• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required. There are daily reviews of audio hardware and expert members to help answer your questions. Click here to have your audio equipment measured for free!

PGGB upsampler: DeltaWave null analysis

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,520
Likes
10,058
Location
North-East
PGGB is an expensive off-line upsampler that promises an extremely high resolution reconstruction of a lower-res recordings (like Redbook), using up to 256-bit floating point computation and avoiding the need to use sinc interpolation (according to the author).

If you read the often glowing reports from customers, it makes an amazing audible difference, despite taking some very serious PC hardware and significant time to upsample existing files, often taking hours or even days to complete and requiring 4x or more times the storage of the original files. Maybe it's really worth it?

Archimago did an evaluation of PGGB previously, and used my DeltaWave software to do the null analysis among other tests. I decided to see if I can reproduce his results and provide a little more details on what PGGB can or can't do. To do this, I picked an 44.1k album to upsample and left PGGB running overnight on an 8 core, 3.6GHz I9 64GB RAM computer. In about 7 hours, I only got 3 of the 10 tracks converted, but that was enough for some initial tests. These were upsampled to 705.6k with all the default/automatic settings.

My first test was to compare the results produced using PGGB and that produced by DeltaWave built-in upsampler. DW uses standard upsampling algorithms and 64-bit floating point compared to the 256-bit of PGGB. Both upsampled the same track to 705.6k. PGGB file was used as a reference, meaning that it wasn't modified during DW processing, only the comparison file is adjusted to match gain and phase of the reference file in DW. Assuming PGGB does something amazing, we should see clear differences between the DeltaWave upsampler result and that done by PGGB.

Spectrum plot... shows no obvious difference. The RMS level of the error after subtracting the two files (in time domain) is -187dB! A-weighted, it's down to -204dB:

1677159972173.png


Any differences when we subtract the two spectra? Nope. Looks like they are the same:
1677160115567.png


Phase differences? Doesn't get more a straight-line than this:
1677160157497.png


Let's see what the actual difference waveform looks like (again, this is computed in the time domain). Notice that even the highest peak value is below the level of thermal noise and overall, seems to be related to dither or quantization error at an extremely low level:

1677160237285.png


Here's the spectrum of the actual difference (null) file:
1677160621913.png


PK Metric audibility plot is just as you'd expect from all of the above:

1677161132789.png


Also notice the numbers across the bottom:
  • 85% of the samples in the two upsampled files match perfectly when expressed as 24 bit samples. Note that even simple random TPDF dither at 24 bits would result in only around a 50% match

  • PGGB reduced gain by 0.1565dB, supposedly to eliminate intersample overs. DeltaWave adjusted the comparison track to match

  • Again, the RMS of the difference is -187dB, A-weighted to -204dB...
I'll listen to these tracks a bit later, but I'm a bit skeptical... ;)
 

tuga

Major Contributor
Joined
Feb 5, 2020
Messages
3,984
Likes
4,277
Location
Oxford, England
PGGB is an expensive off-line upsampler that promises an extremely high resolution reconstruction of a lower-res recordings (like Redbook), using up to 256-bit floating point computation and avoiding the need to use sinc interpolation (according to the author).

If you read the often glowing reports from customers, it makes an amazing audible difference, despite taking some very serious PC hardware and significant time to upsample existing files, often taking hours or even days to complete and requiring 4x or more times the storage of the original files. Maybe it's really worth it?

Archimago did an evaluation of PGGB previously, and used my DeltaWave software to do the null analysis among other tests. I decided to see if I can reproduce his results and provide a little more details on what PGGB can or can't do. To do this, I picked an 44.1k album to upsample and left PGGB running overnight on an 8 core, 3.6GHz I9 64GB RAM computer. In about 7 hours, I only got 3 of the 10 tracks converted, but that was enough for some initial tests. These were upsampled to 705.6k with all the default/automatic settings.

My first test was to compare the results produced using PGGB and that produced by DeltaWave built-in upsampler. DW uses standard upsampling algorithms and 64-bit floating point compared to the 256-bit of PGGB. Both upsampled the same track to 705.6k. PGGB file was used as a reference, meaning that it wasn't modified during DW processing, only the comparison file is adjusted to match gain and phase of the reference file in DW. Assuming PGGB does something amazing, we should see clear differences between the DeltaWave upsampler result and that done by PGGB.

