HQ Player

[Miska]

5 minutes ago, DomiJi said:

Can you may tell me the benefit of Sinc-Mx compared to Sinc-M? Is Sinc-Mx technically superior than Sinc-M?

 

sinc-Mx operates similar way as other filters, where filter response is same regardless of conversion ratio. IOW, the filter length scales with the conversion ratio. While sinc-M is fixed length and thus filter response depends on the conversion ratio. This also means that sinc-Mx has constant delay, unlike sinc-M.

 

They are the same at 16x conversion ratio.

 

[Miska]

2 hours ago, GMG said:

I thought that it is better to upsample 44.1 to 705.6 because it is exactly x16. And that upsampling to 768 which is x17.4 is not only higher stress on the cpu but more importantly not optimized for the samples interpolation. 

 

The higher CPU load is mostly due to higher RAM load, not because there would be notably more processing. Quality is the same though.

 

2 hours ago, GMG said:

Are you saying that if cpu is not an issue it is better to use fixed 768?

 

Yes, especially when DAC is anyway running it's conversion out of constant clock independent of source rate (Chord and most ESS Sabre).

 

This is also anyway what you'd have in most DSD upsampling cases, because most DACs support DSD only at multiples of 44.1k.

[Miska]

2 hours ago, NoNaim said:

If you have multiple NAA's they show up under "Device" in the above settings. If I want to change, select a different device without logging in remotely to the Mac mini M1 that runs HQPlayer is it possible at all to change/select device with the HQPDcontrolv4 app?

 

No, the regular settings are only available through other means. One reason is that when you change the device also your control connection is lost because the engine gets restarted.

 

2 hours ago, NoNaim said:

I've been able to use this to change most things I need, inc changing pipeline profiles for my headphones but couldn't see / find an option to select devices. I have a feeling it isn't but just wanted to check. If not, I may have to setup some sort of VNC access for my iPhone. Thanks

 

HQPlayer Embedded is easier for this since you can change such settings using Safari.

 

[DomiJi]

1 hour ago, Miska said:

 

sinc-Mx operates similar way as other filters, where filter response is same regardless of conversion ratio. IOW, the filter length scales with the conversion ratio. While sinc-M is fixed length and thus filter response depends on the conversion ratio. This also means that sinc-Mx has constant delay, unlike sinc-M.

 

They are the same at 16x conversion ratio.

 

 

Thanks a lot :-)

 

 

With my Pro iDSD i like Closed-Form-M a lot but with the Hugo feeding in PCM i am able to use the "poly-sinc-gauss-xla" Filter and it sounds absolutely gorgeous! Everything sound so realstic and authentic. Wonderful!

 

I'm just a bit unsure if i should use LNS9 or LNS15... can you may explain what is the difference between the two? I think i like LNS9 a bit more but that could be just a placebo 

[Miska]

17 minutes ago, DomiJi said:

I'm just a bit unsure if i should use LNS9 or LNS15... can you may explain what is the difference between the two? I think i like LNS9 a bit more but that could be just a placebo 

 

NS9 and LNS15 have different kind of noise shaping profile. And LNS15 is what I call a "linear noise shaper". LNS15 is optimized for higher output rates, while NS9 is optimized to work also at lower output rates.

 

You can use either one, if NS9 sounds better to you, then it is better to use that one.

 

[DomiJi]

Thank you very much!

 

Regardless which filter or which modulator/noise shaper. HQPlayer was the best investment i have ever did to my system 

[micheloupatrick]

i’m using HQplayer with Roon on Ubuntu 20.04 to upsample everything to DSD128.

Using Sinc-M all is fine but today I gave sinc-Mx a try and something wierd is happening : there’s a 4-5 seconds delay between the time I hit the stop button in Roon and the time the music actually stops playing. I tried many filters before (sometimes with a huge delay before playing the music, but never  after).

Any idea why it’s happening ?

 

edit: no idea why the font is so big

[k6davis]

4 hours ago, Miska said:

Yes, especially when DAC is anyway running it's conversion out of constant clock independent of source rate (Chord and most ESS Sabre).

 

This is also anyway what you'd have in most DSD upsampling cases, because most DACs support DSD only at multiples of 44.1k.

 

My Lampizator DAC (which has no DAC chip) does support 48k base rates, so I have "Auto Rate Family" & "48k DSD" checked in HQP OS.

 

Is that what you'd recommend in my case?

[k6davis]

2 hours ago, DomiJi said:

Thank you very much!

 

Regardless which filter or which modulator/noise shaper. HQPlayer was the best investment i have ever did to my system 

 

I'm running HQP Embedded/OS, which currently has @Miska's latest updates. I expect that they'll be in the forthcoming release of HQP Desktop. 

