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

[Fernando_Pereira]

@R_C2

Now, based on the cited post, it seems that high frequency ringing is a thing.

The term “ringing” means very different things in audiology (study of hearing) and in filter design. My left ear “rings” at around 2kHz (long-term tinnitus most likely caused by a very serious respiratory infection acquired in the Andes in 2007), but that has nothing to do with the high frequency, low-amplitude “ringing” that precedes an impulse in linear phase filters (see this good explanation). One reason some folks prefer minimum phase filters is that they lack pre-impulse “ringing” and curiously, your Roon settings use a minimum phase filter.

Some people prefer minimum phase, some linear phase. Neither (as far as I know from the literature I’ve come across) causes tinnitus. However – and this I know from personal experience – tinnitus is highly variable, triggered by particular sound mixes (and by your level of tiredness, stress, what you ate, caffeine, alcohol, …), and yes, different filters lead to different sound mixes. Headphone listening is especially tricky, because it may “feel” not loud but actually push a lot of energy into the inner ear at some frequencies. This book is informative and sobering on how we mess up our hearing.

[IgorSki]

@Fernando_Pereira

some folks prefer minimum phase filters is that they lack pre-impulse “ringing” and curiously, your Roon settings use a minimum phase filter.

yep, this is exactly what I have pointed out in my earlier post that @R_C2 Roon setting is “minimum phase” filter.

Once again personal non-scientific observation - I have tried several ESS based devices and experience was not always the same. For example - awesome Perreaux Audio DP32 DAC was “ringing” for me with HQP however was brillant with Roon (with linear phase). Note, it could not do more than PCM 24/192. And I loved that one on Roon. On contrary, Topping 50S was very receptive to HQPlayer in my set up. I really liked it. But these two DAC were connected to different poweramps, which also plays a role. I tried few other ESS’s they were all different…

@R_C2

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.

There are several ways available to change the settings. It is possible to change settings “on the fily” - I personally use HQPDControl application, running on my iPad or iPhone (note, I use HQPlayer Embedded). There are few other ways. Some are truly creative and brilliant, like this one here: Commanding HQ Player… The fun way - Audio Gear Talk / HQ Player - Roon Labs Community

[Jeffrey_Glacken]

Maybe I missed it, but have you had a hearing test lately? This may show substantial hearing loss in one or both ears. Tinnitus as I understand it, is the brain trying to fill in the frequencies that are not heard well or at all.
I am not suggesting that your system is not producing frequencies that are triggering
More ringing. Could be both issues.

[Daniel_Margoliash]

Respectfully, tinnitus has many physiological mechanisms that can involve peripheral and/or central nervous system origins, or non-neuronal sources that affect the cochlea. You can’t just guess. But JG’s suggestion is spot on, R_C2 get your hearing tested. You said hearing maybe impaired down to 18 kHz (but I scanned the topic so quickly); that sounds to me like you are guessing (and hoping). In fact most adults have far poorer frequency response than that (and still get by fine day-to-day, even enjoying music). If you have presbycusis (age induced hearing loss) or other forms of hearing loss there can be effects both on frequency response and amplitude accommodation. And yes, tinnitus is serious and should be diagnosed if possible. This is an important feature of health that is mostly ignored by the public and even to some degree health professions, and in your case might help focus your attention on possible sources (e.g. lower frequencies than you image that you cannot hear, sudden attack in music - does this happen more in chamber music?).

BTW, I am not pushing an agenda and I am not an audiologist! (And yes, I do love classical music.) But I am a neuroscientist who works in part in audition.

Best of luck!

[R_C2 (R C)]

Thanks for your concerns everyone. Indeed the title of the thread refers to tinnitus, albeit VERY mild. Like everyone in the roon/HQP thread I care about sound, listen carefully and, just to be sure, carry foam earplugs in my pockets at all times: in case of roadworks, long journeys or even a party. So I was shocked to discover something I couldn’t hear - but could sort of feel (a tingling) - was causing mild tinnitus, noticeable the following day on waking. Wished to investigate that here.

As has been pointed out 7th order Modulators are contraindicated for ESS Sabre DACs. At the time of (skim) reading the manual had no idea that was what was in a Merging Anubis. Merging don’t advertise that but later gleaned it from a forum via google. This is a micro-example of how involvement with HQP requires a certain level of proficiency in a variety of domains. The calibre of the contributions in this thread is further testament to that.

