[roderickvd]

I've had a small period of really enjoying listening to my Holo May in simple, un-upsampled NOS. Even though a NOS dac in Jussi's words "completely fails at reconstructing the original music signal" there's a certain allure to the sound.
However it irks me that 24 bit music files are introducing linearity errors in my music, and I miss being able to use digital volume control. I tried closed-form-M because it leaves the original samples intact and only interpolates, so I figured it could be some sort of NOS+. In the end decided that sinc-L sounds superior to my ears.

However I came across this Stereophile review of some ridiculous 50k dac:
https://www.stereophile.com/content/ch-precision-c12-da-processor-measurements
It uses an ultrashort polynomial interpolator for "perfect time domain performance". I then vaguely remember something about polynomials in the HQPlayer manual and indeed there's two polynomial filters there with either 0 or 1 cycle of pre and post ringing, appended with the comment "not recommended". Intrigued by the idea of recreating a 50k dac experience I gave them a try and.. I think i've finally found the NOS++ i've been looking for. Retains the character but adds clarity to my ears.

Anyone with a NOS dac with similar experiences? I'm really curious whether others also perceive these filters as enhanced NOS or whether i'm just placebo'ing myself

Thanks for posting this. Took me a while to respond, as I it took me a few days through the rabbit hole to try it out.

I thought it was quite thought-provoking, in a good way. Funny because the closed form/sinc-L and polynomial options are pretty much at opposite ends. Question is, when we talk "NOS", what are the traits you're looking for? Here's some of them, regardless if you like them or not:
1. early, slow roll-off
2. little to no ringing
3. non-harmonic distortion
4. bit-perfect reproduction
Not trying to start a NOS discussion here - but as an interlude to look at the filters.
And for starters, let it be known that I also like NOS and have a DAC that can disable filters.

The closed form filters keep #4 but are otherwise long filters (strike through #2) with very high attenuation (strike through #1). Doing so they remove #3 because there are no longer mirror images summing in the audible band at uncorrelated frequencies (that I would always call a "fix" - I can understand euphonic from harmonic distortion, but non-harmonics? No.) Wasn't it Schiit who said about their "comboburrito" closed form filter that it's like a marriage of the best that NOS and OS have to offer? I'd agree...

Sinc-L loses the big thing that closed form brings: it's not bit-perfect (strike through #4). It has in common that's is also long with very high attenuation. It's not NOS at all, actually.

Then the polynomial ones... interesting case. Compared to full NOS, roll-off is quite a bit later but still slow (#1 improved, retaining #2) and with poor attenuation (#3 improved but still there). Pretty sure it's not bit-perfect though (strike through #4). If you like full NOS I can see how you can appreciate this as "NOS, improved" because it will have lower non-harmonic distortion - although still some of it.

Both options really do seem better than "true NOS" with simple zero-order hold, objectively, with strong suspicion that they will hold up subjectively with all NOS aficionados.

Thanks for showing me this path!

 

[plumpudding2]

Thanks for posting this. Took me a while to respond, as I it took me a few days through the rabbit hole to try it out.

I thought it was quite thought-provoking, in a good way. Funny because the closed form/sinc-L and polynomial options are pretty much at opposite ends. Question is, when we talk "NOS", what are the traits you're looking for? Here's some of them, regardless if you like them or not:
1. early, slow roll-off
2. little to no ringing
3. non-harmonic distortion
4. bit-perfect reproduction
Not trying to start a NOS discussion here - but as an interlude to look at the filters.
And for starters, let it be known that I also like NOS and have a DAC that can disable filters.

The closed form filters keep #4 but are otherwise long filters (strike through #2) with very high attenuation (strike through #1). Doing so they remove #3 because there are no longer mirror images summing in the audible band at uncorrelated frequencies (that I would always call a "fix" - I can understand euphonic from harmonic distortion, but non-harmonics? No.) Wasn't it Schiit who said about their "comboburrito" closed form filter that it's like a marriage of the best that NOS and OS have to offer? I'd agree...

Sinc-L loses the big thing that closed form brings: it's not bit-perfect (strike through #4). It has in common that's is also long with very high attenuation. It's not NOS at all, actually.

Then the polynomial ones... interesting case. Compared to full NOS, roll-off is quite a bit later but still slow (#1 improved, retaining #2) and with poor attenuation (#3 improved but still there). Pretty sure it's not bit-perfect though (strike through #4). If you like full NOS I can see how you can appreciate this as "NOS, improved" because it will have lower non-harmonic distortion - although still some of it.

