HQ Player

[Zauurx]

19 hours ago, Miska said:

 

Something there that makes you run out of RAM? Although that usually just freezes Windows. Some memory or power issue is another possibility. Or a driver issue, I had similar issue once when Windows update changed the Windows graphics stack, and Intel graphics driver wasn't automatically updated. Resulted in strange crashes.

 

Generally, applications shouldn't be able to crash the OS, no matter what. So this is either power, or low level OS issue. Typical consequence from voltage rail sag due to power consumption spike is system reboot.

 

And the diagnosis was right !

Congratulations !! @Miska 😉 and best wishes for a successful convalescence ! 
After trying to restore the OS, I had the same problem.
Same thing with a usb key running winPE.
I decided to rewire my server with its internal HDPlex Nano AC-DC + HDPlex DC-ATX power supply, replacing the large external HDPlex 400 ATX power supply.
Well... no more problems. 
In fact, when I put my nose over the ATX linear 400w case, the smell was a little suspicious, even though everything seemed to be working... a component at the end of its life?

Anyway, here's what it looks like  :

 

[semente]

Is it possible to create a very slow (e.g. 3dB/octave) high-pass filter on the Matrix (not that Matrix 😁)?

 

 

Edit: if I understand correctly the syntax should be for example iir:type=hp;f=80;q=1 but what's the tilt of Q=1?

[GoldenOne]

So with an HQP Pro trial you can test a few bits.

Here's an input with a -180dB 1khz signal on ASDM7ECv2, AMSDM7 512+fs and AMSDM7EC 512+fs

Seems like actually the pseudo-multibit modulators have slightly lower dynamic range than ASDM7ECv2?

I can't test ASDM7ECv3 or the super modulators cause those aren't in HQP pro and there's no way to record either direct DSD512/1024 or a high enough PCM rate to do DoP with a virtual cable unfortunately. But seems like in theory the non-512+fs modulators might be the technically better choices? Slightly better dynamic range and noise shaped further out

[copy_of_a]

1 hour ago, semente said:

Is it possible to create a very slow (e.g. 3dB/octave) high-pass filter on the Matrix (not that Matrix 😁)?

 

 

Edit: if I understand correctly the syntax should be for example iir:type=hp;f=80;q=1 but what's the tilt of Q=1?

no. HP and LP filters are fixed 12dB/octave. "Q" defines the resonance (not "tilt"), not the steepness of the filters.

You could create a Dirac Impulse and apply a 3dB/octave HP with a dedicated Equalizer and add that to your existing EQ set in the matrix.

BTW: a HP filter at 80Hz with 3dB/octave slope would affect the whole spectrum 20Hz - 20kHz.

[StreamFidelity]

5 hours ago, Kalpesh said:

though I seem the only one at this stage of development, my appreciation of AHM7EC5L @1024 with the MAy

not really 😉

 

I experience the same sound improvement!

 

4 hours ago, Apollo said:

a lot of us are waiting on Embedded 5.3 to be able to try it out.

It is already available. 

 

 

[Triplefun]

Please can someone confirm if I have a rtx4090 on windows 11 pro and I am stuck with amd am5 CPUs then the 4950x with the higher clock speed would be better than the 4950x3d. 

[Triplefun]

Or since I am stuck with windows rather than upgrading the CPU I should switch to Intel and get a 14900k and z790 mobo.

[semente]

2 hours ago, copy_of_a said:

no. HP and LP filters are fixed 12dB/octave. "Q" defines the resonance (not "tilt"), not the steepness of the filters.

You could create a Dirac Impulse and apply a 3dB/octave HP with a dedicated Equalizer and add that to your existing EQ set in the matrix.

BTW: a HP filter at 80Hz with 3dB/octave slope would affect the whole spectrum 20Hz - 20kHz.

 

If Q doesn't affect the tilt why/how is it used in a HP or LP?

[copy_of_a]

36 minutes ago, semente said:

 

If Q doesn't affect the tilt why/how is it used in a HP or LP?

Q defines the resonance. See screenshot.

 

q=0.71 (actually 0.709) is standard (same as S=1)

q=1 creates a boost (resonance) around the corner frequency

q=2 creates a stronger boost (resonance) around the corner frequency

etc.