Spectrum plot... shows no obvious difference. The RMS level of the error after subtracting the two files (in time domain) is -187dB! A-weighted, it's down to -204dB:

View attachment 266863

Any differences when we subtract the two spectra? Nope. Looks like they are the same:
View attachment 266864

Phase differences? Doesn't get more a straight-line than this:
View attachment 266865

Let's see what the actual difference waveform looks like (again, this is computed in the time domain). Notice that even the highest peak value is below the level of thermal noise and overall, seems to be related to dither or quantization error at an extremely low level:

View attachment 266866

Here's the spectrum of the actual difference (null) file:
View attachment 266868

PK Metric audibility plot is just as you'd expect from all of the above:

View attachment 266870

Also notice the numbers across the bottom:
  • 85% of the samples in the two upsampled files match perfectly when expressed as 24 bit samples. Note that even simple random TPDF dither at 24 bits would result in only around a 50% match

  • PGGB reduced gain by 0.1565dB, supposedly to eliminate intersample overs. DeltaWave adjusted the comparison track to match

  • Again, the RMS of the difference is -187dB, A-weighted to -204dB...
I'll listen to these tracks a bit later, but I'm a bit skeptical... ;)

Are you skeptikal because you have an adamant belief in your testing script's ability to express audible differences?
 

HarmonicTHD

Major Contributor
Forum Donor
Joined
Mar 18, 2022
Messages
3,326
Likes
4,815
PGGB is an expensive off-line upsampler that promises an extremely high resolution reconstruction of a lower-res recordings (like Redbook), using up to 256-bit floating point computation and avoiding the need to use sinc interpolation (according to the author).

If you read the often glowing reports from customers, it makes an amazing audible difference, despite taking some very serious PC hardware and significant time to upsample existing files, often taking hours or even days to complete and requiring 4x or more times the storage of the original files. Maybe it's really worth it?

Archimago did an evaluation of PGGB previously, and used my DeltaWave software to do the null analysis among other tests. I decided to see if I can reproduce his results and provide a little more details on what PGGB can or can't do. To do this, I picked an 44.1k album to upsample and left PGGB running overnight on an 8 core, 3.6GHz I9 64GB RAM computer. In about 7 hours, I only got 3 of the 10 tracks converted, but that was enough for some initial tests. These were upsampled to 705.6k with all the default/automatic settings.

My first test was to compare the results produced using PGGB and that produced by DeltaWave built-in upsampler. DW uses standard upsampling algorithms and 64-bit floating point compared to the 256-bit of PGGB. Both upsampled the same track to 705.6k. PGGB file was used as a reference, meaning that it wasn't modified during DW processing, only the comparison file is adjusted to match gain and phase of the reference file in DW. Assuming PGGB does something amazing, we should see clear differences between the DeltaWave upsampler result and that done by PGGB.

Spectrum plot... shows no obvious difference. The RMS level of the error after subtracting the two files (in time domain) is -187dB! A-weighted, it's down to -204dB:

View attachment 266863

Any differences when we subtract the two spectra? Nope. Looks like they are the same:
View attachment 266864

Phase differences? Doesn't get more a straight-line than this:
View attachment 266865

Let's see what the actual difference waveform looks like (again, this is computed in the time domain). Notice that even the highest peak value is below the level of thermal noise and overall, seems to be related to dither or quantization error at an extremely low level:

View attachment 266866

Here's the spectrum of the actual difference (null) file:
View attachment 266868

PK Metric audibility plot is just as you'd expect from all of the above:

View attachment 266870

Also notice the numbers across the bottom:
  • 85% of the samples in the two upsampled files match perfectly when expressed as 24 bit samples. Note that even simple random TPDF dither at 24 bits would result in only around a 50% match

  • PGGB reduced gain by 0.1565dB, supposedly to eliminate intersample overs. DeltaWave adjusted the comparison track to match

  • Again, the RMS of the difference is -187dB, A-weighted to -204dB...
I'll listen to these tracks a bit later, but I'm a bit skeptical... ;)
Nice thanks a lot ... another debunked claim. But I am sure, someone will come along claiming they can hear -180dB and more - as we all know, there is a (in) famous designer, who claims -300db ;)

Edit: Just looked up the licence costs - up to 1030 USD.
 
Last edited:
OP
pkane

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,520
Likes
10,058
Location
North-East
Are you skeptikal because you have an adamant belief in your testing script's ability to express audible differences?
No, I’m skeptical because of the extreme low level of the difference. This is far beyond any audibility claims, and any known audibility thresholds. The RMS of a null difference is not perceptually weighted.
 

tuga

Major Contributor
Joined
Feb 5, 2020
Messages
3,984
Likes
4,277
Location
Oxford, England
No, I’m skeptical because of the extreme low level of the difference. This is far beyond any audibility claims, and any known audibility thresholds. The RMS of a null difference is not perceptually weighted.

I would expect different DACs to 'react' differently to the upconverted file. Your software doesn't (seem to take) that into account.
 
OP
pkane

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,520
Likes
10,058
Location
North-East
I would expect different DACs to 'react' differently to the upconverted file. Your software doesn't (seem to take) that into account.
What? The differences are below 24 bits. How did you expect DACs to react to these?
 

manisandher

Addicted to Fun and Learning
Joined
Nov 6, 2016
Messages
648
Likes
598
Location
Royal Leamington Spa, UK
I believe I can hear differences where 'regular' measurements suggest they shouldn't be audible. BUT...

There's no getting around null differences, as measured by DeltaWave. If the difference between 2 files is -185dB, there's not a cat's chance in hell that any DAC on the planet would 'react' differently to these.