Both options really do seem better than "true NOS" with simple zero-order hold, objectively, with strong suspicion that they will hold up subjectively with all NOS aficionados.

Thanks for showing me this path!

Thanks for your thoughts! To me the NOS sound is very natural though a bit less detailed. OS enhances detail but can lead to things feeling "smoothed over", I like to draw an analogy to vision how our eyes are extremely sensitive to bad image upsampling and reject overly blurry upsampled images as "vaseline smeared", sometimes visually you'd prefer a rough coarse but sharp image to a smooth blurry one.
Sinc-L is indeed mathematically the furthest thing from NOS except for one sonic trait, the ringing artifacts provide some sort of edge-enhancement that make it sound almost short again, see this nice visual description from Wikipedia:

I'm quite sure all interpolating filters keep the original samples intact, both polynomial and closed-form.

Since the concept of bit-perfect is out the window when upsampling to DSD my explorations of the ultrashort end of the filter spectrum led me to try minringFIR as a filter for DSD. In the end I settled on poly-sinc-mp as it has similar transient quality but is apodising. I previous discarded it since most people say minimum phase is bad for headphones (which are linear phase devices) and I thought it was superseded by the more efficient Gauss filters, but it sounds really good right now.
Not sure yet what I prefer yet between PCM-polynomial-2 and DSD-poly-sinc-short-mp. Depends on headphones and amp and music

 

[Ghoostknight]

Thanks for posting this. Took me a while to respond, as I it took me a few days through the rabbit hole to try it out.

I thought it was quite thought-provoking, in a good way. Funny because the closed form/sinc-L and polynomial options are pretty much at opposite ends. Question is, when we talk "NOS", what are the traits you're looking for? Here's some of them, regardless if you like them or not:
1. early, slow roll-off
2. little to no ringing
3. non-harmonic distortion
4. bit-perfect reproduction
Not trying to start a NOS discussion here - but as an interlude to look at the filters.
And for starters, let it be known that I also like NOS and have a DAC that can disable filters.

The closed form filters keep #4 but are otherwise long filters (strike through #2) with very high attenuation (strike through #1). Doing so they remove #3 because there are no longer mirror images summing in the audible band at uncorrelated frequencies (that I would always call a "fix" - I can understand euphonic from harmonic distortion, but non-harmonics? No.) Wasn't it Schiit who said about their "comboburrito" closed form filter that it's like a marriage of the best that NOS and OS have to offer? I'd agree...

Sinc-L loses the big thing that closed form brings: it's not bit-perfect (strike through #4). It has in common that's is also long with very high attenuation. It's not NOS at all, actually.

Then the polynomial ones... interesting case. Compared to full NOS, roll-off is quite a bit later but still slow (#1 improved, retaining #2) and with poor attenuation (#3 improved but still there). Pretty sure it's not bit-perfect though (strike through #4). If you like full NOS I can see how you can appreciate this as "NOS, improved" because it will have lower non-harmonic distortion - although still some of it.

Both options really do seem better than "true NOS" with simple zero-order hold, objectively, with strong suspicion that they will hold up subjectively with all NOS aficionados.

Thanks for showing me this path!

how does syncM (or better syncMGA) fit into this from tour perspective? i liked it a lot more than syncL but i might retest, tho my memory is blurred because i compared syncMGA

apodizing errors seem to be something that is worth getting rid of from what i heared, even if we are all used to it from normal PCM playback, there are only a few filters that are apodizing

songs with large numbers of apodizing errors (i actually found some that reach 50k apodizing errors in a few minutes) sound less like what would you expect if someone says "this sounds analog etc"

 

[Erick Muller]

how does syncM (or better syncMGA) fit into this from tour perspective? i liked it a lot more than syncL but i might retest, tho my memory is blurred because i compared syncMGA

apodizing errors seem to be something that is worth getting rid of from what i heared, even if we are all used to it from normal PCM playback, there are only a few filters that are apodizing

x2, I also liked a lot both sync-M and sync-mGa, tho currently using sync-Ll because I liked a lot the sound I'm getting with my system, tho apodizing errors galore I will admit.

