A toast to PGGB, a heady brew of math and magic

*******************************************************

(original post)

With the embargo (finally!) lifted, wanted to start a topic to share impressions and feedback on the new PGGB music remastering/upsampling program.

• Main site: http://www.remastero.com
• PGGB: http://www.remastero.com/pggb.html
• FAQ: http://www.remastero.com/faq.html
• Guide: http://www.remastero.com/guide.html
• Onboarding: http://www.remastero.com/onboarding.html

The project started 6-7 months ago to do some experiments in upsampling music, and has grown into something WAAYY beyond anything any of us would have thought possible.  There are ~25 folks that have joined the journey from dev to alpha to private beta. With the impending public launch, very much looking forward to discussing PGGB as more folks are able to hear what has been hiding in their digital content all along.

Thank you @Zaphod Beeblebrox for pushing the math and sound quality to its ultimate limits!

(Disclaimer: although I'm delighted to have been helping on the project since basically day one, I have no financial interest in the project, I paid full price for my license, etc.)

12 minutes ago, ASRMichael said:

Is this effectively another HQPlayer? What are the benefits? 

I view it as complementary to HQPlayer.  While HQPlayer is fantastic for streaming content, PGGB is preprocessing (effectively remastering) music prior to playback.  That opens up a level of processing and compute that just isn't practical in a streaming context (note that HQPlayer Pro is capable of doing preprocessing of content, my comments are with respect to the more commonly used HQPlayer).  I have been and continue to be a delighted HQPlayer customer, even with PGGB.

The project originally started as some experiments to see how far you could push idealized filters for reconstruction (billions of taps vs millions).  The results were stunning, which led to a very long and VERY rewarding series of optimizations.  When used optimally, PGGB is remastering digital content to be aligned as much as possible to the DAC stage of your DAC (think NOS PCM DAC).  Results will vary from DAC to DAC, but with my Chord DAVE, the results are extraordinary and a complete game changer.

The website linked above has more info.  Hopefully the folks that have been in the private beta can share some first hand impressions

There aren't many products where I can say "I was present at the inception." But with PGGB, that is definitely the case. For several months, @ray-dude, @romaz, @Zaphod Beeblebroxand I had been comparing notes on how to optimize the SQ of our Chord DAVEs using various strategies with HQPlayer and other options. In the course of these, @Zaphod Beeblebrox came up with the seed of PGGB, built a prototype, and the rest, as they say, is history. The last few months, as PGGB went from prototype to alpha to beta, and imminently, release, have been a whirlwind. I am amazed at just how big an impact this process makes on SQ. 

The website describes the principles around which the tool is based, so I'll focus instead on the end result. What do PGGB remastered files sound like on my DAVE?

When I compare a native PCM track to a PGGB remastered track, the areas of improvement are:

1. Density. Instruments and voices have far more realistic body and dimension to them
2. Timbre. Instruments sound like real instruments. When a drum gets thwacked, you sense the volume of the instrument, the texture of the "skin" being struck, whether it's fingers or mallets being used, etc.
3. Soundstage. PGGB portrays a holographic, 3-dimensional sense of the performance space.
4. Coherence and detail. There is just a wealth of detail that rises out of the lowered noise floor, that you just didn't notice before. You hear new things in your recordings, new sounds, new bass lines.
5. Bass articulation. The presentation has a tighter, more articulate bass.

And what's amazing is that the biggest impact on SQ in many cases has been on 1FS (redbook) recordings that had been languishing for years in my library. Many of my 1FS albums are now sounding as good as, or even better than, some more recent hi-res purchases.

Finally, and most important of all: PGGB has led me to rediscover, and fall in love all over again, with my library. That's priceless.

I was fortunate to get an invite to join the PGGB beta from @ray-dude.  To say that I was profoundly blown away by the sound quality improvement would be an understatement.  
 