 

All I can say is that you guys are in for a treat. I have always thought that HQP is fantastic, but to my ears, the latest version is the best it's ever sounded. The release notes say there are "sound quality improvements". I have no idea what he did, but the music sounds (even) more natural than before. And the improvement I'm describing has nothing to do with the new "v2" modulators - which are also great. I hear it when I use other modulators too. The function of the software is also improved - connectivity to Roon, network connectivity, overall stability, even the UI. It's a great release all around.

 

So hats off to you @Miska. The program is amazing. Your efforts and support here on the forum are much appreciated. 

[BrownMagic]

I am using an intel mini stick running linux and HQP NAA. 1.5M on my Holo May works great when I am using Qobuz and Roon. 

My wife on the other hand likes to use spotify. Is there a way for me to beam spotify via spotify connect to the HQP NAA. I do not expect any upscaling or anything. I want to know if I can also install some linux based spotify connect receiver on the Intel Mini Stick. Can both of these coexist on the same stick?

[Reg19]

I have a question regarding volume in HQP.

 

I feed HQP Desktop (on a Mac M1 Mini) to NAA (either the SoTM SMS-200Ultra Neo (with SBooster power supply) or a Mele Intel PC Stick (with stock power supply)) to the Holo May L2 DAC.

 

If I only upsample to PCM (768KHz using sinc L filter and LNS15 noise shaper), do I have to still have -3dB as the max volume in HQP? Can it bet set to 0dB?

 

Or is that -3dB setting only used by HQP if I choose upsampling via DSD and does not apply if one uses PCM?

 

Also: if I do use the Intel PC Stick, do you recommend I upgrade the power supply to the Intel PC Stick?

[Miska]

4 hours ago, micheloupatrick said:

i’m using HQplayer with Roon on Ubuntu 20.04 to upsample everything to DSD128.

Using Sinc-M all is fine but today I gave sinc-Mx a try and something wierd is happening : there’s a 4-5 seconds delay between the time I hit the stop button in Roon and the time the music actually stops playing. I tried many filters before (sometimes with a huge delay before playing the music, but never  after).

Any idea why it’s happening ?

 

What is your output rate? sinc-Mx adds about 1 second extra delay. With sinc-M it depends on your output rate.

 

[Miska]

3 hours ago, k6davis said:

My Lampizator DAC (which has no DAC chip) does support 48k base rates, so I have "Auto Rate Family" & "48k DSD" checked in HQP OS.

 

Is that what you'd recommend in my case?

 

It is fine. Auto rate family is mostly useful to deal with filters that cannot convert between rate families. When filter can convert between rate families it just affects CPU/RAM load to some extent.

 

[Miska]

1 hour ago, Reg19 said:

If I only upsample to PCM (768KHz using sinc L filter and LNS15 noise shaper), do I have to still have -3dB as the max volume in HQP?

 

Yes...

 

1 hour ago, Reg19 said:

Can it bet set to 0dB?

 

No, it will keep triggering the limiter due to inter-sample overs.

 

1 hour ago, Reg19 said:

Or is that -3dB setting only used by HQP if I choose upsampling via DSD and does not apply if one uses PCM?

 

It applies to both cases. If you use also DSD mode of Holo Audio DAC, set gain compensation to -6 dB. This way both PCM and DSD have same output level.

 

1 hour ago, Reg19 said:

Also: if I do use the Intel PC Stick, do you recommend I upgrade the power supply to the Intel PC Stick?

 

You can try, but Holo Audio has isolated USB interface.

 

[micheloupatrick]

1 hour ago, Miska said:

 

What is your output rate? sinc-Mx adds about 1 second extra delay. With sinc-M it depends on your output rate.

 

Output rate is 44.1k x 128. Delay with sinc-M is less than 1 second, but with sinc-Mx the music stops playing 4-5 seconds after I hit the stop button in Roon (the 'time played' value in Roon stops immediately but HQPlayer keeps playing music for a few seconds).

[Miska]

4 hours ago, BrownMagic said:

My wife on the other hand likes to use spotify. Is there a way for me to beam spotify via spotify connect to the HQP NAA. I do not expect any upscaling or anything. I want to know if I can also install some linux based spotify connect receiver on the Intel Mini Stick. Can both of these coexist on the same stick?

 

Possibly yes, you could use for example minimal Ubuntu or Debian as basis, and install necessary packages there. networkaudiod package is straightforward, and likely Spotify Connect client too, although likely not available through the standard package repository.

[Miska]

30 minutes ago, micheloupatrick said:

Output rate is 44.1k x 128. Delay with sinc-M is less than 1 second, but with sinc-Mx the music stops playing 4-5 seconds after I hit the stop button in Roon (the 'time played' value in Roon stops immediately but HQPlayer keeps playing music for a few seconds).

 

Yes, sinc-M is always 1 million taps, regardless of rate. So at DSD128 it is about 1/12th of second delay (one million samples). sinc-Mx is 8 million taps at DSD128 rate, so about 3/4th of second delay. They are equal length at 16x. Then you need to add some buffering and processing block size things on top for the delay.