Having played around some more it seems that even when using ASDM5v3 or ASDM5-light (appropriate for ESS Sabre) with an -mp filter (poly-sinc-gauss-hires-mp - to minimise ringing in the analogue signal) my ears can begin to tingle (the first signs of discomfort) at certain moments. ** What are those certain moments? It might be a soprano delivering a sustained note, an interplay of voices, the woodwind section of an orchestra… (Latest tests were with 44.1kHz/24bit Monteverdi: Il pianto della Madonna - La compagnia del Madrigale. Doesn’t get tamer than that!) Note that I adjust the volume of my amp as low as I can whilst still being able to appreciate the piece. If I play the same piece using Roon’s Muse to convert to DSD256 (with 5th order Clans, smooth linear filter) there is NO PROBLEM. Except that it doesn’t sound as good! Nor is there any problem if I use HQP in PCM mode for PCM. So that’s what I’ll do. To my mind that suggests the blame lies in the interaction between the ESS Sabre DAC and HQP’s modulators (those that I’ve tried).

Perhaps my next step should be to audition a Holo Spring? I’ll be compute constrained at that point, but what the heck.

** The choice of appropriate modulator and -mp filter BOTH HELP REDUCE the “ear tingling problem” (which leads to mild tinnitus). Just not enough. Which suggests that the acoustic phenomenon leading to auditory overload DOES stem from Filter ringing and that is somehow exacerbated by SDM Modulation (worse in the case of 7th order, still problematic with 5th) when passed through the DAC of a Merging Anubis. And perhaps electrostatic headphones are a factor. That’s the hypothesis to test!

[Chunhao_Lee]

@R_C2

Having played around some more it seems that even when using ASDM5v3 or ASDM5-light (appropriate for ESS Sabre) with an -mp filter (poly-sinc-gauss-hires-mp - to minimise ringing in the analogue signal) my ears can begin to tingle (the first signs of discomfort) at certain moments.

Based on your Roon’s upsamping experience, have you tried poly-sinc-mp3/mqa-mp? 5EC-light with this filter is my everyday combo. It has the shortest post-ringing among the HQP’s filters and also has the earliest roll-off starting point.

[dabassgoesboomboom]

@R_C2

Perhaps my next step should be to audition a Holo Spring? I’ll be compute constrained at that point

If you can home demo for no cost, definitely yes try. Same filter and modulator will not need more compute

[Marin_Weigel (Bram Stoker's Cappuccino)]

@R_C2

Perhaps my next step should be to …

As others said, getting your hearing checked is No.1

Secondly, as I have a strong feeling that contributions of different upsampling/filtering techniques are usually at or below the auditory threshold and thus rather unlikely to trigger such discomfort, I‘d recommend to get your Stax amp and earphones thoroughly checked by a competent service partner, to verify that they‘re not driven into oscillation by some, even mild in absolute terms, high frequency noise content.

Good luck and take care of your hearing.

[Jack_Stark (Jack S)]

I already have moderate Tinnitus. My tinnitus is in part due to high frequency hearing loss in the vicinity of 2KHz, but my high frequency hearing returns to some extent at 4KHz and higher (beyond the range of most hearing tests, they do not test the higher frequencies, the claim is there is no baseline at those frequencies to compare to ‘what is normal’). I find that both the fidelity and frequency content of music I listen to will make the ringing worse (bilateral, single frequency). Your hearing ‘system’ can develop an adaptation to hearing loss called ‘recruitment’. This causes a band of frequencies in your hearing to have increased perceived loudness, usually near the ringing frequency. I have this condition, it is not hyperacusis - which is very rare. If I listen to music that is in the ‘recruitment’ range, it can be uncomfortable and exacerbate the ringing. You may be experiencing a mild form of this condition, as I sincerely doubt it is due to something your ears are being exposed outside the high frequencies of your hearing range. I find that ‘low fi’ is especially objectionable to my ears, even at low volume. I have invested in premium headphones for listening, and this definitely helps. I also tried notch filtering out the ‘recruitment’ band with only mild benefit - if any.

