Jump to content

Extreme filtering software upscaling


Recommended Posts

2 hours ago, rmpfyf said:

Still just fine for room correction. 

 

Well, it depends what the outcome intended is. 

 

I have spent over 100 hours building my own Linux for audiophile use. I enjoy it, I get a kick out of it. SnakeOil OS does a lot of what I do, it does a lot of other stuff too, is very slick and totally free. If someone was to want to steward rolling some of @Ittaku's work and other extant DSP stuff into a Linux distro that did just that and was a kind of SnakeOil OS-for-DSP intended to sit on a separate PC, I'd probably contribute $$. And I'd think there's a market for it just because HQPlayer isn't free and multi boxes are expensive, but there is a market for both, and many of us have a spare PC kicking about.

 

If I was selling active speakers and looking for a solution capable of DSP work for whatever application I'd sooner spend time with a Danville-type organisation through their OEM channel (replete with consultancy) than DEQX - there's more functionality, less cost, more margin, more customisation and more performance to be had, though that's just me. This being said there are other motivations at play - some customers, and I imagine there are a lot of them - really just want a DEQX box.

 

Again, fast times.

I admit to liking DEQX boxes partly because they are very convenient to use with someone else having done much of the hard work. However I have not found them entirely straight-forward as the final SQ is very dependent on the initial measurements one makes of the raw drivers (which seems very sensitive to the postion of the microphone etc) and the choices one makes as to positions and slopes of Xovers etc. But I suspect this will be a problem no matter which DSP system one uses - even though the changes can be made quite quickly via computers, it is very time consuming to optimise SQ with many variables.

 

However my choice of DEQX is also partly because of the SQ they are capable of - and of course this is due to more than just room-correction. Alan Langford's personal/home active-speaker system using an HDP5 is the best I have heard EVER regardless of technology - and certainly better than my best passive system with the Qutest and either Simon's M-Scaler or @Ittaku's extreme upsampling.

 

So I am obviously intrigued by the possibility of combining best upsampling with best active-DSP systems and despite my current reluctance I will probably follow your advice - though not until after I have received my M-Scaler and see how it integrates with my DEQX-active speakers - and after I have recovered from the past month or so experimenting with upscaling that I found very useful but time-consuming!

  • Like 1
Link to comment
Share on other sites



17 minutes ago, Music2496 said:

@powerav after you've disabled auto family rate, when you play PCM44.1/88/176 does the DAC display 768? It should.

 

And then when you play PCM48/96/192, it should still show 768 too.

It doesn’t produce static when I set the up sampling rate to 512 and I change between cd  and hires tracks no problem only when I select the higher 700 settings 

Link to comment
Share on other sites

1 minute ago, powerav said:

It doesn’t produce static when I set the up sampling rate to 512 and I change between cd  and hires tracks no problem only when I select the higher 700 settings 

 

When you select PCM512, I assume the DAC is displaying either 353khz or 384khz?

 

I mentioned earlier it's an issue at PCM705/768, so this is consistent.

 

Try re-installing the Pro-Ject driver. It should install the ASIO drive and this option to be available to you in HQP.

 

If you can select ASIO (might need a PC reboot) then go back up to PCM768 and re-try.

 

I asked earlier - which filter/s in particular are causing static?

Link to comment
Share on other sites

Guest rmpfyf
7 minutes ago, legend said:

the final SQ is very dependent on the initial measurements one makes of the raw drivers (which seems very sensitive to the postion of the microphone etc) and the choices one makes as to positions and slopes of Xovers etc. But I suspect this will be a problem no matter which DSP system one uses - even though the changes can be made quite quickly via computers, it is very time consuming to optimise SQ with many variables.

 

Correct... it's a time and detail game.

 

8 minutes ago, legend said:

However my choice of DEQX is also partly because of the SQ they are capable of - and of course this is due to more than just room-correction. Alan Langford's personal/home active-speaker system using an HDP5 is the best I have heard EVER regardless of technology - and certainly better than my best passive system with the Qutest and either Simon's M-Scaler or @Ittaku's extreme upsampling.

 

For what it is DEQX is a sorted system and deserves its brand reputation. I don't think I've ever heard anyone say a bad thing about their product or service. Many happy customers.

 

10 minutes ago, legend said:

So I am obviously intrigued by the possibility of combining best upsampling with best active-DSP systems and despite my current reluctance I will probably follow your advice - though not until after I have received my M-Scaler and see how it integrates with my DEQX-active speakers - and after I have recovered from the past month or so experimenting with upscaling that I found very useful but time-consuming!

 

The nice thing about having a business as your own is that you could genuinely make enquiries of a DSP vendor and see what's possible relative to your specific outcomes, and get at least tangible answers not privy to the rest of us plebs. 

 

The only thing I'd be wary of with the actives is the resample workflow - you'll have the M-Scaler doing a very high frequency digital out, then into a DAC, then back to digital through the DEQX? Or if digital all the way through, I assume the DEQX will resample at some point to maintain a 96kHz or 192kHz workflow? Would seem what you'd really want is a multichannel DAC at your resample frequency... or did I miss something? The real challenge for me would be what a fully-active 784kHz workflow looks like (a few people here saving PCM1704U-K's for a rainy day are having NOS dreams surely...)