 

Try switching for example to poly-sinc-gauss-xla and you'll get notably less delay.

 

[micheloupatrick]

22 minutes ago, Miska said:

 

Yes, sinc-M is always 1 million taps, regardless of rate. So at DSD128 it is about 1/12th of second delay (one million samples). sinc-Mx is 8 million taps at DSD128 rate, so about 3/4th of second delay. They are equal length at 16x. Then you need to add some buffering and processing block size things on top for the delay.

 

Try switching for example to poly-sinc-gauss-xla and you'll get notably less delay.

 

OK, thanks for the explanation.

[jdubs]

New Desktop has to be soon, right?  Soooo looking fwd to that RIAA plugin :)

 

-Jim

[OzarkMtn]

13 hours ago, jdubs said:

New Desktop has to be soon, right?  Soooo looking fwd to that RIAA plugin :)

 

-Jim

I was thinking yesterday we might have seen Desktop but saw where @Miska needed to release a update to embedded version, maybe today??

Have read good things about the embedded release. Can’t wait - but have to .🤔

[Gilberto]

Hello everybody and sorry for introducing this topic here in the middle. Because my system is too revealing and on the bright side, and after trying various filters without reaching a clear conclusion, I would appreciate if someone with knowledge of HQP combinations could indicate the darkest possible filter combination. I upsampling all of my music to dsd512 and I love it, but I hate playing EQ on my system and would rather help with the filters in this software. Thanks for any feedback. Greetings to all ...

[ted_b]

11 minutes ago, Gilberto said:

Hello everybody and sorry for introducing this topic here in the middle. Because my system is too revealing and on the bright side, and after trying various filters without reaching a clear conclusion, I would appreciate if someone with knowledge of HQP combinations could indicate the darkest possible filter combination. I upsampling all of my music to dsd512 and I love it, but I hate playing EQ on my system and would rather help with the filters in this software. Thanks for any feedback. Greetings to all ...

What is your dac, and what about DSD512 do you love?  I ask cuz maybe 512 is pushing your dac to brightness, or your server to brightness?  All of the signal processing (dither, sample and/or bit rate, filters, cpu load) has an affect, so don't leave one out. 

[copy_of_a]

2 hours ago, Gilberto said:

Because my system is too revealing and on the bright side, and after trying various filters without reaching a clear conclusion, I would appreciate if someone with knowledge of HQP combinations could indicate the darkest possible filter combination. I upsampling all of my music to dsd512 and I love it, but I hate playing EQ on my system and would rather help with the filters in this software. Thanks for any feedback.

The "darkest" filters (at least for source files at 44.1 + 48kHz) are the poly sinc mqa variants.

But these also won't help if the source is HiRes (88.2kHz and above).

I fully understand that you don't want to play with EQ but if the system is too bright in your room I would suggest to use a convolution filter in HQPlayer (which actually is some kind of "EQ").

If you read about "room target curves" you'll find that a smooth roll off from the low end towards the high end is actually preferrable to a linear curve.

 

"Brightness" is in the range between 5kHz and 16kHz. Anything above is "air" and has a certain impact on definition, but not really on "brightness". On the other hand upsampling filters actually only affect the highest end of the frequency range of the source files. Therefore they do affect the overall representation - but really only in a subtle way. There's no (reasonable) upsampling filter that turns too "bright" music into vintage tube warmth or so.

A flat tilt filter fully preserves the relation of the instruments (unlike many EQs) but makes everything "brighter" or "darker" (depending on the filter you use). I'd suggest to give this route a shot instead of playing around with the different upsampling filters.

So what you need is flat tilt as a *.wav file (visualization attached - in this case the slope is 2db per decade) and load it in the convolution enginge of HQPlayer. You should try different slopes (steeper and flatter) and select the one you prefer.

I myself have stored 3 variants and switch them depending on the music (80th Pop/Rock can be really bright so I often use slighty steeper filter here).

 

 

 

[Gilberto]

Thanks guys for your opinion. What you told me, "ted_be", about upsampling to dsd512, since honestly after having read it here and in the Head-Fi forum that upsampling to dsd 512 is the holy grail because I have followed that path even though I am little friend of equalizations and manipulations is the delivery of sound, so sometimes, like you, I wonder if I am doing the right thing.
    My dac is a LKS 004 with the chip saber es9038pro, which is also shiny and I don't know if it has been a good choice because it seems that I have added fuel to the fire ...🥺

[oldfirm]

I ran the USBDeview and ASIO driver not listed? Only Xmos for the SMSL M500.

 I can play files from HQP playlist , imported DSD files  , but DSD is not displayed on the M500 screen.

Getting stuttering - So laptop - Ryzen 5 3500 overclocked ,  is not coping with HQP load .