• 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 many 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

-b <74 .. 99.7> = -3 dB percentage
SoX can go to 99.9999 and many billions of taps with a patch:
 
Is it clear that PGGB and SoX are more accurate than DeltaWave?

(Not that it matters in the least bit for all practical purposes. Purely out of interest...)

Mani.

Ha! Not so fast! :)

First, it's not DeltaWave, but DeltaWave resampler that was tested here. The resampler is normally not part of DeltaWave analysis and was used here just because it was convenient to do everything in one step, without leaving DW. The result deviates below about -180dBFS from the PGGB or SoX -u results, although which one is more accurate remains a question, since we don't have a 705.6k original digital recording to compare to :)

Just like SoX, DeltaWave has settings to increase the tap size for low pass filter up to 16M. What I measured in the OP was using the default 256k tap filter. If you increase filter size, I suspect the differences will become even smaller.
 
Ha! Not so fast! :)

First, it's not DeltaWave, but DeltaWave resampler that was tested here. The resampler is normally not part of DeltaWave analysis and was used here just because it was convenient to do everything in one step, without leaving DW. The result deviates below about -180dBFS from the PGGB or SoX -u results, although which one is more accurate remains a question, since we don't have a 705.6k original digital recording to compare to :)

Just like SoX, DeltaWave has settings to increase the tap size for low pass filter up to 16M. What I measured in the OP was using the default 256k tap filter. If you increase filter size, I suspect the differences will become even smaller.
I think it shows that all these options do the job accurately.
 
PGGB now uses 256 bit precision:

"We managed to reduce the noise floor in the audible range from an already low -540dB to -740dB."

Before dismissing this outright as being totally crazy, can anyone here steelman a reason for doing this?

Mani.
 
PGGB now uses 256 bit precision:

"We managed to reduce the noise floor in the audible range from an already low -540dB to -740dB."

Before dismissing this outright as being totally crazy, can anyone here steelman a reason for doing this?

Mani.

The same reason one might want to do DSD65536 instead of DSD256 -- bigger is better!
 
"We managed to reduce the noise floor in the audible range from an already low -540dB to -740dB."

Sounds like time well spent. Who knows, someone might do a trillion sample rate conversions in a row, then this sort of reserve comes in handy…
 
"We managed to reduce the noise floor in the audible range from an already low -540dB to -740dB."

Before dismissing this outright as being totally crazy, can anyone here steelman a reason for doing this?
I think there are 2 main reasons someone might want to do this:

  • Resampling data from extremely sensitive scientific equipment like LIGO or something
  • Because you know dB are important to sound quality, but have no idea how they work or what they are, so more is more regardless of how many more.
This is about as practical as using special car wax to reduce your car's drag coefficient through the interstellar medium so that it can go 0.99999 c instead of just 0.99998 c, you know, while you're driving to get groceries outside of the solar system.
 
Before dismissing this outright as being totally crazy, can anyone here steelman a reason for doing this?
They want Rob Watts to buy it, the previous accuracy of billions tap filter was unlistenable to him.

Besides I bet that the civilization that runs this simulation has much better filters than us humans.
 
  • of how many more.
This is about as practical as using special car wax to reduce your car's drag coefficient through the interstellar medium so that it can go 0.99999 c instead of just 0.99998 c
I'm not an expert on special relativity but considering time dilation and energy needed for achieving this speed might make it actually a pretty decent jump
 
This is about as practical as using special car wax to reduce your car's drag coefficient through the interstellar medium so that it can go 0.99999 c instead of just 0.99998 c, you know, while you're driving to get groceries outside of the solar system
Yeah that's a really really bad example, it would be worth it as the difference is quite big
Screenshot_20230408_025412_Chrome.jpg


Screenshot_20230408_025441_Chrome.jpg


It's kinda funny since your example works in totally opposite way and the closer you get to c at those fractions the effects become more extreme
 
Where is the Deltawave Resampler? My friend just got this software and here I stumbled on the thread. There is a 256b mode as someone mentioned here now.

What about the use of PDDB to apply EQ's and other such things to the library of music? Has that been tested at all? I imagine it could be useful for portable listening or so.
 
Where is the Deltawave Resampler? My friend just got this software and here I stumbled on the thread. There is a 256b mode as someone mentioned here now.

What about the use of PDDB to apply EQ's and other such things to the library of music? Has that been tested at all? I imagine it could be useful for portable listening or so.

DeltaWave is not a resampler, but has a resampler as part of the toolkit (also has filters, deconvolver, variable size FIR and IIR filters, and a ton of other tools and can work with DSD files) . It's designed to perform null comparison of two waveforms. While this is mostly used by comparing two recordings of music at the same sample rate produced with two different devices, you can also tell DeltaWave to resample one of the files to match the other, or resample both files to a desired sample rate.