 

Don't get me wrong, if I could afford an M-Scaler, I would. It's a good idea, executed capably. 

Link to comment
Share on other sites

4 minutes ago, Music2496 said:

 

When you select PCM512, I assume the DAC is displaying either 353khz or 384khz?

Yes that is right.

actually I didn’t install the driver as it automatically came up so yes....see I am used to using Libreelec which you don’t need drivers. Dumb me I was too keen. I will have to try in the morning as I need to go to bed as I start work at 6, thanks for your help I will report back tomorrow.

  • Like 1
Link to comment
Share on other sites



Just now, powerav said:

Yes that is right.

actually I didn’t install the driver as it automatically came up so yes....see I am used to using Libreelec which you don’t need drivers. Dumb me I was too keen. I will have to try in the morning as I need to go to bed as I start work at 6, thanks for your help I will report back tomorrow.

Ah that's probably the problem... you definitely need to use the Pro-Ject driver.

 

Once you install the official Pro-Ject driver (and reboot PC), you can try WASAPI (in Exclusive Mode) or ASIO.

 

Try setting to PCM768kHz again and you can enable 'auto rate family' again.

 

That should work properly.

 

Link to comment
Share on other sites

23 minutes ago, rmpfyf said:

Don't get me wrong, if I could afford an M-Scaler, I would. It's a good idea, executed capably. 

If it supported 705/768 on other inputs, I would buy one too, and not be fiddling with this project myself. However I have no intention of buying a Chord DAC to get its improvements as that would be 10 steps backwards to take one step forwards. That said, fiddling with software is hardly hard work; I'm doing it mostly because it's fun :)

  • Like 1
Link to comment
Share on other sites

9 hours ago, rmpfyf said:

The only thing I'd be wary of with the actives is the resample workflow - you'll have the M-Scaler doing a very high frequency digital out, then into a DAC, then back to digital through the DEQX? Or if digital all the way through, I assume the DEQX will resample at some point to maintain a 96kHz or 192kHz workflow? Would seem what you'd really want is a multichannel DAC at your resample frequency... or did I miss something? The real challenge for me would be what a fully-active 784kHz workflow looks like (a few people here saving PCM1704U-K's for a rainy day are having NOS dreams surely...)

 

Don't get me wrong, if I could afford an M-Scaler, I would. It's a good idea, executed capably. 

Not sure what will happen when I connect the M-Scaler to the DEQX - will find out when it comes hopefully later this week but probably early next week and let you know.

 

Fortunately I can get a dealer's price and also claim back GST (and as a tax expense) - there have to be one advantage of being a loudspeaker manufacturer these days!

Link to comment
Share on other sites

Apparently my 8805 may re digitize my analog from my dac, but even so, the difference between having and not having the m-Scaler is more than enough. It’s more evident with the Brooklyn.  

Edited by Sime V2
Link to comment
Share on other sites

Guest rmpfyf
1 hour ago, legend said:

Not sure what will happen when I connect the M-Scaler to the DEQX - will find out when it comes hopefully later this week but probably early next week and let you know.

 

Fortunately I can get a dealer's price and also claim back GST (and as a tax expense) - there have to be one advantage of being a loudspeaker manufacturer these days!

 

Lucky on the GST :)

 

Just to be clear - the last conversion on your actives is at 192kHz (DEQX) or at 768kHz (@Ittaku)? If the former you'd have to wonder what the latter sounds like, which is (I'd think) what the author is getting.

 

Don't get me wrong there's some advantage, though your filters effects (temporally) are going to run longer - should sound sharper, more separated at higher freq. It's not going to sound terrible or muddled as you're having the same filter characteristic run to all speakers, though it's neither the full advantage.

Edited by rmpfyf
Link to comment
Share on other sites



4 minutes ago, rmpfyf said:

Just to be clear - the last conversion on your actives is at 192kHz (DEQX) or at 768kHz (@Ittaku)? If the former you'd have to wonder what the latter sounds like, which is (I'd think) what the author is getting.

 

Don't get me wrong there's some advantage, though your filters effects (temporally) are going to run longer - should sound sharper, more separated at higher freq. It's not going to sound terrible or muddled as you're having the same filter characteristic run to all speakers, though it's neither the full advantage.

I don't know about his actives. I put the scaler after my DSP and before my DAC, but mine is all analogue after the DAC and derived benefit there.

Link to comment
Share on other sites

Guest rmpfyf
Just now, Ittaku said:

I don't know about his actives. I put the scaler after my DSP and before my DAC, but mine is all analogue after the DAC and derived benefit there.

That's the way to do it. 

 

Gets a little trickier with actives in that I'm yet to see a 768kHz multichannel DAC at a reasonable price, though a quartet of S2's might do it :D 

Link to comment
Share on other sites

14 hours ago, legend said:

However I have not found them entirely straight-forward as the final SQ is very dependent on the initial measurements one makes of the raw drivers  <SNIP>  But I suspect this will be a problem no matter which DSP system one uses

Definitely.

