Review and Measurements of PS Audio PerfectWave DirectStream DAC

[amirm]

My question is more like for what purpose does the DAC exist? For measurements or for listening enjoyment?

Whether it is measurements or subjective listening of DACs, neither has anything to do with enjoying music. It is a comparative task that we don't perform when just enjoying music.

Indeed that is the main problem: people try to compare, focus more on one or the other device and discover new detail, air, blacker backgrounds, etc. that was always there. What brought them out was the focus, not the sound changing.

 

[BDWoody]

My question is more like for what purpose does the DAC exist? For measurements or for listening enjoyment? IMO, neither is excuse to do badly on the other.

Music exists for listening enjoyment. You talk about it like it's a musical instrument.
A DAC exists to be functionally invisible. If a DAC is not audibly transparent (as can be reliably measured), it isn't doing its job properly. If you want to design synthesizers, I'm sure that has its own separate challenges.

Why would I want your idea of what sounds good, when I can get a transparent DAC and hear what George Martin's idea of what sounds good when he signed off on the final product?

I can do any final tweaking to MY taste with DSP and whatever else, but I'll start with what's closest to the original.

 

[Miska]

Miska, which measurements do correlate with the sound quality differenciest that you hear? Obviously TIM is one of them, what others? Do you use AP analyzer or do you know how to perform them with the AP that Amir has? It would be nice if you could help him to develop/find the most important tests!

I use three different analyzers. But for example shape of the filter roll-off correlates, and amount of overshoot analog stages have, and reconstruction accuracy (image levels). And complex behavior of the the SD-modulator. Just to name a few.

I've tried to discuss some of these aspect here, but response has been that it doesn't matter. Only SINAD matters, although you cannot hear it.

I just do a lot of measurements and testing as part of R&D.

Personally, I am mostly interested in performance of redbook and 24/96. Does MP3 or AAC set specific demands/problems?

My focus is also on lossless formats.

MP3 and AAC certainly have their can of worms, AAC improving over many of the defects MP3 had, such as better transient response. In the past I've spent quite a bit of time comparing different formats and encoders. There are also many listening test results between different MP3 and AAC encoders (different encoders use somewhat different approaches regarding psycho-acoustic models etc).

 

[Miska]

Indeed that is the main problem: people try to compare, focus more on one or the other device and discover new detail, air, blacker backgrounds, etc. that was always there. What brought them out was the focus, not the sound changing.

Part of my job is to listen for such, but there are things I specifically know to listen for. Just like when listening for submarines and ships through a passive sonar system.

But there are three levels, one is the non-focus listening where you listen for the big picture. And then the focus listening where you use hearing's capability to block out the surrounding noise - just like when you do when you try to listen what someone is saying in a noisy night club. Then you combine impressions from these two. Third one is long term listening if you notice that something begins to annoy you, or the listening becomes tiresome.

 

[Miska]

Music exists for listening enjoyment. You talk about it like it's a musical instrument.
A DAC exists to be functionally invisible. If a DAC is not audibly transparent (as can be reliably measured), it isn't doing its job properly. If you want to design synthesizers, I'm sure that has its own separate challenges.

Why would I want your idea of what sounds good, when I can get a transparent DAC and hear what George Martin's idea of what sounds good when he signed off on the final product?

I can do any final tweaking to MY taste with DSP and whatever else, but I'll start with what's closest to the original.

Point is that SINAD doesn't tell if it's audibly transparent.

Some DACs are more transparent than others, even if they have exactly same SINAD figure. Or same DAC driven different way, still having same analog stage performance (SINAD, IMD, etc).

 

[digititus]

Some DACs are more transparent than others, even if they have exactly same SINAD figure

Define "transparent"?

 

[Bliman]

Point is that SINAD doesn't tell if it's audibly transparent.

Some DACs are more transparent than others, even if they have exactly same SINAD figure. Or same DAC driven different way, still having same analog stage performance (SINAD, IMD, etc).

What measurement then show if it is transparent?

 

[Miska]

Define "transparent"?

Accurate reconstruction of the original analog waveform, also taking into account deficiencies of the preceding production chain...

 

[jasonq997]

And I'm sorry, but I personally don't really like how the ESS Sabre sounds. No matter what kind of THD figure some Sabre DAC gives.

ESS/CirrusLogic - on your face
AKM/Wolfson - neutral
TI/BB - laid back

I wonder what more needs to be done to finally debunk this nonsense...

 

[Miska]

What measurement then show if it is transparent?

Many, and some very likely still unknown. I've tried to discuss this many times before, but the response has been negative throughout.

In the end you also need to accept that reaching something totally transparent throughout the (re)production chain is going to be impossible. You can only try to get as close as possible.

 

[Bliman]

Many, and some very likely still unknown. I've tried to discuss this many times before, but the response has been negative throughout.

In the end you also need to accept that reaching something totally transparent throughout the (re)production chain is going to be impossible. You can only try to get as close as possible.

Define many. And how do you know that you are missing things?
Are you saying you hear differences between the top tier dacs in blind testing?
Because how do you know if you are missing something in measurements if you can't hear a difference.
Isn't that like saying that elves and trolls exist? We haven't seen them yet but they are out there.