Here, I used PGGB resampler to create the reference wave file, and used DeltaWave to resample to the same rate and then compare the two for differences. The result wound being so tiny as to be well below any audibility and, in fact, well below the ability of any existing DAC to reproduce. DeltaWave uses 64-bit floating point computation (fairly standard for PC-based software) while PGGB claim to fame is the huge number of bits used for resampling "precision".
 
DeltaWave is not a resampler, but has a resampler as part of the toolkit (also has filters, deconvolver, variable size FIR and IIR filters, and a ton of other tools and can work with DSD files) . It's designed to perform null comparison of two waveforms. While this is mostly used by comparing two recordings of music at the same sample rate produced with two different devices, you can also tell DeltaWave to resample one of the files to match the other, or resample both files to a desired sample rate.

Here, I used PGGB resampler to create the reference wave file, and used DeltaWave to resample to the same rate and then compare the two for differences. The result wound being so tiny as to be well below any audibility and, in fact, well below the ability of any existing DAC to reproduce. DeltaWave uses 64-bit floating point computation (fairly standard for PC-based software) while PGGB claim to fame is the huge number of bits used for resampling "precision".
Oh I see, but you cannot output the file directly and upscale it in the same way as PGGB using DeltaWave, correct?

I was just wondering about the EQ features of PGGB and other remastering features which may be useful to musicians and such? I'm just thinking about other uses not for audiophiles specifically since just upscaling music you already have isn't going to do anything unless maybe you have a NOS DAC, but that would be comical to upscale music instead of use a normal DAC.
 
Oh I see, but you cannot output the file directly and upscale it in the same way as PGGB using DeltaWave, correct?
Sure, you can take a file, upscale it, and save it in DeltaWave. Again, it's not what it's designed to do, but it'll do it.

I was just wondering about the EQ features of PGGB and other remastering features which may be useful to musicians and such? I'm just thinking about other uses not for audiophiles specifically since just upscaling music you already have isn't going to do anything unless maybe you have a NOS DAC, but that would be comical to upscale music instead of use a normal DAC.

Don't see a good reason to EQ your music collection at a great cost in time and storage space, ahead of time. I apply all my EQ in real-time, and it costs me next to nothing compared to PGGB. What's more, I can adjust EQ based on device, location, or even music that I'm playing. So, not sure why you'd need PGGB to do any of this.
 
Sure, you can take a file, upscale it, and save it in DeltaWave. Again, it's not what it's designed to do, but it'll do it.



Don't see a good reason to EQ your music collection at a great cost in time and storage space, ahead of time. I apply all my EQ in real-time, and it costs me next to nothing compared to PGGB. What's more, I can adjust EQ based on device, location, or even music that I'm playing. So, not sure why you'd need PGGB to do any of this.
I'm not saying for an audiophile but maybe as an artist to tweak the music to how you want it to sound maybe before full release? Or to adjust it per platform you release it on since some platforms are different than others?
I mean the cost of the software doesn't seem like something that would be for a home user to listen to music.
 
I'm not saying for an audiophile but maybe as an artist to tweak the music to how you want it to sound maybe before full release? Or to adjust it per platform you release it on since some platforms are different than others?
PGGB takes a very long time to compute anything with the 256 bit precision. There is no way to use it for music production without waiting a long while to hear any tiny change after you make it. The author appears to have released a resampling plugin for Foobar, which doesn't use high precision and so works in real time... like nearly every other resampler on the planet. Since it uses standard precision, it doesn't provide any of the originally claimed benefits of PGGB -- it's just another resampler.

I mean the cost of the software doesn't seem like something that would be for a home user to listen to music.
And yet, this is precisely who buys PGGB. People pay a lot for the resampler, and then have to upgrade their computer equipment to run it, and then have to spend days converting their audio libraries to higher resolution... for no real benefit.
 
PGGB takes a very long time to compute anything with the 256 bit precision. There is no way to use it for music production without waiting a long while to hear any tiny change after you make it. The author appears to have released a resampling plugin for Foobar, which doesn't use high precision and so works in real time... like nearly every other resampler on the planet. Since it uses standard precision, it doesn't provide any of the originally claimed benefits of PGGB -- it's just another resampler.


And yet, this is precisely who buys PGGB. People pay a lot for the resampler, and then have to upgrade their computer equipment to run it, and then have to spend days converting their audio libraries to higher resolution... for no real benefit.
Seems kind of pointless to me tbh.
I could see it used as a music production tool especially to record in a high resolution and then remux for things like Tidal master or Qobuz.
 
Back
Top Bottom