HQPlayer and the NOS1

[manisandher]

Guys 'n' Gals,

I just thought I'd start a thread about feeding the NOS1 from HQPlayer rather than XX.

After a wasted (really wasted) couple of hours yesterday, everything is now working perfectly. I'm using XX to minimise the OS but keeping WASAPI services on (required for HQP).

So why would anyone be interested in using HQP instead of XX? I think there are a number of potential reasons:

1) HQP can convert native DSD64 and DSD128 files to PCM  on the fly and feed this to the NOS1 (from 16/44.1 up to 24/384 rates). I've been converting DSD64 to 24/352.8 and then feeding to the NOS1. So if you're getting into DSD downloads or recording, no need to get a dedicated DSD DAC if you already have a NOS1. I have a Mytek DSD DAC but find it very inconvenient to keep switching between this and the NOS1, as I don't want to destroy the sound with a preamp.

2) For a newbie, HQP is easier to get your head around than XX. The GUI is far from perfect though (especially file management, and there's no coverart, etc) but it really is very intuitive and easy to use.

3) HQP has some terrific filter and noise-shaping settings. I think it's worth it just for this. Playing around with various schemes, you'll really get a sense of how various filters and noise-shaping schemes sound. It makes a totally mockery of the 'perfect sound for ever' marketing hype for redbook - there is no  'perfect sound' - all the filters sound different and there's no single one that sounds obviously the best with all different types of music.

4) From a purely personal point of view, I hold HQP's developer in very high regard. Although he has a very different approach, his ethos is in many ways similar to Peter's - he's created a player and hardware solution for himself and is just sharing it with the world.

I'll be using this thread as a sort of blog to share my experiences and thoughts about the HQP/NOS1 combo.

HQP will never replace XX for me. But I think it supplements it well.

Mani.

[PeterSt]

Playing around with various schemes, you'll really get a sense of how various filters and noise-shaping schemes sound. It makes a totally mockery of the 'perfect sound for ever' marketing hype for redbook - there is no  'perfect sound' - all the filters sound different and there's no single one that sounds obviously the best with all different types of music.

Mani - AHUM.

If I didn't miss my own point, then most 101% probably you're missing mine. But, it is VERY nice that you just told what you did, because it might give some insight which otherwise would have been a difficult story (which would not be there in the first place). So ...

We must realize that each of the filters change the sound. No, wrong, they change (recreates) the digital data which creates the sound and therefore obviously the sound will change. But how ? it will be completely subjective as to whether you will like it. And as the next thing - you said it yourself - it shouldn't match all the music types (or just "tracks"). It is just a principle whcih I call wrong for a reason ... the data is changed.

But understand me well ... what happens which such filters is not technically wrong. Actually it is technically correct, but think ... the sole fact that you can choose out of more options ... how can it ever be correct. Only one can be, theoretically (for you to find out).

Arc Prediction does not recreate anything. It only interpolates as interpolation mathematically should work. So, data stays in-tact, but data points points are injected.

I think there will be one filter in there that has a chance not to imply the changes per music type, and this is what "Miska" will have named like something with "Spline Interpolation" in it. But, knowing Miska, he may have attemted to cascade this filter with another one behind it and then all is moot (again). However, this *not* assumed, I think this filter has the best chance to compare to Arc Prediction. And well, it should fail, because Miska already doesn't like it himself. If this is for genuine audible reasons (and not for measurement reasons) then Arc Prediction should win this game.

But ... it would be far more interesting to proove that with this filter you won't have the idea that it is not suitable for all music types etc. while you do have that idea for the other "normal" filters. And well, "not suitable" is wrongly expressed, because it should be about the equality.

This should go along with that with Arc Prediction you also don't have the idea that it's better for Jazz than for Latin or whatever, or at least it never should have slipped you mind before.

Lastly for now, notice that it would be the most easy for me to look at the performance of the filters via the analyser, but I tend not to look at other players, as I think you know.

Peter


PS: Absolute comparison will be diffcult already because of the necessity to listen to WASAPI with its own sound you obviously don't like to begin with, plus the adjacent lacking settings (other SQ parameters) not available.

[manisandher]

Actually, I think Miska refers to them as 'polynomial' filters. But you're right, he doesn't recommend them. And I agree with him - they sound fatiguing. My preference is his 'poly-sinc minimum phase' filter.

However... I'm only talking about comparing the filters in HQP to themselves. Much as I like HQP (for the reasons cited), I have a strong preference for XX and AP. And I agree, the single 16x AP setting is not found wanting with any type of music. It's seems to be universally the best filter to use.

BUT...

On CA, I mentioned the following article: http://spectralaudio.com/bulletins/0806.pdf

By 'recreating the sound' there may be filters that help to eliminate/reduce the distortions created at the ADC stage. A true interpolating filter such as AP wouldn't do this. Now, listening to AP and how great it sounds, it does seem that whatever these 'ADC distortions' might be, they aren't really important... but who knows until you've heard the sound without them?

