Miska

Yes, it will, in next release. Working on a release right now!

It improves correctness of the output signal, explained here, along with the currently supported DACs and output rates: https://signalyst.com/dac-correction-support/ More is coming over time, when I get those done.

You are then likely playing in playlist mode. You could try playing one album from HQPlayer's library in album mode and see if it makes a difference.

Please uncheck the "Verbose metadata" in Client control panel!

The older sacd_extract is known to set DSF sizes incorrectly, so that the tails contain unnecessary data. Which results in snap-crackle-and-pop. I don't know about the other one. Are you able to also reproduce this with most commercially available DSD content, for example from nativedsd.com? Are you playing in album mode or in playlist mode? Pop suppression algorithm in playlist mode can deal with minor pops, but not large ones.

Good that you reminded me to update that! Download buttons are now for Fedora 39 since Nvidia finally moved forward. Embedded is still supported on both 37 and 39 though. But your point is of course still correct, if one wants to use HQPlayer on Linux, I would recommend not to upgrade to the fresh new release until HQPlayer has moved forward. And I can do that earliest when Nvidia does so (if we want to keep CUDA offload capability on Fedora as well).

Higher the filter order, longer it's impulse response is and thus more it rings. So steeper the filter, more it rings (a bit simplified).

Server hosting HQPlayer Embedded is on 500 Mbps internet connection. I have one server also in Hong Kong for some of the things, and likewise my connection to that one is rather slow. Hard to say what kind of things are going on on the routes the traffic happens to take for long distances.

No, you just have multiple configurations you can switch. That's perfectly fine to do.

When Direct SDM is active, only DSP you can perform is speaker delay processing. Everything else is disabled. If you have matrix processing enabled, Direct SDM is ignored apart from the locked volume control. Processing DSD256 -> DSD256 is with convolution or cross-feed is rather heavy operation, so in your case M1 runs out of steam doing that. It would need more processing power. In particular some operations can be slow on Apple silicon due to some hardware differences to Intel architecture. Please check that you have integrator set to for example FIR2. My M1Max 32GB can do DSD256 -> DSD256 with cross-feed without issues. But convolution at DSD256 is too much for it (of course always depends on the particular convolution filter though.

For the correct transient waveform shape yes. Are we now talking about fundamental of the transient, or something else? Since many transients don't have any fundamental frequency. I don't agree with that either. https://en.wikipedia.org/wiki/Ultrasonic_hearing In addition, it is too simplistic to view transient response from plain frequency response point of view. Limiting signal bandwidth to 20 kHz has much wider repercussions than just some disappearing frequency components. Do you hear any differences between any digital filters?

In playlist mode, HQPlayer is performing de-popping algorithm since the tracks are unrelated. This also means it is not completely gapless. In album mode it is assumed that that all the content belong together and are in correct order, and thus this is not needed and the playback can be completely gapless without the de-popping algorithm. So if you get pops in album mode, then you know those tracks have pop issues.

No, I certainly don't agree. No, I don't agree. For best transient reproduction we need at least 50 - 100 kHz bandwidth, depending on instruments. Real DSD256 recordings are best you can get these days, those have the least bandwidth limiting.

Some testing results from today, for those who seek to build lower TDP servers for HQPlayer. My i9-13900T with DDR5-6400 CL32 memory can do DSD1024 with ASDM7EC-light when E-cores are set to filter. While i9-14900K can do the same with ASDM7EC-super. Both cases without GPU's help. So the E-cores offload helps quite a bit this case, since it enables the T-model to reach DSD1024.

Strange, because single tap has been working for me. Auto play is not related to this, it is a feature to let Qobuz automatically continue playing similar music when end of your current queue is reached. I'm using Client on Microsoft Surface tablet with Windows 11. And with iPad. I also have a Linux laptop with touchscreen. This is shown in the album view (cover flow)? I have personally added the version to the album title in HQPlayer's library database. Since I also have for example four or five different versions of Pink Floyd's Dark Side of the Moon album. OK, something I've never used... 😅