[Tjj226 Angel]

I also want to point out that there used to be a bunch of dac chips on the market (90s) that would take PCM, upsample it, run it through a delta sigma modulator, and process the outgoing signal as DSD64.

Some chips sounded really good, other chips sounded like crap.

 

[bigshot]

Transcoding is going to depend on the quality of the transcoder. That doesn't mean that the format itself is the problem though.

 

[Tjj226 Angel]

Transcoding is going to depend on the quality of the transcoder. That doesn't mean that the format itself is the problem though.

You do realize we are almost always in agreement but never on the same page right?

If a chip can convert PCM to DSD64 and sound good, that means it's not the format.

 

[audiokangaroo]

Transcoding is going to depend on the quality of the transcoder. That doesn't mean that the format itself is the problem though.

The problem with DSD64 is not really about transcoding. Engineers knew from the start that there was a good amount of ultrasonic noise that could not be properly filtered out. They considered that it was not a problem because it was above 25 kHz, but it prooved to be audible in spite of its frequency. This is why DSD128 and DSD256 were produced later. They push the ultrasonic noise much higher and thus it can be filtered out more efficiently.

 

[bigshot]

You could also transcode to a higher PCM sampling rate and then roll off like 16 bit oversampling DACs do. I'm told DSD 256 is pretty much useless. It doesn't offer any improvement over lower rates. I'm not sure of the reason behind that, but someone else here probably knows. I remember Gregorio explaining it. It was interesting.

 

[audiokangaroo]

DSD256 is very useful, because it has a much lower noise floor above 25 kHz. The noise is above 500 kHz and it can be easily filtered out with an analogue filter with
a corner around 50 kHz. DSD256 solves the opacity problem of DSD64. It works as if the ultrasonic noise could leak into the audible range and make high frequencies less audible. This is the reason why this noise produced by the delta sigma modulation must be filtered out.

 

[bigshot]

It makes inaudible frequencies even less audible! The only thing you'll find at 500kHz are S.O.S. signals in morse code!

Just do a low pass filter eliminating the noise over 20kHz. Even if it rolls off a bit, it's still rolling off frequencies you can't hear. Number chasing like this is a rabbit hole. Too much is never enough.

If a chip can convert PCM to DSD64 and sound good, that means it's not the format.

And that means the transcoding by the chip isn't a problem either.

 

[audiokangaroo]

It makes inaudible frequencies even less audible! The only thing you'll find at 500kHz are S.O.S. signals in morse code!

Those frequencies are expected to be inaudible, but we need to understand that they participate in the waveshape, and thus they can influence the sound and the way we can hear it. Such frequencies are not audible separately, but the fact that DSD64 und DSD256 sound different is a new element of proof that ultrasonic informations are audible as a part of the waveshape.

 

[bigshot]

There is absolutely no evidence that inaudible frequencies affect audible sound. The only way it could do that is if it is raised to a volume level that it becomes a sonic weapon blowing out your ears. And please link the study that DSD64 isn't transparent. I know people claim to hear differences. Where's the proof? The idea that we need to be able to resolve up to 500kHz clean in order to hear music up to 20kHz is patently absurd. 500kHz is a radio frequency. Do radio waves from local broadcasting stations make your stereo sound worse or better?

 

[Tjj226 Angel]

And that means the transcoding by the chip isn't a problem either.

I never said transcoding was the issue.

 

[bigshot]

You said that some 90s chips did a good job of transcoding and some sounded like crap. I said that the quality of the transcoding was the cause of the problem.

 

[audiokangaroo]

There is absolutely no evidence that inaudible frequencies affect audible sound. The only way it could do that is if it is raised to a volume level that it becomes a sonic weapon blowing out your ears. And please link the study that DSD64 isn't transparent. I know people claim to hear differences. Where's the proof? The idea that we need to be able to resolve up to 500kHz clean in order to hear music up to 20kHz is patently absurd. 500kHz is a radio frequency. Do radio waves from local broadcasting stations make your stereo sound worse or better?

Please don't conflate electromagnetic radio frequencies with air vibrations.

 

[bigshot]

OK. I will wait for you to provide information backing up your claim that DSD isn't audibly transparent until it gets up almost five octaves beyond the highest frequencies humans can hear.

 

[Tjj226 Angel]

You said that some 90s chips did a good job of transcoding and some sounded like crap. I said that the quality of the transcoding was the cause of the problem.

That's not what I said, nor was that even close to the intent of my comment.

All I was doing was pointing out that if a chip can take PCM and convert it to DSD64 on the fly all while sounding good, it's probably not the format. Full stop. I was just pointing out an obvious data point.

 

[bigshot]

Well then the part about some sounding like crap was confusing me because it doesn't follow your point, and since it was the last thing you say, I assumed that was your summary of your point. I was agreeing with you then. The fact that it's possible to transcode with perfect transparency shows there is nothing wrong with the format. And the fact that some chips transcoded poorly isn't the fault of the format either. It is the fault of the chip doing the transcoding.