 

[digititus]

Many, and some very likely still unknown. I've tried to discuss this many times before, but the response has been negative throughout.

Reminds me of the saying "all that counts can not be counted and all that can be counted does not count."

 

[Miska]

Define many. And how do you know that you are missing things?

I suspect... Since there is no easy straightforward correlation, there are likely missing aspects still. I'm totally fine accepting that world is not ready and perfect yet. Maybe in 500 years we know a bit better.

Are you saying you hear differences between the top tier dacs in blind testing?

Define top tier DACs? OK, let's say we take RME ADI-2 as an example. You can switch between different digital filters while still having same SINAD figure. Sounds different. I can switch to DSD256 still having same SINAD figure and switch between different filters and modulators. Same DAC, same SINAD performance all the time, but different sound.

Yes, I know how to measure the difference between these cases. Response here has been that it doesn't matter and it is not audible. OK, fine. Here it only matters if SINAD is 121 dB or 123 dB, except that response here is that it cannot be heard either, but still it matters. OK, fine.

 

[GrimSurfer]

Isn't that like saying that elves and trolls exist? We haven't seen them yet but they are out there.

Sort of.

I am not aware of any evidence to support the thesis that trolls and elves exist. Therefore, I have high confidence in the proposition that they do not exist.

Conversely, an absence of evidence on the existence of trolls and elves technically does not prove that they do not exist. It simply means that there is no evidence to support their existence. Therefore, I will continue to have high confidence that trolls and elves do not exist until evidence to the contrary is presented.

It is certainly possible that we lack the knowledge or means to measure all audio characteristics. Is it likely, given the many advances we have made in science? It appears unlikely but I cannot discount the possibility of that which I cannot disprove.

It is this kind of logic that ensures science is questioning. Accepting that an unknown measure might exist is good science. Believing that an unknown measure does exist, in the absence of any evidence, is illogical.

It's a fine line.

 

[Bliman]

You have a good argument (with the knowledge that I have) that different filters sound different (I have not a dac with different filters, but I believe you).
Grimsurfer: Yes you never can discount anything. But I thought that we know much about hearing. So do you say that in the future we can have dacs that maybe are now tested as transparent are not to be the case in the future? I mean in the way that it makes a difference to what we hear from the dacs.
I thought that other people say that Dacs are already transparent and it doesn't make a difference anymore if we make them measure better.

 

[Hugo9000]

The only people I see going on about SINAD and nothing but SINAD are those setting up a strawman and mischaracterizing this forum and the members here. And Amir does lots of measurements beyond SINAD, and does extra ones when people ask. I've never seen anyone claim that two devices with the same SINAD sound the same. If I've missed those threads, I hope someone will link them. It's a boring Sunday evening, and I could use some entertainment. Giving my hearing a break from music right now as I just finished listening to three complete operas.

 

[MZKM]

Why would I want your idea of what sounds good, when I can get a transparent DAC and hear what George Martin's idea of what sounds good when he signed off on the final product?

Don’t forget the Circle of Confusion, what if the system Martin heard the on was overly warm or bright?

 

[BDWoody]

Many, and some very likely still unknown. I've tried to discuss this many times before, but the response has been negative throughout.

Well, I would say it hasn't been very receptive...but that happens when you make claims that don't always make sense. Your focus on what is outside the audible range is an interesting hobby I suppose, but until you can correlate your claims of actual audibility to actual evidence, rather than the old because I've been doing it so long I don't have to...and I don't care what anybody thinks... I'm not sure what you expect?

 

[GrimSurfer]

So do you say that in the future we can have dacs that maybe are now tested as transparent are not to be the case in the future? I mean in the way that it makes a difference to what we hear from the dacs.

It depends.

It could be an issue of resolution. We could be measuring the right thing, but unable to identify the distinguishing characteristics from the noise floor etc.

It could be an issue of human hearing. We could be measuring the right thing to the right resolution... but people could perceive the sound differently. That perception could be influenced by many factors which we do not yet understand.

It could be an issue of conditionality. The measurement of characteristics could be accurate and reliable in that test environment... but if you change the environment then the perceived sound changes.

These are just three propositions one could use to frame the issue. But, to be clear, I don't think anyone can say that some kind of missing characteristic of sound exists... one that we've yet to discover/measure. It might exist. But it cannot be proven, so it cannot be said to exist.

 

[GrimSurfer]

The only people I see going on about SINAD and nothing but SINAD are those setting up a strawman and mischaracterizing this forum and the members here. And Amir does lots of measurements beyond SINAD, and does extra ones when people ask. I've never seen anyone claim that two devices with the same SINAD sound the same. If I've missed those threads, I hope someone will link them. It's a boring Sunday evening, and I could use some entertainment. Giving my hearing a break from music right now as I just finished listening to three complete operas.

Very good point.

It's plausible that the unique interaction of noise, distortion, and frequency/linearity is what gives different DACs of comparable measurement slightly different sound characteristics.

I raise this because, while we accurately measure all these things, I'm not aware of any exhaustive study into how the correlation of these 3 or 4 factors change anything.