Which HQP Filter are you using? [2015-2023]

Thank you, didn’t know this feature.

@jussi_laako Having been in the upsampling universe for several years now with HQP and a DirectStream Junior which itself upsamples to DSD 1024, I’m curious to try a non-upsampling sound. I’ve ordered a Border Patrol R2R tube rectifier DAC that will arrive in a week or so. It maxes out at 24/96 and uses an old Philips chip (1541?).

What HQP settings are best for playing PCM and MQA (unfolded in Roon)?

Also, it is possible to set HQP to not upsample at all but still apply filters and dither?

Thanks

TDA1541A works up to 176.4 kHz at least, likely including also 192k. But old CD players used it at 176.4k with SAA7220 digital filter (upsampling) chip.

I would run it at max rate and TPDF or Gauss1 dither, and remember to set DAC Bits to 16! If you use for example poly-sinc-ext2, you can keep adaptive output rate unchecked and thus everything getting upsampled to for example 96k.

Yes, by selecting same output rate as source rate for each case. But running it such way doesn’t make much sense.

If upsampling to 384khz, is there a reason to use closed-form-M, sinc-M or sinc-XTR over other options?

It seems these three are the best filters, but is there a specific reason why one of those is the “best”? And if so which is it?

Also, at 384khz is it better to stick to LNS15 or move to NS9?

Curious why setting it to 16 bits is important if the chip goes to 24 bits?

The idea with having the no upsampling option is just to experiment with no upsampling. Specifically how do I do that? When I click on the sample rate drop down menu I don’t see an option for none.

Thanks.

Just having a mess about with PCM vs DSD again. I had a similar issue before and not sure how I rectified it.

When in PCM im using Sinc-M LNS15 and auto sample rate. From that im getting 352.8/384 (family ticked)

For DSD, ive tried Sinc-M again, ASDM7EC and auto sample rate. From that im getting dsd128.

Now, those DSD settings only work with 16/44, 24/44 and DSD formats. 48, 96 and 192 will not playback.

Im sure theres a schoolboy error from me somewhere there, but I just cant think

Ta

Select the same output rate as your source and there will be no upsampling. This requires you to manually change the output rate to match as the source changes.

If that sounds inconvenient, it is because no-one does this. If you don’t want to upsample then you wouldn’t use HQPlayer at all. The filters and dither are part of the upsampling process, not independent of it.

These are all at the long end of the filter choice. That means they are steeper in the frequency domain but longer in the time domain meaning transient response may not be as good as with other options. See this post above:

The thing to do is try each of the filters you have named, and possibly some others such as ext2. The one that sounds best to you in your system listening to your music is the one to use.

The issue is I find that the “best” filter depends on what i’m listening to.
ext2 I love the transients and soundstaging it gives but find that it can often be a tad dry.

What options would you recommend?
Essentially, If I had to just pick one filter, aside from personal preference, what are the pros/cons to the different filters in hqplayer?

I can’t seem to find much info about what might be better at what other than personal preference

1 Like

Sounds like you are using HQP 3 ?

I think ticking “family” means you are telling HQP to output multiples of the same sampling rate family as the source, however few DACs have 48k DSD oscillators. Try unticking family and HQP should convert to the closest rate your DAC can handle.

Thanks. It’s just an experiment! I’m questioning whether or not upsampling provides the most natural sound.

1 Like

Oversampling/Upsampling is somewhat inevitable.

Your dac is almost certainly an oversampling dac anyway.
And if you wish to use NOS, then that has its own analog low pass filtering anyway

Yes, it’s a NOS DAC. Border Patrol SE-i. Wanting to experiment with as little upsampling as possible.

NOS sound will depend on the analog implementation of your dac, and so will vary from dac to dac. Typically NOS is known for sounding “faster” particularly in the lowend. But also darker

1 Like

Hi,

I am using 4.18

Just tried de selecting “Family” and still doesnt work with Sinc-M. Ive just tried poly sinc ext2 and thats fine. I wonder why that is? I do prefer the sound of Sinc-M in DSD to Ext2 tbh

EDIT - Ah, Ive just selected 48k DSD and now we are up and running…thank you

What I do notice is that the CPU (All cores average of 6 cores) can be as high as 60% and the Antipodes CX gets quite hot, wheres as Poly Sinc etx2 has lower CPU load using DSD and 352.8/384 PCM Sinc-M is very low at 10%

Is 60% anything to be worried about?

Percent usage isn’t a problem of itself. What CPU temps are you seeing ?

What DAC are you using ? Forcing 48k multiple DSD on a DAC which doesn’t actually have a 48k oscillator results in a slowed lower pitch to the music.

Hey,

Hmmm not sure of the temps, I’m just aware of the percentages. How do I check?

My DAC is an Aqua La Scala mk2 Optologic r2r DAC. Now I know that it’s best to use pcm for an r2r but dsd12 sinc-m and asdm7ec Sounds super impressive.

Also, I was advised to use DAC bits “20” for my a DAC by Jussi. Does that make a difference here when trying to playback 192 in dsd?

I use CAM to monitor my Windows 10 server.

The pitch issues when forcing a 48kHz family output onto a 44.1kHz oscillator are pretty obvious. If you’re not hearing them then your DAC has the 48kHz oscillator. I couldn’t tell from the website or spec sheet whether it did or not.

1 Like

I can’t tell a difference tbh. If there is it’s not “obvious”. I’m not actually sure what forcing 48khz does but would make sense to select it if I want to convert a file from a 48/96/192 to dsd…but I thought that’s what “family” did, but family doesn’t work for me using dsd

TDA154x is 16-bit part, that’s why. It can only understand up to 16-bit, rest are just thrown away.

Many USB interfaces in such DACs however claim to have 24 or 32-bit. This results in those extra bits being thrown away (truncated) which causes distortion. For this reason it is important to make sure you don’t send more bits to the DAC than the actual D/A conversion stage can deal with.

If you still want filter being applied, you need to select output rate from Client to match source rate of the material. Alternatively you can set filter to “none”, in which case that part is skipped altogether.

Those settings in HQPlayer Settings-dialog are only startup defaults. You can control them during runtime also.