worst when combined with metal dome tweeters that have 20+ dB resonance peak around 25 kHz or so. In worst case I get tinnitus from that.” Don’t understand all that is being said there but it seems the blame is squarely with the Filter and not the Modulator.

It seems to be totally unacceptable to be getting tinnitus (hearing damage, basically: hearing cell death) from listening at modest volumes because of high energy ultrasonic waves. Can we not find a technical solution for this?

Btw. do you have a shortcut system for importing settings according to repertoire? Or go into settings and do it manually each time? I should think that setting Preset Buttons in HQP would be a popular request.

[Stefano_Antonelli]

@R_C2

Have not yet understood why sometimes it’s “Oversampling 1x” and other times “Oversampling Nx” that is applied during PCM conversion

1x is the filter used for input file resolution up to 48k
Nx is the filter used for input file resolution > 48k

@R_C2

For example I can have “sinc-LI” on 1x no problems, but on Nx playback (in Roon) hangs.

This filter is integer therefore can be used only if you upsample to a integer multiple of the original resolution (ie 44.1 → 352.8 or 48 → 384)

To fully understand filters and modulators you should read the HQP help page (or the user guide) where they are described with their limitations

[ipeverywhere]

Don’t have a great solution for you but, yes, you’re experiencing listening fatigue and its manifesting itself as a ringing. Age will play into this as well as we naturally lose our hearing. I only mention this because I don’t faul HQPlayer here. It’s doing what it is advertised to do. How that plays (pun intended) downstream with the rest of your gear is why there are so many options and combinations.

However… you mention headphone listening… what amp are you using? What tubes are you using (if its a tube amp)? Harmonics within the amp may be a cause as well. There are some steep filters in HQPlayer that will fall off quickly if your amp isn’t handling the ultrasonics nicely.

[Fernando_Pereira]

@R_C2

“Both, too leaky filters, and also the long filter ringing gives me “digitus”, which is listening fatigue after a while. This is usually worst when combined with metal dome tweeters that have 20+ dB resonance peak around 25 kHz or so. In worst case I get tinnitus from that.” Don’t understand all that is being said there but it seems the blame is squarely with the Filter and not the Modulator."

That is in reference to other filters than the ones you were using, not specified in the paragraph but discussed later in that post, for instance filters that preserve original samples.

[Chunhao_Lee]

@R_C2

For example I can have “sinc-LI” on 1x no problems, but on Nx playback (in Roon) hangs.

Merging DACs only support up to 44.1x256 so integer ratio filters won’t work on 48 family contents. Since your system can cover ultra sonic regime, better start with gentle combinations for example slow roll-off filters with 5th order modulators on Merging DACs. Also verify your headphone amp’s analog LPF can cut clean before DSD noise comes up.

Here’s the ultra sonic noise floor from XLR output of Merging Anubis Premium showed on my PicoScope:

As you can see on Merging side the DACs output include these ultra sonic noise. If your system is very sensitive to ultra sonic noise you might consider to change your audio gears…

[Fernando_Pereira]

@R_C2

It seems to be totally unacceptable to be getting tinnitus (hearing damage, basically: hearing cell death) from listening at modest volumes because of high energy ultrasonic waves.

Where are you finding “high energy ultrasonic waves”? According to the post you linked to poly-sinc-gauss-long is “flat to 20 kHz and reaches -300 dB by 22.05 kHz” and the graphs given earlier on this thread show modulator effects down below -70dB, which is below audibility.

[R_C2 (R C)]

You are right.

My understanding is that ultrasonic is anything above 20kHz (and in my case inaudible is perhaps 18kHz or less). Something inaudible seems to cause chronic listening fatigue - and in a short period of time. Retrospectively it was when certain instruments/voices/notes kicked in and suddenly I’d feel my ears tingling and then an urge to take the headphones off. Initially I didn’t realise there was anything going on so I didn’t stop. Now, based on the cited post, it seems that high frequency ringing is a thing. Whatever the technical definitions of filters - whatever their specs may say - my ears were not lying. (And they were slightly ringing the following morning. This happened more than once.) Perhaps indeed it’s DAC-related as well (hasn’t happened with Roon upsampling). For the sensation of intense noise (although inaudible) to be so evident as to produce a noticeable reaction in me it would be as though the acoustic sound were well over 90dB. Whilst I listen rather more quietly than that. Especially classical. I guess dB is a measure of energy. By “high energy ultrasonic” I was just emphasising that, as frequency increases, so does the energy of a wave (E=hv and all of that). That was perhaps an exaggeration but based on the result. I expressed myself badly.

I guess HQP cannot know what the DAC is going to do with the data. But the question remains, might there be a technical solution to prevent sound pressure intense enough to induce tinnitus?

(Sadly I don’t have a tube amp yet. Just a Stax SRM-d50 (as amp) and an Anubis SPS Premium (as DAC, but selected as a tool for recording DSD))

[Chunhao_Lee]

@R_C2

Just a Stax SRM-d50 (as amp) and an Anubis SPS Premium

I forgot to mention… don’t use shielded ethernet cable. Use CAT5e or CAT6 UUTP for your Anubis only. Shielded cable will bring you unpleasant noise especially you’re using Stax headphone.

[R_C2 (R C)]

Am using Merging’s supplied cable (no markings). Funny I thought shielding could only be a good thing on an ethernet cable.

On reflection, regarding, ringing (or at least inaudible sound pressure) it seems that, for a given DAC:

STEP 1: choice of Filters may be important during PCM playback to avoid later ringing problems at the DAC stage;
STEP 2: choice of modulator may be important for SDM conversion in order to avoid ringing later in the DAC;

In my case the Filters were (probably) OK. The primary problem was the use of a ASDM7EC-super with a Sabre 9026 DAC. Certain notes resulted in uncomfortable inaudible sound pressure.

It seems to be more common for a poor choice of Filter to be responsible. Varying filters I may have experienced this too.

Perhaps if others contribute their stories we can get a better understanding of what can go wrong and how to avoid it. Whether by making appropriate choices during configuration (a table of what doesn’t work for various DAC) or by being aware that it’s important to stop if you feel anything odd, note the configuration used for playback and review it with a certain logic.

[Chunhao_Lee]

@R_C2

Perhaps if others contribute their stories we can get a better understanding of what can go wrong and how to avoid it.

I would say… read the manual thoroughly.

The manual already said so. Pushing to 7th order modulator on ESS is personal choice but deviating from recommendation you have to take the risk by your own. If you feel uncomfortable with 7th order just go back to manual’s then find the best combo based on recommendations.

[Rugby (aka Daniel or Big Dan)]

@R_C2

Funny I thought shielding could only be a good thing on an ethernet cable.

It is the opposite, shielded cables can introduce ground loops. They are for use in commercial settings. Home use should always, imho, be un-shielded STP cables.