[clark8888]

JPlay IOS can not stream Tidal to HQPlayer Embedded 4 by upnp (with red alarm: error sending action to renderer). JPlay IOS can stream Tidal and Qobuz to HQPlayer Embedded 5 by upnp. BTW, Mconnect can stream both to HQPlayer Embedded 4 and 5.

 

[Rayon]

I noticed that while ASDM7EC-super seems to pair best with Susvara, with TC I still prefer ASDM7ECv3. I mainly listen to very fast music, like metal, with TC and ECv3 feels like a faster/snappier filter.

Btw an update on ASDM7EC-super with 13900k. I had overclocked it to run ASDM7ECv3, but I have switched to using super instead. I had some stability issues with OC, so I removed the explicit OC (changing ratios). ASDM7EC-super works just fine. CPU load is 180% @ DSD1024 (44.1k family). GPU handles the filters (cross-family oversampling for 48k content), leaving modulator for the CPU.

EDIT: Well this is a positive surprise! ASDM7ECv3 seems to work without OC as well now. I wonder if Jussi has done some performance optimizations in 5.1.0

 

[dusk]

I am thinking about grabbing a SRC-DX to hook it up to a Hugo2, what cables did you use? Are they the same as a TT2? I am planning on upgrading to a TT2 in a couple months.

What modulator/filters do you like best for PCM?

I heard HQPlayer > DSD > Chord products is not ideal because it will convert back to PCM and destroy the DSD benefits, is that true?

 

[shwnwllms]

I heard HQPlayer > DSD > Chord products is not ideal because it will convert back to PCM and destroy the DSD benefits, is that true?

Yes. You’ll want highest rate PCM instead (but the real magic happens at DSD256 and beyond)

 

[dusk]

ya, I am getting a Holo Audio May KTE for NOS and will run DSD on that, and leave the Hugo TT for PCM as the general consenus on forums is that's the way (but I'll test it a bit more)

DSD now on the chord is a much more relaxed and flat presentation to my ears, transients don't hit so hard, things don't attack me, I listen to it more at night when I'm trying to relax

PCM for when I want to wake up

 

[Failed Engineer]

Can the HQPlayer library do auto updating scans? I didn't see anything about it in the documentation.

 

[bogi]

It is intentionally not automatic. You can manually re-run Scan at any time you wish.

 

[Patu]

I've been using HQP for around a year now. It runs on my server machine in another room where my listening happens. I stream over to Holo Red with HQP NAA image and from there with USB to Holo Spring 3. I have the older HQP 4 license on my desktop PC to be used with my modest headphone rig when I listen to music on my PC. HQP 5 license runs on my server machine and serves my main speaker rig.

I've experimented with lots of different setting combinations and can't really say which one is ultimately the best. It depends a lot about system synergy and own preferences. I currently play with DSD256x48 - ASDM7ECv2 (I actually slightly prefer the older v2 sound to the newer v3) - IIR2. My long time favorite combo was DSD256x48 - ASDM7EC-super - poly-sinc-gauss / poly-sinc-gauss-long. ASDM7EC-super and v3 are both slightly sharper sounding than the older v2, which sounds fuller and slightly smoother IMO. My current server machine (Lenovo Thinkcentre M70q Gen2 SFF with i5-11400T) can do DSD512 but only with very limited modulator and filter options. Modulator is easily as important as the filter itself. Difference between DSD256 and DSD512 is smaller than the difference between different modulators and filters. I think DSD256 is kind of a sweet spot when it comes to the bitrate selection. I might upgrade my server in the future but currently playing happily with DSD256x48.

 

[plumpudding2]

Modulator is easily as important as the filter itself. Difference between DSD256 and DSD512 is smaller than the difference between different modulators and filters. I think DSD256 is kind of a sweet spot when it comes to the bitrate selection. I might upgrade my server in the future but currently playing happily with DSD256x48.

Jussi has stated that in-band noise performance is optimal at 256 for Holo dacs, and that 512 trades 3db of in-band noise performance for 20db ultrasonic noise reduction. That corresponds to you saying 256 is the sweet spot.
Measurements aren't everything though, there's a few people here that swear by DSD1024, even with a non-EC modulator if you can't run these.

 

[bogi]

I have the older HQP 4 license on my desktop PC to be used with my modest headphone rig when I listen to music on my PC. HQP 5 license runs on my server machine and serves my main speaker rig.

You can use v5 license on both if you are not listening on both at once.

ASDM7EC-super and v3 are both slightly sharper sounding

ASMD5-super may be also worth of try, if you find it a bit less sharp.

You can try also the 512+ versions of EC-super at DSD256, especially with source content up to 96k sample rate. 512+ modulators bring higher dynamic range. Miska wrote: "Remember that for example 96k sampling rate PCM has hard bandwidth limit of 48 kHz. So running 7EC512+fs at DSD256 already exceeds well beyond what 96/24 PCM can do." https://audiophilestyle.com/forums/topic/19715-hq-player/?do=findComment&comment=1246430

 

[rkt31]

Sinc long skips with i7 processor during pcm to pcm 16x upsampling. Sinc medium works fine. Any idea of proper setting for sinc long to work?

 

[genefruit]

Sinc long skips with i7 processor during pcm to pcm 16x upsampling. Sinc medium works fine. Any idea of proper setting for sinc long to work?

more compute

 

[rkt31]

more compute

You mean it won't work with i7 computer or there is some setting which will work? I found sinc medium to be best sounding so may sinc long is even better.

 

[genefruit]

You mean it won't work with i7 computer or there is some setting which will work? I found sinc medium to be best sounding so may sinc long is even better.

I'm not aware of any setting that is going to work. One needs more compute power. I have an i9-14900K machine (windows) I built this weekend and sinc-medium @ 1.536 is no problem. Sinc-long continually stutters @ 1.536 and @ 768 but runs @ 705. All based on 1x original content. 4090 Nvidia card arrives tomorrow and I'll provide new results once installed and tested.

Edited for additional content - please see https://community.roonlabs.com/t/which-hqp-filter-are-you-using-2023/244185/414

What’s so special about these new sinc filters as compared to the sinc filters we already had before? Their number of taps is not mentioned in the manual.

These can be used also to convert between rate families, unlike the other ones. But otherwise they are related to sinc-L family.

Therefore, if you change the toggle to match rate family, one could theoretically use sinc-L family. My test allows me to use Sinc-Ll @ 1411.2 with 44.1 content and Sinc-Ll @ 1536 with 48 content.

 

[bogi]

Sinc long skips with i7 processor during pcm to pcm 16x upsampling. Sinc medium works fine. Any idea of proper setting for sinc long to work?

Longest filters need not to be best sounding. They may bring more weighty sound and better timbre with slower music content (classical etc.), but they have also drawbacks. Since the actual sample value is computed from very long series of previous and next samples, long filters introduce a smearing, which is audible especially in faster genres. On systems, which are well treated against computer noise, short to middle length filters may sound better with pop/rock recordings. But of course, it is very individual, it is about what for sound a listener likes.

Are you using HQPlayer 5.20? sinc-short, -medium and -long were converted to 2 stage filters and now require less computer resources.

If sinc-long is not working properly, as stated above, staying in the same sample rate family (44.1x or 48x) as source content may help. Conversion between families requires more computing resources.

Mentioning i7 is not telling enough about performance. 6th gen vs 13th gen brings much difference.