@austinpophas described above the benefits I hear as well.  Density is what I would have listed first too.  I’d just add that the density increase benefits can be heard in both the frequency and time domain.  There’s an increase in the solidity of percussive sounds but that weight is delivered with more realistic rise and fall times.  Anyone who’s heard the Chord HMS will understand this benefit - though to my ears, PGGB takes it to the next level.

If you have Chord DAC, you gotta hear this.  For other brands of DACs, the max input sample rate will likely have some say over how impressed you’ll be.  16fs sounds a lot better than 8fs though 8fs still sounds a lot better than not processed at all.

Lastly I should point out that the improvement from PGGB is far greater than what we hear from buying high res versions of recording.  This is what high res should have been all along.  

1 hour ago, rando said:

Since you are acting ambassador(s) for this commercial entity, correct?

Echoing @ray-dude's point: I have no commercial affiliation with this product, and paid the same $500 fee as all customers do/will.

I've actually known @Zaphod Beeblebrox  a long time as a fellow audiophile and DAVE owner, and while I knew he did "DSP stuff" in his day job, I did not realize he had such hidden talents! So yes, I do wish him well in this enterprise, but in truth, all I'm suggesting is people try this for themselves. The results will speak for themselves.

1. There is a trial version.  It will do five files at a time.

2. Read Hitchhikers Guide to the Galaxy

3. I paid the full license fee and I am sure others did also.  Ray Noted this in his post above!

4.The software runs a very complex set of operations that to be frank eats RAM while running.

5. All I did was add RAM to my  Ryzen 7300 machine and a 4tb SSD for input and output conversions. 16fs files are huge!

6.  If you read the documentation you should see that the system needs Internet access for time checking to validation license, etc.

7.  I do not see any reason this would not run on the CAPS machine.  You would want more RAM and ensure that you have fast storage for conversions.

8. I have found that the results with the three Chord DACs I have is rather impressive.

I am NOT being compensated for my words here.

@ray-dude has been of great help to me during the testing phase. 

1 hour ago, Exocer said:

Sounds interesting but I wonder if I can utilize this effectively with the Yggdrasil DAC. I have a rather small library of flac files which sound really good in their default redbook format. Would the idea be to upsample these files using PGGB to the highest input sample rate of the DAC? Or, would it not be so worthwhile with a DAC that accepts only up to 24/192?

 

Cheers,

-Rob

 

 

Hey Rob,

I'd strongly recommend folks check out the website at https://www.remastero.com/pggb.html, as there's a fair amount of detail there. Specifically, the DACs section will address your questions: https://www.remastero.com/pggb.html#dacs

This project started with the DAVE as the target, but we (collectively) are learning that this approach works on other DACs too. While this is not a definitive association, the improvement due to PGGB on a DAC seems to be higher:

• as the max sample rate increases. Chord DACs go up to 705.6/768 or 16FS. Users have reported great results with 32FS DACs like the Terminator+ and the May - although the latter is difficult to get functional at 32FS. I have personally only tried PGGB on 3 DACs, and the improvement I observed was:
  • DAVE (16FS): massive
  • iFi Pro iDSD (16FS): large
  • Ayre QX-5 Twenty (8FS): marginal.
• if the DAC does NOS when presented a signal at the max rate. For example, the Chord DACs will skip the internal WTA1 filter stage when presented 16FS.

The Yggy, being only a 4FS DAC, may not benefit as much - I don't know. But download the trial and give it a try to see what you think.

2 hours ago, Exocer said:

Sounds interesting but I wonder if I can utilize this effectively with the Yggdrasil DAC. I have a rather small library of flac files which sound really good in their default redbook format. Would the idea be to upsample these files using PGGB to the highest input sample rate of the DAC? Or, would it not be so worthwhile with a DAC that accepts only up to 24/192?

 

Cheers,

-Rob

 

 

We have not tried with 4FS DACs before, but the best way to think about it as a Hires remaster. From your experience, if Yiggy benefits from Hires files compared to redbook then it is worth giving a go. Another use case is if you had access to DXD or DSD tracks, you can use PGGB to remaster them to 4FS rates in a transparent fashion. 

