Octave Music Don Grusin High Resolution Music Analysis (Video)

[Dogcoop]

Octave Music Don Grusin High Resolution Music Analysis (Video)

But then you are not staying in native 24-bit! :D It is not native 24-bit processing anymore! :facepalm: Windows audio pipeline uses shoggy 32-bit floating point just like macOS CoreAudio too. HQPlayer uses 64-bit, 80-bit, 128-bit or arbitrary precision math for PCM as necessary. If you...

audiosciencereview.com

Oh yeah, and no DSD involved

Miska has posted measurements on a few forums over the past many years.

Like he said, you need to look around.

Even various threads on ASR.

Join those threads and ask questions and others with the time can answer.

Well since you’ve decided to enter the conversation, maybe you would be kind enough to answer the many questions that I’ve asked that have not been answered. I would welcome your input. What exactly is the audible benefit of hqplayer. I know that it is a audio player with a multitude of conversion and upsampling features. Do those provide the audible benefit, or is it something else, or is there no audible benefit?
Many measurements have been posted, I’m sure, but I have not found one objective test result. Point me to those.

 

[Miska]

Oh, I fully understand their architecture. Just because there is noise shaping in majority of ADCs does not mean at all that they equal DSD64 1-bit encoding. Easy to prove. Here is RME ADI-2 Pro FS being fed a 1 kHz tone at 96 kHz sampling:

Oh yes, ADI-2 Pro, with 44.1 kHz PCM input:

You can see some images of the 44.1 kHz source around 44.1 kHz and 88.2 kHz sampling rates.

Same, but with DSD256 input from HQPlayer, DSD Direct modet:

 

[Blumlein 88]

SMSL M400 DAC, 1 kHz 44.1k PCM input:
View attachment 193217

You can nicely see how the noise shaper bump peaks at about 1.4 Mhz. And all the digital images of PCM at much higher level, first image at -65 dB.

Let's then run 0 - 22.05 kHz sweep through it:
View attachment 193218

Oh yeah, nice series of images. Peaking at about -45 dB, giving reconstruction accuracy of about 7-bits.

And if we take into account the loudspeaker response at 400 khz what do we get? Plus what about the filtering of the air at 400 khz not to mention anything higher. There is no way that is altering what we hear in those images you show. To call that 7 bits is ridiculous. No wait it is disingenuous is what it is.

 

[Miska]

And if we take into account the loudspeaker response at 400 khz what do we get? Plus what about the filtering of the air at 400 khz not to mention anything higher. There is no way that is altering what we hear in those images you show. To call that 7 bits is ridiculous. No wait it is disingenuous is what it is.

And then some left over DSD noise at same frequencies, but several tens of dB lower is a problem? That looks roughly same as the left over noise bump from the internal modulator there is a problem? Because you can find it in digital analysis of the file. While you can see this only once you measure the DAC output, because you cannot see the digital domain data from inside of the DAC chip between it's DSP and conversion stages.

I guess it is convenient to ignore what happens inside a modern DAC. HQPlayer is designed to be replacement for the in-chip DSP.

 

[amirm]

Oh yes, ADI-2 Pro, with 44.1 kHz PCM input:

The heck you are talking about? I measured the *ADC* which was topic of the discussion. You are showing DAC output and complaining about imaging spikes that are -105 dB below the signal at 40 to 45 kHz? You either lack this much sensibility or really trying to sell us misinformation.

 

[Blumlein 88]

You need to Google around for that

And here is where it always ends with Miska it seems. It is a Zeno's paradox. If you move half way there, and then half of that so on and so forth somehow you don't quite ever get there. With Miska you don't quite ever get a useful answer. He has blind tests results with panels of listeners showing things at -120 db are audible, but he cannot tell you about it. He is prevented in some way. He can write a DAW to do anything you do with PCM for DSD, but has no interest in doing that. So claiming DSD isn't something you can edit is not a valid complaint well except you can't for the most part edit it. So on and so forth and you can't quite ever get there. He himself said he's been at this for 20 years. Maybe another 20 years will do it. Maybe we are halfway there now.

 

[amirm]

Maybe another 20 years will do it.

He will never do it because science is not on his side. There was a reason the industry abandoned DSD.

 

[Blumlein 88]

And then some left over DSD noise at same frequencies, but several tens of dB lower is a problem? That looks roughly same as the left over noise bump from the internal modulator there is a problem? Because you can find it in digital analysis of the file. While you can see this only once you measure the DAC output, because you cannot see the digital domain data from inside of the DAC chip between it's DSP and conversion stages.

I guess it is convenient to ignore what happens inside a modern DAC. HQPlayer is designed to be replacement for the in-chip DSP.

So granting the DSD noise is not a problem what are the audible advantages of DSD over PCM?

We know the files are large, we know that DSD is unwieldy.

 

[Miska]

The heck you are talking about? I measured the *ADC* which was topic of the discussion. You are showing DAC output and complaining about imaging spikes that are -105 dB below the signal at 40 to 45 kHz? You either lack this much sensibility or really trying to sell us misinformation.

OK, I already have shown how it looks like. ADI-2 Pro ADC looks essentially just like DSD128 converted to PCM. Which it pretty much is.

ADI-2 Pro, silence recorded at 705.6/32 PCM:

ADI-2 Pro, silence recorded at DSD256:

Here's another view from 705.6/32 PCM input from ADI-2 Pro ADC:

 

[Miska]

So granting the DSD noise is not a problem what are the audible advantages of DSD over PCM?