 

[Quadman]

5 hours ago, StreamFidelity said:

though I seem the only one at this stage of development, my appreciation of AHM7EC5L @1024 with the MAy

I have and continue to be a huge fan of DSD1024, I have been telling all when I post magic happens here.  Now with the new AHM modulators many people can finally experience DSD1024 with EC modulators.  I have played quite a bit with these AHM and other EC modulators at 1024.  I have not definitively decided if I prefer the AHM modulators over the non AHM at 1024 with EC.  I am pretty confident that I prefer AHM5EC5L to AHM7EC5L with my Holo May.  I certainly really like the new AHM modulators and have spent about 80% of my listening time with them since 5.3's release.  I like that my PC with a 3090 GPU draws close to 100W less power running the AHM modulators vs "normal" EC modulators.  I like the sound of the new AHM modulators.  But I am not partial to quick judgements, I need to to see how I feel listening to my music, am I more engaged, am I chair dancing more am I listening more total hours, do I NOT want to turn my stereo off in the early mourning hours.  So far AHM is impressive.  Got to get back to listening.

 

My prayer's to you Jussi and a to quick recovery.  Your program has brought me more musical joy in the last 8 years than anything else.

[Miska]

15 hours ago, GoldenOne said:

Here's an input with a -180dB 1khz signal on ASDM7ECv2, AMSDM7 512+fs and AMSDM7EC 512+fs

Seems like actually the pseudo-multibit modulators have slightly lower dynamic range than ASDM7ECv2?

I can't test ASDM7ECv3 or the super modulators cause those aren't in HQP pro and there's no way to record either direct DSD512/1024 or a high enough PCM rate to do DoP with a virtual cable unfortunately. But seems like in theory the non-512+fs modulators might be the technically better choices? Slightly better dynamic range and noise shaped further out

 

Now don't mix ASDM and AMSDM... Yes, AMSDM has slightly lower SNR, but it has some other advantages when running certain types of D/A conversion stage electronics, due to the way it operates. AHM has even lower SNR, while it can counter some adverse effects of the conversion stage electronics running at higher speeds.

 

Now, putting AMSDM modulators aside and talking just about the regular EC modulators, 512+fs modulators have better SNR in audio band than the regular ones, but less bandwidth. This is on purpose, since at DSD512 with regular modulators you have 200 kHz of extremely low noise audio bandwidth. The 512+fs ones meanwhile focus on providing even more dynamic range in <100 kHz region while trading some bandwidth from >100 kHz with gentler noise slope.

 

So if you are doing DSD512 or more, you can choose between SNR (512+fs modulators) or frequency bandwidth (regular modulators). The difference is some 10 - 20 dB. (in practice, the "passband" noise floor is anyway way below what is possible in analog domain 😀)

 

[StreamFidelity]

9 minutes ago, Miska said:

The difference is some 10 - 20 dB. (in practice, the "passband" noise floor is anyway way below what is possible in analog domain

What about the AHM7EC5L modulator? After all, we are talking about DSD1024. Nevertheless, digital volume control is not recommended due to the low SNR. 

 

How critical is hybrid volume control? At the moment I have set -20dB in the HQPlayer and 60dB in the SDV 3100 HV. This allows me to increase the volume by about 15dB, which is enough for me.

[Miska]

7 minutes ago, StreamFidelity said:

What about the AHM7EC5L modulator? After all, we are talking about DSD1024. Nevertheless, digital volume control is not recommended due to the low SNR. 

 

How critical is hybrid volume control? At the moment I have set -20dB in the HQPlayer and 60dB in the SDV 3100 HV. This allows me to increase the volume by about 15dB, which is enough for me.

 

AHM matches what you regularly get for example from typical power amplifiers. So you don't have extra headroom. If there would be a DAC that would support DSD2048 you would get more.

 

So with AHM I would go with analog volume control at leave HQPlayer at -3 dBFS. This way everything was perfectly fine for my listening tests, both speaker systems and headphones.

 

With regular modulators you can use as much digital attenuation in HQPlayer as you like, since you will be limited by laws of physics in analog world rather than available digital headroom. Hybrid approach is still good at least just to make sure there cannot be volume accidents, especially if one outputs audio from something like Windows or macOS (or Linux desktop).

 

[Kalpesh]

3 minutes ago, Miska said:

 

AHM matches what you regularly get for example from typical power amplifiers. So you don't have extra headroom. If there would be a DAC that would support DSD2048 you would get more.