1 hour ago, ecwl said:

I mean, after people commented that HQPlayer filters don’t sound like Chord’s or particularly Chord M-Scaler, HQPlayer eventually added their Sinc-M and Sinc-L filters to mimick the M-Scaler’s longer tap length filters.

sinc-M was because people were asking for such number. I don't use it myself though.

But M-Scaler does such only up to 16x output rate. While sinc-L does 16M taps to 256x output rate. Whole different thing to process 1 million taps at low 768 kHz than 16 million taps at 12.3 MHz.

And yeah, there's undocumented unofficial way to tell HQPlayer to use more or less taps for sinc-M. So you can do 32M or 64M or whatever silly figure even at 50 MHz (1024x) rates or such, if you have enough CPU/GPU power.

1 hour ago, ecwl said:

And now, we basically have a full-blown offline upsamples that can do many, many more taps, run at 32-bit (instead of 24-bit) with noise shaping.

You don't need offline upsample to do 32-bit PCM. HQPlayer can do 32-bit noise-shaped output on the fly. It could as well do 64-bit noise-shaped PCM if you can just find a DAC to support such. (and USB Audio Class is not fit for it either)

1 hour ago, ecwl said:

And then when people still say M-Scaler sounds different and then Rob Watts mentioned that he doesn’t dither and uses noise shaping for M-Scaler, HQPlayer then added the LNS15 noise shaper.

HQPlayer had NS4, NS5 and NS9 noise for maaany years before M-Scaler existed. LNS15 noise shaper was more recent addition.

1 hour ago, ecwl said:

Call it piracy. Or call it Imitation is the Sincerest Form of Flattery.

Chord digital filters have pretty average about 120 dB stop-band attenuation which equals to about 20-bit reconstruction precision. While I prefer million times higher, 240 dB stop-band attenuation which equals to about 40-bit reconstruction precision. And less than 0.1e-8 dB pass-band ripple.

And in addition, sinc-M is apodizing filter while Chord's WTA is non-apodizing.

So there is not much similarity except number of taps (which is pretty meaningless figure).

Let's see when Chord will get anywhere near the performance...

1 hour ago, ecwl said:

1) HQPlayer did not have Sinc-M or Sinc-L long tap length filters until long after M-Scaler came out

Yes it had, for example closed-form-M and closed-form-16M (16 million taps), and so on. But I considered number of taps pretty pointless figure so was not (and still I'm not) talking much about tap lengths. Especially because HQPlayer is not for any fixed output rate like 16x or similar, unlike M-Scaler. Instead it can do various different conversion ratios, including ones that are not multiples of the source sampling rate. Such things usually affect number of taps, so quoting a single figure is not sensible.

1 hour ago, ecwl said:

2) HQPlayer did not use noise shaping for 16fs upsampling with LNS15 as an option until a couple months after Rob Watts mentioned on Head-Fi forum that he doesn’t dither and uses noise shaping after 16fs upsampling

Oh yes it certainly did, already about 10 years ago. NS4 and NS9 were introduced in January 2011. And those have never been limited to 16x rates.

1 hour ago, ecwl said:

So while the IP is in public domain, Jussi took the easy way out and took Rob Watts tuning and incorporated it into his software without giving any credits to Rob Watts

Which tuning are you talking about? Can you explain what I should give credit for? Some number of taps? While the taps I have are way much better than the ones created by Rob Watts.

Since HQPlayer has been doing upsampling since 1998 and those noise-shapers instead of TPDF and shaped dithers since 2011, I would counter argument that Chord is just doing poor copy of what HQPlayer has been doing doing without giving any credits to HQPlayer.