Time domain improvements, because it doesn't have the time domain damage caused by brickwall filtering. And no need to shuffle the sampling rate up and down as much at least. So you get rid of the unnecessary decimation part.

 

[amirm]

ADI-2 Pro, silence recorded at DSD256:

You don't dare posting DSD64 from commercial content, do you? I don't trust your signal generator for one millisecond and certainly not at DSD256.

 

[amirm]

Time domain improvements, because it doesn't have the time domain damage caused by brickwall filtering.

You worry about time domain "damage" with PCM at 96 kHz? Any more folklore you want to share?

 

[Miska]

He will never do it because science is not on his side. There was a reason the industry abandoned DSD.

Hmmh, I don't think so. 20 years ago it was hard to find DACs that support DSD. Today it is hard to find DACs that don't support it. Same goes for ADCs. Really nice to see how well it is supported. Of course the implementations could be improved still.

You don't dare posting DSD64 from commercial content

I somehow feel that you don't dare posting DSD256 or DSD512.

Any more folklore you want to share?

You don't seem to be short of folklore about DSD.

He can write a DAW to do anything you do with PCM for DSD, but has no interest in doing that.

I just don't have time to do more. There are limits how much single man can do, even though I work 16 hours a day 7 days a week.

 

[amirm]

For another data point, here's another PCM to DSD comparison from samples by Sound Liaison. Note that this is raw spectrum, before performing a match operation in DeltaWave.

DSD64(white) vs. PCM192(blue):
View attachment 193221


DSD128(white) vs. PCM192(blue):
View attachment 193219


DSD256(white) vs. PCM192(blue):
View attachment 193220

This really closes the door for good on Miska. Commercial content produced by a top label clearly showing the deficiencies of DSD64 and DSD128. Only when you get to DSD256 that it matches PCM at just 192 kHz sampling. DSD256 files are crazy big. Here is my Blue Coast DSD Sampler in DSD 256:

Nearly 10 Gigabytes for one album!

Here is the same album in 24/96 kHz which is plenty for high-res:

Nearly 5 times smaller!

 

[Miska]

This really closes the door for good on Miska. Commercial content produced by a top label clearly showing the deficiencies of DSD64 and DSD128. Only when you get to DSD256 that it matches PCM at just 192 kHz sampling. DSD256 files are crazy big. Here is my Blue Coast DSD Sampler in DSD 256:

View attachment 193240

Nearly 10 Gigabytes for one album!

Here is the same album in 24/96 kHz which is plenty for high-res:

View attachment 193241

More than 5 times smaller!

And less than half the resolution... OTOH, likely your PCM is FLAC while DSD is uncompressed. How about checking the DSD size after WavPack compression? Not that I would be running out of space on my NAS...

But when I'm upsampling my PCM content to DSD256, DSD512 or DSD1024 for playback through the DAC, it is not being stored anywhere anyway. Just glorious sound coming out to my ears!

 

[kongwee]

He is debunked left and right. But if you can't follow the argument, I can see you not getting that.

Get the exact same hardware and software and debunk him straight in his face. I am pretty sure this kind of diagnose is nothing new to you. This is the industrial way to resolve a conflict that cost in millions. However, you just need a person that you know that can debunk in converter knowledge. I don't wish to see you right, i'm wrong. Just take that way. There are threads that do not need to behave like this, I'm enjoying it.

 

[Blumlein 88]

Time domain improvements, because it doesn't have the time domain damage caused by brickwall filtering. And no need to shuffle the sampling rate up and down as much at least. So you get rid of the unnecessary decimation part.

Can you give me a song or two from some recordings where the difference is rather easy to hear?

 

[Blumlein 88]

This really closes the door for good on Miska. Commercial content produced by a top label clearly showing the deficiencies of DSD64 and DSD128. Only when you get to DSD256 that it matches PCM at just 192 kHz sampling. DSD256 files are crazy big. Here is my Blue Coast DSD Sampler in DSD 256:

View attachment 193240

Nearly 10 Gigabytes for one album!

Here is the same album in 24/96 kHz which is plenty for high-res:

View attachment 193241

Nearly 5 times smaller!

I prefer 600% larger. That is all just a little over 600% file size increase. pfffft!

 

[Dogcoop]

And less than half the resolution... OTOH, likely your PCM is FLAC while DSD is uncompressed. How about checking the DSD size after WavPack compression? Not that I would be running out of space on my NAS...

But when I'm upsampling my PCM content to DSD256, DSD512 or DSD1024 for playback through the DAC, it is not being stored anywhere anyway. Just glorious sound coming out to my ears!

Miska said:
Yes, exactly, not about what I hear. Or what you hear.
Really.

 

[amirm]

Get the exact same hardware and software and debunk him straight in his face.

What software and hardware? The topic of this thread is commercial content from Octave music. I showed its spectrum which is full of ultrasonic noise. This is backed precisely by the concept of DSD 1-bit encoding. I also showed the output from real players and high-end hardware player. There is nothing to debunk here if you understood the topic.

His equipment by the way is horrible. A scope is a poor instrument for audio measurement or everyone would spend $500 and $1000 to do what I am doing.

Do you understand the above? If not, tell us in your words what the argument is.