[R_C2 (R C)]

That’s very interesting. Your case. It’s certainly true that I have an intolerance for loud sounds that I don’t remember having before. The thesis that there is some kind of psychological amplification going on makes sense. Whether that or tinnitus (the psychological failure to adapt to the absence of an input signal from a now-dead or otherwise disconnected sensor cell) - makes me immediately wonder how one can hack the system. Some things right themselves directly. Allegedly if one puts prism glasses on that turn the whole world upside down, maybe 10 minutes later the brain suddenly flips everything the right way around. Rather terrifyingly there is a similar delay when you take them off! Some things are trickier. Take “phantom limb syndrome”, for example, where an amputated limb continues to give pain. Some guy found that using a mirror to create the illusion that the arm was still there enabled the sufferer to “move” it into a more comfortable position thus relieving the pain.

In the meantime I’ve done more tests - including the legendary poly-sinc-mqa-mp3-mp filter! - to no avail. I’m wondering if the Merging Anubis - designed as a recording / studio tool which even allows outputs to be routed into inputs and all sorts of clever things - is in fact missing some kind of basic low pass filter that I guess would be a part of a consumer DAC designed for listening? Unfortunately am travelling and don’t have the USB cable to connect to the DAC in the srm-d50 STAX amplifier (it’s also an ESS - a 9018) to see if it performs differently with an SDM stream from HQP.

Back on the ultrasonic possibility it turns out there’s also a small percentage of the population who are sensitive and suffer in places in the modern world where ultrasonics are present “invisibly” (including public address systems, it seems). https://royalsocietypublishing.org/doi/10.1098/rspa.2015.0624

Have also become aware (online) of the Holo Cyan 2. It’s a perfect match (quality/stripped down functionality/portability/HQP proven-ness) at a price point that might make ordering it without even trying not a bad thing to do. If with that I still can’t bear - amazing as it sounds - upsampled SDM, well at least I’ll have a great laddered PCM experience!

[jussi_laako (Jussi Laako (Signalyst))]

@Chunhao_Lee

As you can see on Merging side the DACs output include these ultra sonic noise.

It is ESS chip, so it has very gentle filter. Thus 5th order modulators are recommended.

@R_C2

But the question remains, might there be a technical solution to prevent sound pressure intense enough to induce tinnitus?

This random noise that is inherent to any DSD (or SDM DAC in general) is not usually as much problem in this respect. But rather lower frequencies around 22.05 kHz (Nyquist of RedBook content) or 24 kHz (Nyquist of 48k content). Plus other filter effects such as ringing around those 22.05/24 kHz frequencies that happen more, longer the filter is.

So I would rather look for another filter, rather than modulator. Try for example with poly-sinc-short-mp-2s for a comparison. But with ESS based DAC, best getting started with 5th order modulator.

For me, listening fatigue is mostly related to filter choice.

[R_C2 (R C)]

Wow.
I guess @Chunhao_Lee I should have asked what the “ultrasonic noise floor” looks like for a “normal audiophile DAC” for comparison. Does it still include that massive peak at, I guess, 22.05kHz? Do I take it that the “noise floor” is just the inherent background noise present in the system before one even starts playing a track? Do we infer that it is indicative of where noise will be found (or even amplified by ringing/resonance/wave superposition?) once music playback is underway?

In the ESS9026 datasheet there seem to be lots of filtering options. Perhaps in the Anubis Merging has chosen not to turn them all on!!

[jussi_laako (Jussi Laako (Signalyst))]

You have two types of ultrasonic content appearing in DAC outputs. Modulator noise which appears on SDM converter outputs and it is static (should be) regardless of your signal. Similar output appears at much much higher level on output of any class-D amplifier. This is uncorrelated noise, sounding like a tape hiss. So it is not related to the music content. It is there same way even if you listen to a track containing silence. On a generally good DSD DAC, the level is not higher than inherent audio-band noise floor of RedBook (44.1/16) PCM. On better ones it is much lower (beyond what you can find as inherent audio band noise floor on best hires recordings). Whether these exist on audio output of your system depends on what you use for listening. Generally none of the headphones or speakers can reproduce frequencies beyond 100 kHz at sufficient levels.