Jussi has every right to defend his work, his IP and his experience.  I just wish he didn't have to do it, but it seems that every time a new product is released with some fanfare and positive review there are folks who need to bash competitors, or call the efforts "trivial."  WTF!!  C'mon folks, lets talk about PGGB, ask questions of the developer, and share experiences.  If you haven't heard the product please don't bash it.  If you want to, unprovoked, put down the competition do it somewhere else...or expect a war of words (and wars are not productive).  My $.02

First of all, a thank you for the developer @Zaphod Beeblebrox to work out a unique development of computer audio processing. He found a problem with the processing of the Chord DAVE and decided to do something about it, hats off to you. When reading through the FAQ and the guides, let's get to the gist of things: 

i)   How much will it cost 
ii)  What work do I need to do to setup PGGB 
iii) Could there be any benefits to SQ

Cost

a) Purchase another 32GB ECC RAM
b) Purchase another 10 TB HDD to store the processed WAV files
c) purchase another 10 TB backup drive to backup the processed wav files
d) Buy the PGGB license

That's close to USD1500,  one mortgage payment or a month's rental. 

Work involved

Adding another Hard drive, replacing RAM is less than a morning's activities, procuring often takes 3 times as long, since I can't usually buy all items from the one place at an optimised price. This results in more freight charges. Internationally, that's a big deal, but over a certain weight, the additional items are trivial.

The wait to process some 125,000 tracks, Hitch Hiker's Guide to the Galaxy would be the intro to the complete books of Game of Thrones.

SQ Benefits

From reading the responses so far, owners of Chord DACs have the greatest benefit. I don't usually bag other user's DACs since that's their choice to buy, freedom and enjoyment is what its all about, huh. Does this mean then the algorithms in the Chord devices are 'not optimised' to begin with? The Ayre QX-5 20 has marginal effect with PGGB. So far a range of DACs the effect can be marginal or massive. I suppose in time there will be a list of more DACs as users take up PGGB, so the choice of implementation is easier to make.

As noted from a long time ago, some DACs are great at Redbook and if you have one of these, higher up natural rates including DSD also will sound great. Seems that the internal high upsampling rates to PCM DACs and then using their filtering is technique that can be bettered. I don't own an NOS DAC, so cannot comment.

I also have doubt that increased processing will make the PC any quieter as far as conducting noise emissions is concerned. Forget the wasted hours in trimming processes on Windows, PGGB makes up for that processing noise and then some.

PGGB needs to convert to PCM always, I understand the complexity.  It will also mean direct DSD conversion by the DAC is bypassed. Boo, hiss. 

Upsampling

The reason I gave up on HQPlayer long ago now, there's a great difficulty in finding the 'right' filter and upsampling rate to DSD or high rate PCM, when the source material can vary so considerably. From the PGGB Guide notes:

"Choose ‘Moderate’ or ‘Dense’ when you need more body. This is especially useful for genres such as classic rock, big band jazz or CDs that may not have the best downsampled quality."

I listen to a wide variety of music, so would I need to change the filtering to suit, if I could ever find that medium. That means constant fiddling and tweaking, sorry there's a major negative. For at least three years or more, I gave up on upsampling processing within the software, since the DACs I've owned take care of all that elegantly and let the player software play, as is.

Conclusion 

The final analysis is the USD1500 could be better spent on noise reduction from the Music Server, like the XE JCAT NET Femto, fanless designs, mechanical isolation, that kind of optimisation is 'real and tangible'.
The added drive for backups and drive replacements is tedious. 
I don't use direct streaming from Qobuz, Tidal, Spotify, those that do will be disappointed (especially with Chord DACs)

PGGB based on what's written above, is not for me, sorry about that.

PGGB was not initially intended as a commercial venture but rather a passionate pursuit among 4 like-minded audiophiles.  @Zaphod Beeblebrox, @austinpop, and @ray-dude are electrical engineers.  I am the odd man out but as a physician-scientist, I have strong technical leanings and so you can imagine the endless "nerdy" conversations.  One thing we each had in common was our admiration for the Chord DAVE and for Rob Watts' philosophies on PCM upsampling.  This is not to say that the approach of others is less valid but for our group, this is what galvanized us.  