I don't see this as a reflection on the DEXQ, as much as it is a reflection of speaker design in general.

 

It comes back to the basic question of:    "Is a correction being applied to the driver based on measurements which aren't representative".

 

As we increase the capability to "correct things" (eg. by using IIR and FIR digital filters) .... we increase the ability to "incorrectly correct things".

 

When building a speaker with a passive crossover filter (ie. resistors capacitors conductors) .... the same issues are in play ....  it's just that the capability to make corrections to the driver is limited - meaning in many ways it's harder to make "mistakes".

 

14 hours ago, legend said:

So I am obviously intrigued by the possibility of combining best upsampling with best active-DSP systems

The DAC used in the S2 (9038pro) is actually an 8 channel part.... but it seems very few people have actually implemented it that way.   Acko and ? TwistedPear might have something, but I assume you mean not-DIY.

Link to comment
Share on other sites

Currently (before M-Scaler arrives) I am using HQplayer to upsample to 192k before DEQX. As I understand it DEQX then downsamples this to 96k so its Sharc chips can do their DSP and create 6 channels of digital output (2xBMT) then DEQX upsamples  these to the original incoming sample rate for its internal DACs (hence why I initially upsample to 192k).

 

However the DEQX can output its 6-channels digital after DSP (before its DACs) so one could do the upsampling there (by 3 M-Scalers!) and then use 3 external DACs.   Project S2s would be very cost effective for this as I already have one of them.  Or a 6-8 channel DAC as @davewantsmoore suggests.

 

But I bought the M-Scaler mainly so I can hopefully forget about the front-end in the development of passive speakers! Proper upscaling for my DSP active-speakers probably requires @rmpfyf's approach or something similar?  However I will also play about with the M-Scaler when it arrives in DEQX-active systems.

hdp4-back-3.jpg

Edited by legend
  • Like 1
Link to comment
Share on other sites



Guest rmpfyf
46 minutes ago, davewantsmoore said:

The DAC used in the S2 (9038pro) is actually an 8 channel part.... but it seems very few people have actually implemented it that way.   Acko and ? TwistedPear might have something, but I assume you mean not-DIY.

 

Not sure it’ll run to 768kHz in 8 channel mode (or will it?) Audial is about to release something based on it soon at 8 channels.

Link to comment
Share on other sites

20 hours ago, Music2496 said:

Ah that's probably the problem... you definitely need to use the Pro-Ject driver.

 

Once you install the official Pro-Ject driver (and reboot PC), you can try WASAPI (in Exclusive Mode) or ASIO.

 

Try setting to PCM768kHz again and you can enable 'auto rate family' again.

 

That should work properly.

 

No unfortunately that didn’t work still have white noise if a different resolution is played, all works fine on the 512 k setting bit not on 768. I have also tried many filters, frustrating and a shame as it sounds magic on 768 here is a screenshot of settings.

12D622D9-2C41-4A5B-999A-9044E808A2E5.jpeg

Link to comment
Share on other sites

20 minutes ago, powerav said:

No unfortunately that didn’t work still have white noise if a different resolution is played, all works fine on the 512 k setting bit not on 768. I have also tried many filters, frustrating and a shame as it sounds magic on 768 here is a screenshot of settings.

12D622D9-2C41-4A5B-999A-9044E808A2E5.jpeg

Use the Pro-Ject WASAPI driver in Exclusive Mode.

 

And disable 'auto family rate' and give that a crack at 768k and report back.

 

Also set dither = TPDF 

Edited by Music2496
Link to comment
Share on other sites

Use the Pro-Ject WASAPI driver in Exclusive Mode.
 
And disable 'auto family rate' and give that a crack at 768k and report back.
 
Also set dither = TPDF 

Sorry no good, I have a suspicion that the pc just isn’t fast enough to cope with the change.
Link to comment
Share on other sites



1 minute ago, Music2496 said:

No worries. I think I've run out of things to try. Sorry can't be of more help on this one.

 

 

Thanks for your help, I just realised I had a spare Pentium computer doing nothing hence why I tried I will definitely be buying a more powerful computer like @legend did and will try again but as long as I just reset the settings each time I change formats it’s perfect and sounds unreal. Just played alanis Morrisset hand in pocket and you would swear the bass was a double bass played in the lounge room.

  • Like 1
Link to comment
Share on other sites

7 hours ago, rmpfyf said:

Not sure it’ll run to 768kHz in 8 channel mode (or will it?) Audial is about to release something based on it soon at 8 channels.

Yes.

There are 8 inputs and 8 outputs.    You can use them up to 1.536MHz if you want (but I wouldn't).

Link to comment
Share on other sites

8 minutes ago, powerav said:

I just an ab comparison from hq player to Kodi same dac, speakers and amp and just wow,  I now just have to get this hq player working, I can see why all you guys are so nuts about it.?

I use an intel nuc with windows running roon hqp, netflix etc. A capable computer makes it all very stable. Then you can also add room/speaker correction in hqp.

  • Like 1
Link to comment
Share on other sites



  • Recently Browsing   0 members

    • No registered users viewing this page.
×
×
  • Create New...
To Top