MQA bails on Rocky Mountain Audio Fest*

So the Schiit colours the output from the DSD layer? How does the same layer sound via a different DAC?

 

Every PCM only DAC I've tried "colors" the sound when using software (in my case JRiver or Foobar) to convert DSD to PCM. It's just the nature of the beast. The sound gets softened and soft focused like that when converting DSD to PCM. DACs that accept native DSD can maybe do a little bit better, maybe do worse. It's up to your ears to decide. I've heard native DSD supporting DAC do worse, I've heard some that maybe do better. I haven't heard anything like the PS Audio DirectStream yet.

 

I would if I had the CD...not that I need any convincing.

 

No, no, no...
If you believe in Stairsteps, then it's only logical to believe that DSD has no resolution, because it is just one bit.

Now, you see these stairsteps? The signal becomes more analog the more bits, the more stairsteps. Because, if you believe in multibit-PCM hi-res audio, it has more bits and more and finer stairsteps.
The common audiophile believe is, the advantage of multibit-PCM hi-res is not just noise, but finer details because of the finer stairsteps. 24 bits is more than 16 bits.
If you believe that, how could you believe in DSD with only one bit resolution?
To me, therefore, DSD with its just one bit of resolution, sounds very grainy and unnatural. Like, you get too much reverb tails, which is quite annoying.
How can't you not hear that grainyness of DSD with its only one bit of resolution?

These staristeps figures point into the right direction:
We don't need less bits, we need more bits! We don't need lossy compression like MQA, but instead true hi-res audio!
True hi-res audio only starts at 32 bits, but what is really needed is 128 bits of resolution.
And we don't need 192kHz, not 384kHz, but the full 1536kHz of sampling frequency.
Everything less is just a bad compromise, when you compare to full analoge resolution.
MQA instead reduces bits and sampling rate, which is the wrong way to go.

 

At this point you're just arguing to be constrain.

You have made no point.

And for the record, I actually know how digital works. In detail. I've read through all of Pohlmann's "Principles of Digital Audio" and other books about digital audio. I understand the difference between PCM and DSD. I don't need your contain and incorrect lessons on the subject.

 

I used to work for Rolls-Royce, in their civil engine group. Some years ago, I helped reduce the stock on the shelf by having some parts re-machined and thus able to be redeployed as a needed new part on an aircraft engine.

I spent a lot of time reading blueprints, and once I probably could've told you the unit price for each part on an RB211 type engine.

Didn't make me an engineer though. I was the guy from IT on secondment. It's not enough to just do the reading, but to know how and why something works as it does and the subsequent effects that that has.

 

Thanks. I actually double majored in engineering. I think I can figure out the maths and theory to understand digital audio enough to at least make it through the Pohlmann book. If I want to understand it further than that I'll ask someone like Black Elk.

 

Good for you! Congratulations. Maybe you should design your own DAC then. We'd be all ears I think...

 

I’m not sure why you answered @HamSandwich or @Markc the way you did, so I’m not sure that your point was understood. I plead with you and some other members to please not - directly, sarcastically, satirically or earnestly - repeat or promote the nonsensical stairstep myth here. The stairstep analogy has been repeatedly disproven. It’s simply not an accurate graphical representation of a digital waveform. Also, DSD samples rates are, at a minimum, 2.8MHz, which is 64 times the sampling rate of Redbook CD. The DSD process is a form of pulse-density modulation stored as a delta-sigma modulated file. The file is built out of data with one-bit values, sampled at (at least) 2.8225 MHz. Your one-bit alarm is pointless because it’s irrelevant in such a high-rate bitstream.

I think it is best not to confuse or conflate bit-rate with word length. The two things are related but different. More inaccurately, you’re drawing a word-length comparison between pusle-code modulation and pulse-density modulation. In a pulse-density modulation system, the shortest practical word length (one-bit) at extremely high sample rates (2.8225 MHz or higher) is precisely what the system calls for in order to create the highest resolution bitstream commonly available today for audio recording.

IMO, we should not use the stairstep analogy whether we’re being critical, sarcastic or not. The digital stairstep analogy is ridiculously inaccurate, it’s promoted by people who are repeating nonsense but who’ve never worked at a test bench, never paid attention in school, and never read an accurate textbook on the subject.

 

Because you all conspired to confuse me with bits I decided that to migrate all my music library and future purchases to MQA. That will teach you a lesson!

 

With due respect, your intuition is taking you in the wrong direction.