Mani.
 
OP
pkane

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,520
Likes
10,058
Location
North-East
I believe I can hear differences where 'regular' measurements suggest they shouldn't be audible. BUT...

There's no getting around null differences, as measured by DeltaWave. If the difference between 2 files is -185dB, there's not a cat's chance in hell that any DAC on the planet would 'react' differently to these.

Mani.

Exactly. The product seems to be based on the premise that ‘any difference, no matter how small, must be audible’.

My audio library contains about 5TB of mostly redbook content. Imagine how painful it would be to convert this all using PGGB to 705.6k/32 bit!

The tiny difference here seems to cost $1k+ and at a great inconvenience requiring all my files to be slowly and painfully converted into an unwieldy disk-wasting format.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,782
Likes
35,671
Location
The Neitherlands
What? The differences are below 24 bits. How did you expect DACs to react to these?

I would expect different DACs to 'react' differently to the upconverted file. Your software doesn't (seem to take) that into account.

Tuga does have a point here. Not about the Deltawave of course.

play a 44.1 file on a filterless NOS DAC and play the same but upsampled file over the same DAC (assuming it can do 705kHz) and you'll hear and will measure a difference on the analog out.
That said... upsample the same 44.1 to 176kHz using a decent upsampler and the expensive and freeware upsampled ones will sound the same on said crappy DAC.

There are many folks around believing filterless NOS DACs are the best sounding. It would be a bit daft (but highly recommended) to upsample first and then convert with a 'crappy' DAC. IF one thinks those DACs are the bestest/musical/fast/[fill in your audio fantasy] sounding ones.
 
Last edited:
OP
pkane

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,520
Likes
10,058
Location
North-East
Tuga does have a point here. Not about the Deltawave of course.

play a 44.1 file on a filterless NOS DAC and play the same but upsampled file over the same DAC (assuming it can do 705kHz) and you'll hear and will measure a difference on the analog out.
That said... upsample the same 44.1 to 176kHz using a decent upsampler and the expensive and freeware upsampled ones will sound the same on said crappy DAC.
Of course, upsampling is not black magic. You don’t need to go through the pain of PGGB conversion and pay $1k for the privilege to get the same result.
 

manisandher

Addicted to Fun and Learning
Joined
Nov 6, 2016
Messages
648
Likes
598
Location
Royal Leamington Spa, UK
I gave PGGB a go last year and really couldn't understand why anyone would use it. To my ears, 16/44.1 files upsampled in PGGB sounded pretty much identical when upsampled in Sony's Audio Studio. And in any event, 16/44.1 already sounds absolutely fine.

Mani.
 

solderdude

Grand Contributor
Joined
Jul 21, 2018
Messages
15,782
Likes
35,671
Location
The Neitherlands
Of course, upsampling is not black magic. You don’t need to go through the pain of PGGB conversion and pay $1k for the privilege to get the same result.

Indeed, some decent freeware upsampling is faster and could be good enough.
 

tuga

Major Contributor
Joined
Feb 5, 2020
Messages
3,984
Likes
4,277
Location
Oxford, England
I believe I can hear differences where 'regular' measurements suggest they shouldn't be audible. BUT...

There's no getting around null differences, as measured by DeltaWave. If the difference between 2 files is -185dB, there's not a cat's chance in hell that any DAC on the planet would 'react' differently to these.

Mani.

I think you mean an ASR cat. :p
 
OP
pkane

pkane

Master Contributor
Forum Donor
Joined
Aug 18, 2017
Messages
5,520
Likes
10,058
Location
North-East
Indeed, some decent freeware upsampling is faster and could be good enough.
As demonstrated, DeltaWave upsampler is good enough to match PGGB to 24 bits and beyond. And it takes seconds to convert a track instead of hours, no specialized hardware or lots of CPU cores + tons of memory required. Maybe I should turn it into a product and sell it for $900 to compete with PGGB :)
 

GXAlan

Major Contributor
Forum Donor
Joined
Jan 15, 2020
Messages
3,778
Likes
5,838
As demonstrated, DeltaWave upsampler is good enough to match PGGB to 24 bits and beyond. And it takes seconds to convert a track instead of hours, no specialized hardware or lots of CPU cores + tons of memory required. Maybe I should turn it into a product and sell it for $900 to compete with PGGB :)

Might as well. You have not proven that there is no difference to the sound. You have just proven that your upsampler is as good and dramatically faster.

That money can help fund all your free tools…
 

fordiebianco

Senior Member
Forum Donor
Joined
Mar 11, 2020
Messages
348
Likes
738
Location
British Isles
Dear @pkane ,

many thanks for running these tests. They seem self-explanatory. Nevertheless, can someone please explain to me (I am neither an electrical- nor audio engineer) how upsampling a track that is based on 16-bit data is supposed to improve my listening experience? Does the software find the missing bits that got lost during the 16-bit recording/sampling?

Sorry for the stupid question.
 
Top Bottom