Then you have correlated ultrasonic images, that are result of PCM sampling, appearing around multiples of the source sampling rate. Level of these depend on oversampling ratio and attenuation of the oversampling filters. These sound more like harmonic and intermodulation distortion and are fully correlated with the source signal. If you listen to silence, they don’t exist, but immediately when you have some source signal, these appear. Choice of filter affect these.

Intermodulation products of uncorrelated noise sounds like a tape hiss, or radio background noise. While intermodulation products of correlated images/errors sound like distortion.

In addition, choice of filter affects things like transient speed, ringing around the source Nyquist frequency, etc. And choice of modulator affects things like transient speed, reconstruction quality, etc.

Plus, then in the source content you can have errors that can certainly cause listening fatigue. Some of these can be corrected with apodizing filters. HQPlayer as indicator for these kind of errors. The “Apod” counter that increments during playback whenever such are detected. Most of these errors live in 16 - 22.05/24 kHz region. This is provided to as guidance for selecting suitable oversampling filter.

In addition, ESS chips don’t have DSD Direct path (to bypass on-chip DSP), so any input, PCM or DSD, will always go through the ESS’ on-chip DSP. Just different path depending on PCM or DSD input.

Main difference I notice in your settings is minimum-phase vs linear-phase filter choice. CLANS modulator creates some amount of correlated ultrasonic “whisper”, but that shouldn’t be the issue here. So I would suggest first going to check poly-sinc-short-mp-2s for comparison. And for more similar “softer” Roon like character, you can try ASDM7 or ASDM5.

@R_C2

In the ESS9026 datasheet there seem to be lots of filtering options. Perhaps in the Anubis Merging has chosen not to turn them all on!!

Those apply to PCM inputs. For DSD inputs, it has choice of (gentle) 50/60/70 kHz filter.

[jussi_laako (Jussi Laako (Signalyst))]

Of course also keep eye on your volume, that you don’t end up listening louder!

[Chunhao_Lee]

@R_C2

I should have asked what the “ultrasonic noise floor” looks like for a “normal audiophile DAC” for comparison.

Before answer this, I’d like to know that since the Stax SRM-D50 headphone amp is RCA input only for analog signal, had you checked your cable wiring? Balanced XLR to RCA cable sometime I saw it had very high faulty rate due to incorrect wiring… And also suggest to try SRM-D50’s built-in DAC + HQP upsampling?

[R_C2 (R C)]

The cables are well made - Mogami / Enoaudio (Berlin) https://enoaudio.de/en/product/m2534bk-2rm3xf-p/ (They replaced a pair I bought on amazon to test the concept: the improvement in clarity and detail was extraordinary!)

Eager to try HQP upsampling with the integrated DAC but am travelling and didn’t bring the appropriate USB B cable.

Oh, this is what the 50kHz/60kHz/70kHz filter for DSD looks like from the datasheet for the chip in the Anubis. It seems as though 50kHz actually means something like “-4dB at 50kHz”. Not a hill of beans.

[R_C2 (R C)]

By way of contrast, for the filtering, found this for the first generation Holo Cyan. The note on the graph says “slow” (from 20kHz, I guess) but, beyond 22.05kHz it drops off extremely rapidly. Now I understand why the ESS is “gradual”. I wonder what the Cyan 2 does?

[Ron_Klein (Ron K)]

It sound to me you need an audiologist and some hearing testing.

[Chunhao_Lee]

@R_C2

Does it still include that massive peak at, I guess, 22.05kHz?

The strong peak you saw that was 19KHz & 20KHz two-tone test main signal. I prefer use that to see the severity of IMD. Stereo DAC I have is iFi micro iDSD BL. It’s TI/BB chip inside (old but good). Although 7th order works well on that but iFi’s FAQ mentioned using 5th order is better. Those ultra sonic noise are included in the RCA output.
Here’s the same 19&20KHz two-tone test, 5ec-light:

And here’s 7ec-light:

The micro iDSD’s ultra sonic noise floor looks better than Merging Anubis. But however ultra sonic noise from Merging DACs are still can be cut clean by analog LPF by analog side. So my common sense tells me your problem might caused by STAX headphone amp’s LPF design. If you have another headphone, it’s worth to try Anubis’ headphone amp and verify your ear ringing issue.

[R_C2 (R C)]

Many thanks. It’s all very interesting. I certainly have some testing to do.