My advice is to read, learn, and understand this (yes, it's math. And to many, it's hard math). I'd be willing to bet a substantial amount of money that once you truly understand the FFT, you'll give up on the stair steps.

 

If You are ironic or not somehow gets lost.

 

Anyone gave my test a try, by any chance?..

 

Seems to be a lot of snotty comments flying around in this thread, even some aimed at the host. Wow guys, do you really care that much about others perceptions of high res audio? Dig what you dig and let others do the same. Jeeze.....

 

Ah, math. You want to believe in math and theory. But, I don't want to analyze the music with an oscillogram, I want to enjoy the spirit of the music, the emotion. And that's something missing in all your theory and formulas.

It is common audiophile wisdom that 24 bits are more than 16 bits, and 96kHz is better than 44.1kHz. With earnest concentration, everybody can hear the difference, provided the equipment is good enough and the subject is able to hear the difference.
But, why leave it with 24 bits?
32 bits is better than 24 bits, and 128 bits is even more better.
And the same is true for the sampling rate: 192kHz is more than 96kHz, and 3072kHz is even more better than 192kHz.

I always was wondering, why such obvious differences between 16 and 24 bits are so hard to hear?
My conclusion is, it's not the bad ears of the subjects, nor the bad equipment. But simply, the improvements in details and sonic textures are just too little to be worth the effort.
Only a real substancial increase in both bit depth and sample rate will be able to reproduce all the fine details of the analog source.

As I understand it, DSD (SACD) is only one bit, but at 64x oversampling. That 64x oversampling alleviates some of the disadvantages of the DSD one bit resolution, but you cannot have a benefit without paying for it.
And what you pay, it is a feedback loop in the Delta-Sigma, required for the noise-shaping of higher orders.
I feel, that feedback loop introduces artifacts into the audio.
SH has reported about prolonged reverb tails in DSDs, and my strong feeling is, these are just spurious sonic events, something like "ghost audio" (similar in effect like print-through in analog tape). So, this is not a proof for more details in DSD, but quite the opposite, technical artifacts.

To overcome all such shortcomings of present digital audio, I feel, only a real substantial increase of both bit depth and sample rate is able to do so.
Not engineering tricks like "noise shaping" and Delta-Sigma like with DSD-SACD, not tricks like lossy compression like MQA. You cannot fool physics, that is.

 

@Tim Müller why not sample in the terahertz range if oversampling is such a great idea? I know, some of us might not like glowing in the dark or getting incinerated if the waves become coherent, but this is only a small vocal minority of perpetual discontents.

 

Simply, there is the law of diminishing returns.

Now, consider sampling with just one bit.
Fine.
Add one bit, and you end up sampling with two bits. That's a double.
Now, add another bit, and sample with three bits of resolution. That's not a double, just a 50% increase.
Now add one more bit, ....

You surely get the idea.

So, my feelings are, with double or 4x the DSD sampling rate, you should be fine. But, who knows...

What sampling rate and bit depth do you propose for optimum audio quality, covering all the fine details of the analog source?

 

The "not sure if serious" is getting strong in this thread.

 

@Tim Müller You’re really sending this thread off the deep end. None of this has anything to do with MQA, and to top it off you’re dead wrong about DSD because you’re fundamentally misunderstanding how it works. Please stop posting nonsense, as you’re looking more and more foolish with each post. Several members here have graciously given you detailed explanations of why one bit is sufficient for DSD, yet you persist in your erroneous postings.

Your postings also contradict the basic reality of how we hear DSD files. If there were “no resolution” no one would be using it. You have to apply common sense.

 

Yup. There's that pesky Rational Thought again. Who needs THAT?

Math doesn't care if anyone "believes" in it or not. Simply, it exists. I don't "want" to believe in math, and math does not belong to me.

My only point is that your "stair steps" thesis isn't supported by scientific fact (you know, that "math" thing).

 

Mr Müller, I somehow get the impression You are all over the place.

 

This whole thread

 

Wrong. What our host was stating was that DSD fully captures the reverb tails that ARE ON THE MASTER TAPE while 16 bit/44.1 digital truncates those tails somewhat. To prove his point, he listed a Creedence Clearwater Revival song he mastered in both formats.

 

That's different from what I hear. Maybe your ears react different, I dunno.
I feel that DSD is not really an authentic representation of an analog source. It lacks detail, and adds artifacts.

 

I love fallout