Electronics and reality part company below 20kHz?

[counter culture]

Electronics and reality part company below 20kHz.

The other day I came across the following statement from Earl Geddes, a prominent contributor here.

'The funny thing is that no one cares. You see nobody believes any measurements in audio anymore so the fact that THD and IMD are meaningless is of no real significance. And replace it with something meaningful!? That would only confuse people more and well, marketing doesn't really want something meaningful, there is no way to put a "spin" on that.

You have to ask yourself what's wrong with a marketplace that prefers to use things that are completely meaningless when absolutely valid representations are available. And its not just distortion, it's in everything about audio.'

I started to think about all this stuff because of a recent review on Head-Fi: Head-Fi TV, Episode 011: Fostex HP-P1 Portable DAC for iPhone, iPod and iPad (With Built-In Headphone Amp)

This new product from Fostex features a new DAC from AKM, the 4480. It's prominent in the advertising, and it's clearly headined as being a '32-bit DAC'.

If you're a digital engineer, and these days it's difficult to be an electronic engineer without being a digital one, at least to some degree, you've probably come across the following equation:

SNR = (6.2 * n) + 1.7 dB

...where n is the number of bits of sample resolution.

What this means is, the notional SNR of a 32-bit DAC (if I can borrow a term from finance) is 200dB. Looking on the Asahi Kinsei site, they quote a SNR for the AKM4480 of 117dB (mono) and 114dB (stereo).

Similarly TI show a 24-bit DAC with a potential SNR of 150dB as having an actual value of 127dB (the best they offer).

In contrast, 16-bit DACs often achieve close to their theoretical SNR.

So what's going on?

What's going on is that you can take TI's 24-bit DAC and throw 24 bits at it and it will only deliver 20-bit performance at best (better than the [32-bit] AKM4480 @ 18 bits).

What does it mean?

Well, not to put too fine a point on it, it's the end of the world.

Even a huge company like TI has been captured by the admen.

What's going to happen when the rot gets to Intel and AMD? You'll go buy a 128-bit microprocessor, and only 80 or 90 bits of the data bus will do anything.

It's just not possible to engineer in this context.

And let's face it. No engineering... no recorded music... end of the world.

You think I'm joking. Files with identical checksums sound different. Get a grip for xsake.

[thaumaturge]

Originally Posted by counter culture

Similarly TI show a 24-bit DAC with a potential SNR of 150dB as having an actual value of 127dB (the best they offer).

In contrast, 16-bit DACs often achieve close to their theoretical SNR.

So what's going on?

Noise floor. Black body curve. Can only go "down" so far before you encounter an ocean of noise.
Doc

[jcx]

24/44.1 "hi res" Beatles - likely 70-80 dB source at best

I think the more balanced approach to a "hi res" release would probably be 88 or 96 k at 16 bits
the extra BW allows noise shaped dither to give better than 120 dB audio frequency linearitiy, weighted S/N
and eases concerns that maybe someday someone will "prove" 44.1 is too low a sample rate or that reconstruction filters for 44.1 are audibly harmful

but it requires some engineering, psychoacoustic knowledge to appreciate the practical "resolution" added by dither and the advantage of BW to the dither noise shaping

http://www.meridian-audio.com/w_paper/Coding2.PDF

[DF96]

"You see nobody believes any measurements in audio anymore so the fact that THD and IMD are meaningless is of no real significance." ('counter culture' quoting Geddes)

Slight exaggeration here. Some people don't believe any measurements in audio and regard THD and IMD as meaningless. Other people believe some measurements and would not want THD and IMD to be too high, but recognise that they don't tell the whole story. I guess some people still think, or want us to think, that THD and IMD are the only things which count so 0.0001% is 10 times better than 0.001%. Both ends of this spectrum tend to argue by exaggerating the perceived errors of the other end, while hoping we won't notice the sensible people in the middle.

I would be interested to hear what his "absolutely valid representations" are. Does he have an 'audio quality meter', or just a claim to golden ears?

[SY]

Originally Posted by DF96

I would be interested to hear what his "absolutely valid representations" are. Does he have an 'audio quality meter', or just a claim to golden ears?

He has published his metric along with some simulations and analysis tied back to loudspeaker evaluation. I'm unaware of anyone using this metric to distinguish a "good" amplifier from a "poor" amplifier where both have good conventional measurements. Or where the "good" amplifier has horrible measurements (e.g., Wavac, Jadis) and can be expected to be a large, inefficient effects box by benighted souls like you.

[abraxalito]

Originally Posted by counter culture

Similarly TI show a 24-bit DAC with a potential SNR of 150dB as having an actual value of 127dB (the best they offer).

Nope, they have a couple of DACs which do better - 132dB - in mono mode.

You also missed that TI has a '32 bit' DAC - the PCM1795. But compare its datasheet with that of the '24 bit' PCM1796.

So what's going on?

Obviously excessive cynicism.

[vacuphile]

you know, Counter Culture, the next byte up from 24-bit is 32-bit. So it is darned conventional thinking, in collusion with big capital, all over again, which delivered us these monstrosities.

What happened to the baricades, btw?