 

[Ghoostknight]

x2, I also liked a lot both sync-M and sync-mGa, tho currently using sync-Ll because I liked a lot the sound I'm getting with my system, tho apodizing errors galore I will admit.

i might retest some later on... closed-form sounds quiet impressive too... im just not sure if the sub bass region sounds really like "intended", it sounds like high resolution but the bass under 40-50hz seems "bloated" in some kind (it sounds like a 3-6db boost under 30-40hz), i might need to try to adjust my eq settings accordingly, i noticed this with other filters too... it can actually audibly somewhat mess up your perfectly balanced "housecurve/eq"

regarding the eq i might share some toughts because its fitting: i noticed that 0,2db rolloff, lets say starting at 1k and rolling off to 20khz can actually be quite audible and skew the balance towards bass even if it was just 0,2db.... so my believe after a lot eq testing is somewhat that the not always PERFECTLY flat response of dacs could skew with listener preference... even just 0,2db can make a difference ime, so the preference you might have for some dac could be very well objectivly proofen things, different implementations of reconstruction filters and how much "perfect" bandwidth the dac can produce (specially in combination with the applied reconstruction filters)

from audible expierence i can just explain it like that: if you compare a single tone lets say 10khz with 10khz that is 0,2db quieter, you probably laugh but the thing with EQ is the balance between, simple said, lows, mids and highs if you hear low and high tones at the same time then the balance between those two can be quite obvious... i guess it just gives us more information to compare it to...

and most people probably prefer a "slight" balance towards bass ... it just sounds easier to the ear than some high frequency spikes that "stick out" out of the bass

 

[bogi]

apodizing errors seem to be something that is worth getting rid of from what i heared, even if we are all used to it from normal PCM playback, there are only a few filters that are apodizing

Actually, there are more than 25 apodizing filters included in HQPlayer, not yet counting the -2s variants of poly sinc filters.
Apodizing filters are marked with 'x' in the last column of the filter table in HQPlayer PDF manual.

 

[Ghoostknight]

Apodizing filters are marked with 'x' in the last column of the filter table in HQPlayer PDF manual.

ahhh thank you, i missed that ... FIR and IIR is apodizing too, that makes more sense now...

Edit: sincMx is probably my favourite now... upsampling to DSD264 still works but i have to see how to get more going... 512 atleast would be optimal since there are the 512fs optimized versions of modulators

i tried to run DSD512 but hqplayer keeps crashing even tho cpu usage wasnt that high with a intel 12400, is this some bug or do i have to look to upgrade my 1070GTX to some bigger card and use that? (same crash problem if i try to use cuda offloading)

 

[bogi]

crash problem if i try to use cuda offloading

As the 1st thing I would check if you have the latest nVidia drivers installed. Every new HQPlayer version is based on latest nVidia drivers and may crash with older driver.
Use the latest Studio driver, unless you are also playing games on the same computer, in such a case install the latest Game Ready driver.

After the driver update just try to compare PCM to DSD512 with and without CUDA, starting with some light filter like poly-sinc-short-mp-2s. Your CPU is OK for running any modulator, but your GPU may be too slow for processing filters at DSD512, resulting in CPU waiting for GPU to finish its job. If you don't get satisfying results, GPU upgrade is the way.

You can use the 512+ modulators also at DSD256 rate, I mentioned it more it more times in this thread with an explanation too, just try to find it.

 

[Erick Muller]

ahhh thank you, i missed that ... FIR and IIR is apodizing too, that makes more sense now...

Edit: sincMx is probably my favourite now... upsampling to DSD264 still works but i have to see how to get more going... 512 atleast would be optimal since there are the 512fs optimized versions of modulators

i tried to run DSD512 but hqplayer keeps crashing even tho cpu usage wasnt that high with a intel 12400, is this some bug or do i have to look to upgrade my 1070GTX to some bigger card and use that? (same crash problem if i try to use cuda offloading)

I like sinc-Mx, currently listening to sinc-mGa, it's good, i think i'm going to listen for an entire week to each filter and see which I like better.

In order to run DSD512, you need to be using ASIO drivers for your DAC, just make sure you have them install.

 

[bogi]

In order to run DSD512, you need to be using ASIO drivers for your DAC, just make sure you have them install.

He mentions Fedora in signature. Linux supports native DSD too.

 

[Erick Muller]

He mentions Fedora in signature. Linux supports native DSD too.

Didn't notice, honestly I'm completely unfamiliar with linux.