Mani.

[Jud]

I'm going to rush in where even fools fear to tread (that is, into a discussion I'm vastly underqualified to be a part of), because of my great interest in the topic of filtering in particular and of good, accurate music reproduction in general.  (I don't know how widely such things are shared among different countries/cultures - the first phrase above is a reference to an aphorism, "Fools rush in where angels fear to tread.")

- Of course the goal is the most accurate music reproduction, and one of the best ways to tell if your system is accurate is to listen to lots of different things - instruments, types of music, etc. - and see if the system sounds better with certain types of music than others.  If it does, then it is imparting its own sound, and thus deviates from accurately reproducing only what it is given.

- Unfortunately we don't have some of the things demanded by the idealized mathematical model of Shannon/Nyquist, e.g., infinite time.  Therefore any system that samples music into digits and turns those digits back into music will be less than perfect.  There will be trade-offs involving phase shift, pre- and post-ringing, and undoubtedly other stuff I've never heard of.

- Given the premise that the filtering involved in the system won't be perfect, one must choose (or provide choices) between imperfect filters.  It seems to me this must inevitably be, to however (hopefully) small an extent, a matter of taste.  Phase shifts may drive one person crazy, pre-ringing another, and each makes the choices that minimize what bothers him or her most.

- When making (or providing) choices between imperfect filters, one should strive to guard against choosing the euphonic at the cost of the accurate.  But I imagine it is damned hard to tell which is which.  Does it sound good because less phase shift makes it more accurate, or is the touch of extra post-ringing making things "harmlessly" euphonic?

Mighty interesting stuff.

[manisandher]

By 'recreating the sound' there may be filters that help to eliminate/reduce the distortions created at the ADC stage.

Actually here's a snippet of a 2009 TAS article with Richard Fryer and Keith Johnson of Spectral Audio that got me thinking:

Richard Fryer There was lack of understanding among our peers as to why we spent so much time on a CD player. Don’t they understand that CD is now a heritage technology? This is where the music is. They dismiss it because many of the recordings are poor. I’m sorry, but there’s musical history in there. Although we’ve gone past the CD, it was obvious to us that we needed to pull out all the stops and do an “Apollo program” to build the finest reproduction of 16-bit digital.
Keith Johnson My instinct is that ultimately high-resolution will be a coming media, but when and what form it’s going to take is questionable. I’m wearing a record company hat [Reference Recordings] because we’re seeing this interest in it, and it does work. It’s got something a little better, but in my experience it’s not something that’s going to bash you in the face as being different, unless you’re using a converter that has taken into account what we have done in the SDR-4000. We’ve been using the high-res file as the reference. We ended up doing strange things to make the CD as close to the high-res as we could. That, of course, benefits a lot of old recordings.
RF Because we’re audiophiles there’s a tendency to say to ourselves, “Think of the heaven we’d be in with this high-resolution file.” But you have to continually put yourself back in the shoes of every-day music enthusiasts, where the recordings they love were made in any number of different ways, and many are compromised. What the public wants is great reproduction of their music. That’s everything from the Stones and the Beatles to everything recorded in the 60s and 70s. It’s all the stuff. We’ve tried to make a real contribution to keeping music alive by addressing issues that are relevant to music lovers and not just to audiophiles. If we can bring about a more ideal conversion of this massive body of work, everybody wins. To me the heroic effort was sort of archeological—that is, to take a medium that’s been beat around and abused and doesn’t have a great reputation, certainly amongst us purists, and raise it up and discover the really noble performances.
KJ I have a monaural Joan Baez LP that’s a gorgeous, beautiful recording of a great voice in its prime. If you listened to the CD years ago, it was like knitting needles in the ears. But if you listen to the same CD where the time-domain distortions have been reduced, you hear the LP in the CD. For 25 years we’ve learned not to expect from the CD what we hear in LP and analog tape. But when decoded correctly, it doesn’t sound digital. It sounds old and vintage, but it’s beautiful. That quality is on many recordings you wouldn’t expect because you thought the CD had been so badly encoded. That’s reversible to a degree.

[The highlights are mine.]

Mani.

[manisandher]

I'm going to rush in where even fools fear to tread...

Great post Jud.

I don't know why, but I've always been more fascinated with the ADC stage than the DAC stage. (I don't do the lotto, but if I did and won a few tens of millions I would get a team together and build the world's best ADC.) If a recording were encoded 'perfectly' then I think there would be no need for multiple filters - Arc Prediction would be all we need. Done. However, this just isn't the case...

Mani.

[Jud]

Mani, I'm just getting ready to post a topic over at CA on the "chain" from ADC through DAC.  Sending those waves through the aether to my brain again, are you?   

