HQPlayer Embedded Discussion [2021-2022]

Guess we will soon have a plain explanation from the grand master himself. :slight_smile:

1 Like

Iā€™ll learn something new so I appreciate your question.

1 Like

When it comes to HPQe it has come to dominate my HiFi life to a degree where one could really question if I am loosing it ā€¦
I wish I had far better knowledge about the filters as such. It seems that people, self taught or educated, are much more familiar to the doā€™s and donā€™ts, but I am old enough not to be embarrased for asking silly qustions.
The breakthrough came when I compared my rig to those who got the money to pay big time. Still their equipment was not able to master the noise in music, a part that seems to become a familiar part of the sound so when gone, peoples first reaction is that somethingā€™s wrong. I often hear the reaction ā€œlack of dynamicsā€. I usually do not reply, I simply grab the remote and increase to 0dB and look for the reaction when transients go from 0 to 100 in a split second without any tendency of distortion at all. When the noise is gone, the amp energy reserve is adequate to so much more and there is no smearing out of the transient peaks. With just a small DAC from RME bought pre-owned fƶr Euro 400:- while others cost a number of zeroes moreā€¦ :slight_smile: Life is sweet

3 Likes

As alluded to above, the gauss family of filters is inherently two stage, thus why no -2s choices.

1 Like

It is a technical decision to use 2x higher ratio for first stage, compared to -2s, on ext2, ext3 and gauss and keep those still two stage.

There could be single stage version of those too, but for technical reasons I have not done it.

3 Likes

And for the difference, while big boysā€™re showing-off, my fanless NUC on Intel Core i7-8665U, running Ubuntu Server + Roon Server + Latest HQPe (amd) Version 4.29.0 direct USB into MAY

Back in October, when I just installed HQPe on my NUC I wrote the following:

Today same hardware runs ASDM7ECv2@256! I can not do gauss-xla or ext3, these are shuttering. But gauss-long and ext2 can do all common PCM rates from 44,1/48 to 192 smoothly!



It feels like I am rediscovering music again!

And I keep falling to Ā« short-mp Ā» over and over and over again. It feels almost like it was specifically designed in conjunction with 7ECv2 or vice versa. With SDM@256 it gives phenomenally realistic leading edge. Punchy and crunchy percussions with full body. Terrifyingly real, especially in Jazz recordings bass+drums, so real with body and timbre, itā€™s even frightening.

Iā€™m actually wondering what do I misunderstand or mishear in EXT or Gauss familiesā€¦ drums are not that punchy, that is certainā€¦

3 Likes

Nothing, different genres need different emphasis on filter design aspects. And different people are sensitive to different aspect of sound too, so filter choice is always personal too.

Iā€™ve used short-mp for prog rock for a long time. ext2 was a surprise to me that it was also fit for this genre, sort of unexpected result. My number one test track when designing short-mp was San Tropez track from Pink Floydā€™s Meddle album (specifically the first CD edition of it).

Other test tracks, a bit heavier material though:
Anesthetize from Porcupine Treeā€™s Fear of a Blank Planet album (and the album as whole)
Eternal Rains Will Come from Opethā€™s Pale Communion album (and the album as whole)
(for these two, you may also notice common denominator - Steven Wilson, he has good remixes of Yes and other prog rock material too)

Anesthetize is very well known song from itā€™s complex drum composition. You can have also fun watching YouTube videos of people playing it cover.

Then some longer filters seem to work nice for example for Daft Punkā€™s Radom Access Memories album (88.2/24 hires version).

2 Likes

With the new DAC I will stroll around the filters again. Tried the poly-sinc-xtr-short-mp just now, and ā€¦ it was good drive, fast transients and silence as well ā€¦ hmmmm
The mapā€™s changed a bit with RME.

Fiddling around among the short-mp-filters available. And, got to give you cred for your insights. There is some more ā€œnerveā€ in the music and less ā€œshoe shineā€ ā€¦ Exciting, this!

1 Like

It felt somewhat naturally - Prog stands out! Mustā€™ve shared my Ā« full Ā» screen for PCM48 Ā« test Ā», what a coincidence, hahahaha!

Besides with newly discovered unexpected possibilities of my modest NUC, a bit earlier iā€™ve been also enjoying Mr. Wilsonā€™s Ā“To The Boneā€™ and Ā“Raven who refused to singā€™. List goes onā€¦ I probably must continue to HQP filters thread :slight_smile: Some things from my play lists has been totally reimaginedā€¦

3 Likes

Hi @jussi_laako

For ALSA backend why does web config page say default = ā€œ0ā€ for DAC bits and buffer time - but the actually default shown is 24 bits and 100ms buffer.

For RME ADI-2 - should I change bits to 32 bits?

Or simply change both values to ā€œ0ā€

ā€œdefaultā€ for bits means what ever the hardware claims.

ā€œdefaultā€ for buffer time is to use what ever is HQPlayerā€™s built-in default value at a time.

Default configuration file has values differing from these in attempt to ensure sensible results in most cases.

For bits, set it to ā€œ0ā€. For buffer time, Iā€™d recommend to stick to 100 ms.

1 Like

Hi Jussi
HQPlayer Desktop for Mac wonā€™t allow to change (grey out) the CoreAudio buffer time. On RAVENNA network the HQPlayer Desktop for Mac sounds really bad, very noisy, audible pops all the timeā€¦ On the same machine (2020 Intel i5 MacBook Pro 13) has no such problem under Windows 10. Please advise.


Iā€™ve put the RAVENNA buffer to max value already.

CoreAudio determines the buffer time. In the end it is up to the driver. If Windows works better, then it may be better to use that instead.

But build-in DSP of Roon and Audirvana for Mac doing upsampling to DSD playing over CoreAudio RAVENNA have no such issue, tooā€¦

No issues with my Anubis

This is the Embedded thread but Iā€™ve used Desktop also without issues

Maybe those are lighter algorithms and you are just running out of CPU power with HQPlayer algorithms (which you didnā€™t state).

Its a very long thread so if @dabassgoesboomboom @jussi_laako or anyone can simply tell me again, can I plugin my iphone and play apple music to raspberry pi 4b with HQPlayer embedded and upsample to 768k to DAC? I supposed this would work with auto rate switch from iphone?

iphone > Apple camera adaptor > USB-B > Raspberry pi 4b > USB-Bā€“USB-A > DAC

Or I should replace raspberry pi 4b with upboard to enable auto rate switch?

iphone > Apple camera adaptor > USB-B > Upboard > USB-Bā€“USB-A > DAC

I have only tested this functionality on UP Gateway. Theoretically it should work on some other hardware too. But depends a lot on the low level USB device interface implementation and driver.

1 Like

Yes this is what I do - works fantastically well

Its not a USB B that feeds the UpBoard - itā€™s a micro B OTG port

Same cable type that comes with external hard drives normally

You can run Embedded on the UpBoard but it will be limited to PCM upsampling

What I do is use the UpBoard as an NAA input for my more powerful 11900K HQP Embedded server which is in a different room