 

[gammi]

Hi, long time HQPlayer user here but I got a new DAC (Cyan 2) and also used PCM usually on my R26 DAC but I'm switching to DSD on my Cyan so I have some questions.

I did some quick searching but didn't have much success. Sorry for asking a question that has probably been asked a lot, but what are the favorite filters and modulators for DSD currently? I was using Sinc-Mx and ASDM7ECv3 last night. Is V3 still the best or should I be using super or one of the 512/fs modulators?

I have a powerful PC, AMD 7950x 16 core at 5Ghz+ and RTX4090. For reference, with ASDM7ECv3 it takes about 10 seconds to start playback. Is this normal for that modulator or is something suboptimal? If there's another modulator that is as good as ASDM7ECv3 but faster, I would love to know. Would be nice to get playback down to 2-3 seconds. This is with DSD256 btw. Thanks everyone.

 

[bogi]

what are the favorite filters and modulators for DSD currently? I was using Sinc-Mx and ASDM7ECv3 last night. Is V3 still the best or should I be using super or one of the 512/fs modulators?

I recommend you to try 512+ versions of super and light modulator. You could experience improvement in soundstage space presentation. 512+ filters are ok on DSD256 too. Your hardware should run DSD512 without issues.

with ASDM7ECv3 it takes about 10 seconds to start playback. Is this normal for that modulator or is something suboptimal? If there's another modulator that is as good as ASDM7ECv3 but faster, I would love to know. Would be nice to get playback down to 2-3 seconds.

Modulators don't cause noticable delay. That's the extra long filter sinc-Mx you chose. Try different kind of filters, for example:
- long filters: poly-sinc-ext2, poly-sinc-ext3, poly-simc-gauss-xla
- middle lenght filters: poly-sinc-hb-m, poly-sinc-mp, poly-sinc-gauss, poly-sinc-gauss-halfband, sinc-Lm
- short filters: poly-sinc-short-mp, poly-sinc-xtr-short-mp and short versions of the above moddle length filters where available.

There is no such equation that longer (harder to process) filter needs to bring better sound. Filter design is about dealing with compromises. One of them is that although long filters perform very well in frequency domain, they perform weaker in time domain. It results to transient smear particularly with music content containing fast transients (percussions, electronic instruments). Short filters are suitable for pop music whose instruments ar eseparately recorded and then mixed, so no natural space information appears in such a recording and artificial room acoustics is added into the mix. Short filters presents transients very accurately but they lack natural space presentation with acoustic music like classical music or acoustic jazz.

 

[gammi]

I recommend you to try 512+ versions of super and light modulator. You could experience improvement in soundstage space presentation. 512+ filters are ok on DSD256 too. Your hardware should run DSD512 without issues.



Modulators don't cause noticable delay. That's the extra long filter sinc-Mx you chose. Try different kind of filters, for example:
- long filters: poly-sinc-ext2, poly-sinc-ext3, poly-simc-gauss-xla
- middle lenght filters: poly-sinc-hb-m, poly-sinc-mp, poly-sinc-gauss, poly-sinc-gauss-halfband, sinc-Lm
- short filters: poly-sinc-short-mp, poly-sinc-xtr-short-mp and short versions of the above moddle length filters where available.

There is no such equation that longer (harder to process) filter needs to bring better sound. Filter design is about dealing with compromises. One of them is that although long filters perform very well in frequency domain, they perform weaker in time domain. It results to transient smear particularly with music content containing fast transients (percussions, electronic instruments). Short filters are suitable for pop music whose instruments ar eseparately recorded and then mixed, so no natural space information appears in such a recording and artificial room acoustics is added into the mix. Short filters presents transients very accurately but they lack natural space presentation with acoustic music like classical music or acoustic jazz.

I wanted to stay away from DSD512 because SNR performance actually drops when using that with the Cyan 2. I will give a try to using the 512 modulators with DSD256 though, thanks.

Ah, I thought it was the modulator since that's run on the "slow" CPU and I figured the GPU is going to crush any filter given to it considering it does 82 TRILLION operations per second. Very much appreciate that list of suggested filters you gave, I'll give them a try, thank you.

EDIT:
I just set it to DSD256, poly-sinc-gauss, poly-sinc-gauss-hires-mp, and ASDM7EC-super 512+fs
Will be listening to that config later.

 

[Erick Muller]

This is something that i think you'll find interesting.