A quote from that Spectral Bulletin really appeals to me:

In a day and age of maturing digital audio technology, cost reduction and component integration, ( think of todays premium sigma-delta type DAC chips with self-contained filter programs which lower costs and reduce chip counts ) there is the unexamined belief that digital audio filter programs have attained a high degree of accuracy and refinement, especially in highend and studio applications. The unfortunate truth is, virtually all digital audio products from mid-fi to the most expensive high-end luxury and prosound units all use most of the same commodity filter technology.

[Emphasis mine.]

There's so little attention paid to this topic.  It seems to me most people must take lack of knowledge as a barrier to inquiry, whereas to me it's a spur.

[CoenP]

. That’s reversible to a degree.[/]

Like with Phase Alignment...

Regards, Coen

[listening]

And I know some recording engineers which watch their equipment and the combination of components as a treasure - including the recording locations. But it's still very difficult to reproduce the same sound again even at the same location. Virtual dgital locations are very easy to reproduce and some of the best sounding recordings are done in that way.

Georg

[stefanobilliani]

PS: Absolute comparison will be diffcult already because of the necessity to listen to WASAPI with its own sound you obviously don't like to begin with, plus the adjacent lacking settings (other SQ parameters) not available.

I don think HQ Player goes necessarily to WASAPI , if that is what you were talking in the PS .

[PeterSt]

I don't know. But since Mani took all this effort to get it going, something must be "truth" in there.

Peter

[esimms86]

Mani, one of the points you appear to e making is that DSD as played through the Mytek and DSD translated into PCM format and upsampled by the NOS1 are at least equivalent in the quality of the listening experience. I've never heard native DSD files but I've always read that they provide a unique listening experience. Is that not the case or is DSD-> PCM -> NOS1 simply good enough for someone with, admittedly, a high standard to begin with ? Put another way, do you find that you prefer one over the other and, if so, why?

Since you spent a number of hours toiling in futility before working things out, it would be great if you could post your HQ settings for the NOS1 to save the rest of us the trouble of trying to setup the same listening experiment. Thanks for once again taking the inspired road less traveled.

Esau

[manisandher]

An update:

A couple of days ago, tried HQPlayer again with the NOS1. I noticed a small button called '2-wire' that Jussi (Miska on most forums) had recently added in the settings area of HQPlayer. I engaged this and voila, I managed to feed the NOS1 with 32/768 from HQPlayer. Of course, XX has been able to do this since the NOS1 went USB, but it's nice that NOS1 owners now have an alternative to try.

Reading back through this post, I was struck by Peter's comments regarding Arc Prediction. I thought I'd try something. So here's what I did...

I got a 24/176.4 Reference Recording file. I played this natively through XX (all other settings per my signature). Nice sound. I then engaged 4x AP to take the rate to 705.6. The 'character' of the sound fundamentally changes. In a very pleasant way, but it changes nevertheless. Things start sounding more 'alive' - almost as if a tiny bit of reverb had been added to the sound. This sort of chimes with what Peter says about all albums now sounding like live recordings.

I then tried exactly the same thing with HQPlayer. Natively first. Very, very similar to XX. Maybe a tad 'darker', but really not a night-and-day difference. If anything I was expecting it to sound brighter and more 'metallic', the typical WASAPI sort of sound, but this was not the case. (I have no idea if HQPlayer applies any OS tweaks during playback - I don't think it does.) But the really interesting thing (for me at least) was when I applied the poly-sinc-short-minimum-phase filter to take the rate up to 705.6 and also 5th order noise shaping. The 'character' of the sound did NOT change. There was a clear improvement in pretty much every area of SQ, but without any added 'reverb effect'.

So, is the HQPlayer filter/noise-shaper doing a more accurate job than XX's own Arc Prediction? For me, it's opened my eyes to exploring different filtering options.

On a final note, all this is only possible because of the stupendous capability of the NOS1. For anyone who is still sitting on the fence about getting one, just do it. It's the surest bit of kit you'll ever buy.

Mani.

[PeterSt]

Hey, nice ...

What I like to add to this is that whatever way Miska applies to get 768 going is either :

- Very smart because no standard exists for this and he must have done it the same way as I did (which should be the best way);

- Is pure coincidence that it works (Miska has no access to a NOS1 as far as I know);

- Does not work (focus on no stereo or no proper stereo).

And let me also say that we both had zero communication over this.

Let's go for the first option. I sure dedicate Miska that ...

Peter


PS: And of course it is time to try it myself, although it would be the very first time I genuinely am tempted to do so. And oh, I did ask this politely a month or to back ...

[manisandher]

Hey Peter, I'm not sure between 1 and 2, but certainly not 3 - the stereo is no different from playing the 24/176.4 file natively. Miska describes it as '"Dual wire" channel bonding support to reach 2x rates, PCM and DoP (DSD)'.

Another thing that I will try in the next few weeks is playing back a DSD64 and DSD128 file (one of my own vinyl recordings or a suitable download) on the NOS1. HQPlayer will convert these to PCM32/705.6 on the fly. There are a bunch of algorithms and noise filters to try. Could be interesting...

Mani.