We looked at all the things that we thought mattered.  Initially, it was about "taps" and getting as many "taps" as possible.  With real-time upsampling, because of the limitations of hardware, we could only get to 32 million taps initially and with massive time delays (>1 minute).  With offline upsampling, gone are the restraints of hardware and depending on the file and the length of the track, PGGB has provided me new masters with up to 4 billion taps.  Is this audible?  My suggestion is that you hear it for yourself and decide. 

Then the conversation turned to noise shaping which can serve to lower the noise floor but also shape the signature of the sound.  Without noise shaping, the noise floor or what Rob Watts likes to call the "hiss level" of a 16-bit file is -96dB.  At 24-bits, this drops to -144dB.  Rob showed with the Chord DAVE that the human ear can hear noise to a level of -350dB.  With PGGB's noise shaper, re-mastering at 16-bits, the noise floor drops to -330dB and pushed to 32-bits, it drops to -360dB.  With the Chord DAVE, we each have verified through listening tests that this is audible.  Again, the best thing I can suggest is try it for yourself and see.  To my ears, listening to the original files via streaming from Qobuz or Tidal or from internal storage is now considerably less enjoyable and more fatiguing.

By varying parameters, the signature of the sound can be "shaped" from soft to incisive, dense to airy, or transparent to more forgiving.  Depending on the quality of the recording, the properties of your components or your room, or just personal preference, PGGB allows you to create as many different masters of your files or albums as you wish.  With time, I have settled on 2 settings, a very transparent setting for my very best recordings and a more forgiving setting.  In time, PGGB may evolve to include EQ (for headphones) and room correction.  Given @Zaphod Beeblebrox's background in DSP, this is not only possible but something that he has a personal interest in.

As for the ideal platform for PGGB, you will want a DAC that excels in PCM.  PGGB can handle all of your files including any DSD512 files you might have but ultimately, they will be converted to PCM.  The higher the sampling rate of PCM your DAC can handle, the better and there are some DACs today that can read 1,536kHz PCM files but it's important that your DAC adds no further processing to a PGGB file and hence, you will want a PCM DAC that has a NOS (non-oversampling) mode or a digital filter bypass mode.

As mentioned above, all of this effort was initially intended "for our ears only."  While Ray, Rajiv, and I have been involved from the beginning, our involvement has purely been to listen and to provide feedback.  In a way, we helped voice PGGB but to be perfectly clear, it is @Zaphod Beeblebrox who poured hundreds of hours of his own time coding and refining PGGB to what it is now.  PGGB is @Zaphod Beeblebrox's masterpiece and I feel comfortable calling it that.  As purely a passionate pursuit among friends,@Zaphod Beeblebrox never asked for compensation but I suggested $500 to compensate him for his time and so this is what Rajiv, Ray, and I each voluntarily paid.   

As we shared a few PGGB re-mastered files with a few audiophile friends purely for feedback, the feedback was overwhelmingly positive and it became cruel to give some of these friends a taste but not the opportunity to re-master their own library and so @Zaphod Beeblebrox poured more of his time refining PGGB further with a user-friendly interface and a broader range of settings to accommodate for different preferences.  These individuals were offered their own copy of PGGB for the same price that Rajiv, Ray, and I paid and this beta group has been instrumental in further refining PGGB.  This group is comprised of individuals ranging from computer novices to IT experts with DACs from companies like Chord Electronics, Denafrips, Holo Audio, dCS, Gustard, and others.  

As for the word "piracy," yeah, that's a bit harsh.  None of Rob Watts' code is used in PGGB but we did feel his WTA algorithm sounded most natural to our ears and so through listening tests, we attempted to model what we heard.  I consider Rob a friend and I sent him an early version of a PGGB file to get his opinion.  To be upfront, he didn't like what he heard and so I left it at that and so people are free to make up their own minds.  No claims are being made by PGGB that it sounds just like or is better than M-Scaler, HQP, etc.  Ultimately, PGGB is not intended to compete with Rob Watts or with Chord Electronics but hopefully serves as a complement to Chord's offerings.  In my opinion, Chord DACs are the ultimate vehicle for PGGB-remastered files.  