 

So with AHM I would go with analog volume control at leave HQPlayer at -3 dBFS. This way everything was perfectly fine for my listening tests, both speaker systems and headphones.

 

With regular modulators you can use as much digital attenuation in HQPlayer as you like, since you will be limited by laws of physics in analog world rather than available digital headroom. Hybrid approach is still good at least just to make sure there cannot be volume accidents, especially if one outputs audio from something like Windows or macOS (or Linux desktop).

 

due to convolution I'm rather in the 15 20 dB range of constant digital attenuation in HQP with variable volume control vi preamp.

And no SNR issue I can notice and I play as loud as required by the recording

[GoldenOne]

5 hours ago, Miska said:

 

Now don't mix ASDM and AMSDM... Yes, AMSDM has slightly lower SNR, but it has some other advantages when running certain types of D/A conversion stage electronics, due to the way it operates. AHM has even lower SNR, while it can counter some adverse effects of the conversion stage electronics running at higher speeds.

 

Now, putting AMSDM modulators aside and talking just about the regular EC modulators, 512+fs modulators have better SNR in audio band than the regular ones, but less bandwidth. This is on purpose, since at DSD512 with regular modulators you have 200 kHz of extremely low noise audio bandwidth. The 512+fs ones meanwhile focus on providing even more dynamic range in <100 kHz region while trading some bandwidth from >100 kHz with gentler noise slope.

 

So if you are doing DSD512 or more, you can choose between SNR (512+fs modulators) or frequency bandwidth (regular modulators). The difference is some 10 - 20 dB. (in practice, the "passband" noise floor is anyway way below what is possible in analog domain 😀)

 

Thanks

I've managed to get some digital capture of the new modulators and...well...nevermind 10-20dB. How about upto 40-50dB in the upper end of the passband?
Sure, below the analog noise floor but that's pretty wild performance!

 

 

This would mean that the modulator SNR isn't the limit but rather the file itself. 
Feeding a 24 bit file in (-120dB sine in this case, -1dB headroom applied on both Roon and HQP), the results for say Roon vs ASDM7EC-Super 512+fs seem to show that the 512+fs modulator is limited by the 24 bit file itself any any dithering therein. Neat!

[bogi]

16 minutes ago, GoldenOne said:

How about upto 40-50dB in the upper end of the passband?

That's work for analog filter of DAC. If delta sigma modulator (noise shaping) wouldn't be used (like with R2R DACs) then you would get audio band mirror images (spectral replicas) at output. One or the other. Analog filter is always needed as part of audio DAC. This requirement comes from sampling theory. In digital domain you cannot remove mirror images above fs and similarly you cannot remove modulator high frequency noise.

 

Noise shaping as result of delta sigma modulation does not lower amount of digital noise. It just changes its distribution. The goal is to lower noise in audio band and to shift it to high frequencies, where it can be easily removed even with gentle analog filter.

[GoldenOne]

Just now, bogi said:

That's work for analog filter of DAC. If delta sigma modulator (noise shaping) wouldn't be used (like with R2R DACs) then you would get audio band mirror images (spectral replicas) at output. One or the other. Analog filter is always needed as part of audio DAC. This requirement comes from sampling theory. In digital domain you cannot remove mirror images above fs and similarly you cannot remove modulator high frequency noise.

 

Noise shaping as result of delta sigma modulation does not lower amount of digital noise. It just changes its distribution. The goal is to lower noise in audio band and to shift it to high frequencies, where it can be easily removed even with gentle analog filter.

I'm aware

[Miska]

55 minutes ago, GoldenOne said:

seem to show that the 512+fs modulator is limited by the 24 bit file itself any any dithering therein.

 

Yeah, for fun during development I tested that I can get 192/24 file bit-perfect through back and forth conversion. And also for example 48/32 file.

 

(below 10 kHz, those plots are limited by the test method for super 512+fs)

 

[Miska]

22 minutes ago, bogi said:

That's work for analog filter of DAC.

 

And a delta-sigma DAC is analog filter by itself too. For example in my DSC1 there's the conversion stage that works as linear phase analog FIR, followed by 4th order regular analog low-pass with 100 kHz corner frequency. Result is that running at DSD512 the output noise floor is flat throughout at all frequencies (including any MHz range).