Lastly, just like Ray and Rajiv, I have zero financial affiliation with PGGB or with @Zaphod Beeblebrox.  I would have been happy to keep PGGB to ourselves but imho, PGGB is too good not to share.  

Lot to unpack here.  A couple quick comments

* There is literally no engineer working in and around audio that I respect more than Rob Watts. I have owned 4 of his DACs, and will order his next DAC one day 1 preorders are allowed.

* The Whittaker-Shannon theory is as foundational as it gets for modern information theory.  It continues to be curious why more folks aren't leveraging it.  There is no IP/secret sauce here, just solid engineering and math fundamentals.

* Jussi clearly focuses on DSD more than PCM.  He has been very supportive of his PCM upsampling customers asking for various enhancements, and I'm very appreciative that he added the various sinc filters to HQP, and the LNS15 noise shaper (I happily use these for streaming).  He does tremendous work, and being a good engineer with a rrobust understanding of fundamentals and supporting your customers most definitely does not make you a copy cat.

* Having carefully compared Hugo mScaler (Rob's 1M tap upsampler, which I adore...see my review of mScaler here) with HQPlayer running sinc-M and LNS15, to my ear HMS is clearly better with Chord DAVE (I can not speak to other DACs).  There is more than applying 1M taps and a 15th order noise shaper. The details matter, and tuning to your target DAC matters.

* Having heard upwards of 1B tap processed files 6+ months ago, it was a huge step forward from 1M taps, but it can't even hold a candle to where PGGB is today. Rob is very open about using solid engineering fundamentals (1M taps based on Whittaker-Shannon theory), but I can say definitively with no equivocation that the secret sauce is NOT in the taps.  The taps are permission to play (and an easy thing to hang a marketing position on), but Rob's genius is his relentless tuning of the pipelines and filters so they sound the best they can possibly sound.  He does not publish that.  Having gone through this process for 6 months now with PGGB, it is grossly unfair to call what ZB has developed as copy cat.

With all that being said, I would ask that we keep things informational and share impressions.  Jussi/Rob/ZB have no need to defend themselves for solid engineering work, and if HQP or mScaler or PGGB or a stiff shot whiskey give you more enjoyment from music, that is worth sharing and celebrating.  Please, there is no need to impugn others, or ascribe intent or bad faith to people.  

Back to PGGB’s benefits....

I think what has impressed me most is how much better DSD sounds once processed to 16fs.  Processing these files takes a lot longer but I’ve consistently found it worth the wait.  I’m almost inclined to say that it breathes new life into these as transients are much improved.

I was a late stage participant in the closed beta of PGGB, so I've only recently made the plunge to change up my audio chain, which included my usb source, through my HMS and until recently my Hugo TT2. I've spent some time over the years pursuing optimal output from my digital library to my ears.  Learning from folks here on AS, I focused on practicing good digital hygiene to great effect.   I've recently upgraded my TT2 to a DAVE. and after studying the basis of PGGB, and reading beta participant subjective results, I installed PGGB for a test drive.  I bypassed everything between my server and DAVE; just output from my JCAT USB XE, LUSH^3 and DAVE.  Playing critical listening test tracks that I'm keenly aware of the nuance and intricacies within I was able to achieve a similar (if not improved) SQ without multiple intevening cables, devices and power supplies between my HMS and DAVE.  I like the elegance of simplicity and I like the reclaimed space. But it comes at a price, the PGGB license, upgrades to my server's memory, CPU and power supply.  I'm also adding more storage to my NAS.  But I believe PGGB delivers results that justify the cost and effort, for me.  Download PGGB and see if you can find a machine to process some files using the Onboarding workflow to evaluate the various configurations and the resultant outputs.  My server had a i3 processor and 16gb of memory and I was unable to run the program.  Once I decided to take the leap of faith, my 128gb of memory arrived before anything else. After installing the memory I ran the program on the i3 processor and PGGB plowed through the files at a pretty remarkable rate.  

Audio is like a braided stream, there are multiple ways to get downstream.