 

For example Holo Audio DACs are similar.

 

[GoldenOne]

6 hours ago, Miska said:

 

Now don't mix ASDM and AMSDM... Yes, AMSDM has slightly lower SNR, but it has some other advantages when running certain types of D/A conversion stage electronics, due to the way it operates. AHM has even lower SNR, while it can counter some adverse effects of the conversion stage electronics running at higher speeds.

 

Now, putting AMSDM modulators aside and talking just about the regular EC modulators, 512+fs modulators have better SNR in audio band than the regular ones, but less bandwidth. This is on purpose, since at DSD512 with regular modulators you have 200 kHz of extremely low noise audio bandwidth. The 512+fs ones meanwhile focus on providing even more dynamic range in <100 kHz region while trading some bandwidth from >100 kHz with gentler noise slope.

 

So if you are doing DSD512 or more, you can choose between SNR (512+fs modulators) or frequency bandwidth (regular modulators). The difference is some 10 - 20 dB. (in practice, the "passband" noise floor is anyway way below what is possible in analog domain 😀)

 

What particular DACs would you recommend using AMSDM instead of ASDM?

For example those using Holo Audio or a T+A DAC200, should they go to AMSDM or ASDM?

[Miska]

58 minutes ago, GoldenOne said:

What particular DACs would you recommend using AMSDM instead of ASDM?

 

Originally my DSC1 was target, and some other discrete DAC implementations. This was discussed also at the DIY site back then. The benefits appear primarily on bit-perfect / discrete DACs.

 

58 minutes ago, GoldenOne said:

For example those using Holo Audio or a T+A DAC200, should they go to AMSDM or ASDM?

 

One can test which one is preferred. AMSDM is one kind of approach to utilise extra bandwidth/rate available at DSD512 and higher.

 

This is one of the weighting things, which aspects are preferred in particular case and system.

 

For example on my living room system I'm running DAC200 at DSD512 from ASDM7EC-super 512+fs. And on the same rack Marantz SA-12SE at DSD256 from ASDM7EC-light. My Spring 2 and 3 are running with different modulators from different servers. HA200 is my main development DAC so it runs at varying rates from varying modulators depending on what I'm working on.

 

[OzarkMtn]

1 hour ago, GoldenOne said:

What particular DACs would you recommend using AMSDM instead of ASDM?

For example those using Holo Audio or a T+A DAC200, should they go to AMSDM or ASDM?

@Miska First, Please continue to keep your well being and health at the forefront!! Hope that you are comfortable in your ongoing recovery.

 

Would you also please include Denafrips DACS in your recommendations using AMSDM instead of ASDM?

Thanks.

[Miska]

1 minute ago, OzarkMtn said:

@Miska First, Please continue to keep your well being and health at the forefront!! Hope that you are comfortable in your ongoing recovery.

 

I'm doing this at pretty relaxed pace now, trying to recover. Slowly catching up with emails etc.

 

1 minute ago, OzarkMtn said:

Would you also please include Denafrips DACS in your recommendations using AMSDM instead of ASDM?

Thanks.

 

If you run it in DSD mode, using highest possible rate is best. In fact it could be interesting to test AHM at DSD1024 there (didn't do so yet). But since it doesn't seem to be bit-perfect, there is no particular benefit from AMSDM. You can of course still use it if you prefer.

 

I would usually recommend to just send highest possible PCM rate to Denafrips unless you prefer DSD mode from sonic perspective.

 

[Gato]

´I'm running DAC200 at DSD512´ why not at DSD1024, is it due to the spec of the pc running hqplayer or …?

[GoldenOne]

3 hours ago, Miska said:

 

I'm doing this at pretty relaxed pace now, trying to recover. Slowly catching up with emails etc.

 

 

If you run it in DSD mode, using highest possible rate is best. In fact it could be interesting to test AHM at DSD1024 there (didn't do so yet). But since it doesn't seem to be bit-perfect, there is no particular benefit from AMSDM. You can of course still use it if you prefer.

 

I would usually recommend to just send highest possible PCM rate to Denafrips unless you prefer DSD mode from sonic perspective.

 

I didn't do particularly extensive testing but with both the ares 2 and terminator plus, dsd input performed/behaved 100% identically to PCM input so I'm guessing they resample to PCM internally.