Digital sound is not only about ones and zeros

[AlanS]

Alex

Apart from my gut instinct reaction to your post last night I have disected you thoughts and the faulty assumptions you have made in coming to the conclusions you have.

The issue with digital anything is that it can be viewed both as a blueprint and as a final product. Unlike in non-virtual world, where even the least educated person has no trouble distinguishing between a blueprint of a chair and a finished, physical chair one can sit on, in the world of digital that distinction is not that obvious to the uneducated.

Alex who apart from you is making this issue assertion about blue prints and final product? I have not seen that model of thinking used before and from your posts here I note you are not an authority on matters digitial either by qualification or long standing experience. Neither am I but perhaps that is how I recognise you aren't

How did you come to this notion was it in a flash of inspiration or long chewed up ideas?

Can you give practical examples of blueprints?

When musical performance gets recorded, the information captured on tape can be converted to a digital blueprint.

You say CAN be converted how often you think recordings are converted to these digital blueprints you assert.

How does a digital blueprint differ from the original recording to tape or more likely hard-disk? mBoth are 100% copies as you say later.

As I suggest digital blueprints do not exist, kindly indicate what one looks like, if I was looking at one what would I see. I can see a tape or hard disk and directory structure. I can read what is on a tape or hard disk with suitable equipment, what could one read/verify your digital blueprints with?

This digital blueprint can then serve as a recipe that will guide the machinery on how to produce the final product -- in this case, the final product would be the weak electrical current that will be fed into a DAC.

However, the digital blueprint itself can never be listened to; it can only be regarded similar to how we regard a recipe -- i.e. a piece of paper sitting in a drawer, ready to be pulled out and used when preparing a meal. But one cannot satisfy one's hunger by eating that piece of paper.

Ever heard of mixing your metaphors?

In that regard, we see that indeed it all boils down to ones and zeros. This 'recipe', or music blueprint can be copied indefinitely and at will, and each subsequent copy is guaranteed to remain 100% identical to the original blueprint.

However, what is NOT guaranteed is that each time this recipe is used to prepare a meal, that each meal will taste identical to the previous meal made from the identical recipe. Or, if we stay with the blueprint analogy, two different shops can use the same blueprint to build a chair, but these chairs will NOT end up being identical.

So we have digital recording, blueprint or recipe all 100% faithful. If you think analogies then somehow diiferences are introduced, if you think source then no problem.

I think you have fallen foul of ascribing the characteristics of the analogy to the original that the analogy was made to. because food differs digital sound will do.

A meal is made from differently sourced ingredients, prepared by cooks according to their interpretation. They therefore differ because they are a human production task. There is no equivalent in taking a digital recording and producing a CD or file for distribution. It is a copy as you yourself said a 100% copy.

In the above case, we see that it's not only about ones and zeros, and that the final product wildly varies depending on many other factors and conditions.

There is only wild variation in food and chair production from blueprints the analogy, with digital sound, no analogy, unless someone deliberately chooses to alter the recording, pre DAC or CD it is a facimile, Food and chairs introduce the human factor/interpretations.

If I have missed something apart from your faulty use of analogy please put me straight. A digital recording isn't used as a blue print/recipe, it is a digital recording that produces identical copies unless interfered with.

Perhaps you can cite some practical examples of where your variation/loss theory would apply.

[Themis]

Above, dealer supplied but with deluxe box and fancy display - £1,000

Above, Top End dealer, limited production, boutique bits and very expensive casework, probably bodged to hell away from true neutrality, but hey, it "sounds good," - several £K

Dave,you make the assumption that limited production and heavy research are two different things, here.

You also make the assumption that internal parts all cost the same, whatever the quality of the parts and/or the implementation....

If I follow your reasoning : My Denon's internal dacs cost 10$ (research and production included), so, I can conclude taht all dacs above the -say- £40 price are ... what ? Expensive casework and boutique bits ? Well, I doubt.

[DSJR]

May I sort-of agree to disagree here?

DAC chips these days do most of it for you I think and all the designer of the finished product can reasonably do is tinker around with the power supplies and filter/output stage. I dare say the Denon, if it's a multi-channel one, will have had most of the cost gone into casework and costed over a limited lifespan, since the next generation of chips will be but months away. If it's an expensive two-channel DAC, then it's profit all the way - honestly. Don't let these people tell you that it costs hundreds of thousands to design audio equipment. It doesn't, unless it's a very small scale manufacturer (Naim is a minnow in the wide scheme of things in all honesty) trying to have really good casework made (the aforementioned naim cases are superbly done these days), the DAC chip is a few quid - honestly - and passive components are twenty a penny if bought on a gondola/drum, as most are these days for production.


I heard the truly wonderful little Rega Brio-R amp yesterday. The case is full of transformer and everything else is shoehorned carefully round it, although I reserve judgement on the dome-head bolts holding the top on. The Rega DAC is built the same way too and just totally disappears from the audio proceedings in use, not ONCE drawing attention to itself. BOTH can be bought for well under a grand total. "We" already know also what a Gatorised Caiman can do for even less and again it's the casework that IMO is the only thing that could reasonably be tarted up - that and maybe incorporating the power supply inside, giving possibly more scope to tweak the supplies to individual parts of the circuit. I'd love to say that a DCS, Chord or similar expensive creation would blow it away, but I really do doubt it, not these days, since many of the things addressed in sooper-dooper DAC's are almost certainly way below audibility, I suspect largely academic interest only and done for "mine's better than yours," just like the THD measurements were used in the mid 70's before the likes of Linn and Naim came along and brought about a re-appraisal of exactly what it was we were supposed to be listening to - the MUSIC!!!

Sorry to go on and possibly with confused thinking here. There is so much reviewed in the likes of HiFi News that sells for £5K, yet inside has a very few tenners worth of bits, the total cost made up of very expensive casework and greed...



P.S. I justify the above by stating that despite my slightly tongue-in-cheek comments to Simon B before I bought his QED Opto-Digit and Positron, I've been very surprised just how good this ancient Bitstream thing is. The circuit was built exactly to the Philips build-notes and all that was done was the Positron supply, which I believe has twin filtering/regulation on both positive and negative lines to make it "special." The "Twin Supply" mod I'm considering could easily be done in one box and probably with a larger? transformer with two secondaries, all the regulatiuon etc being done on a small slave board alongside.....

[Themis]

Well, lets have another example :

How much to you think this dac should cost (all discrete parts) :


The size is a full-size component (that is 4 times the Caiman).
Now, let's say that someone has to design it, produce it and sell 1000 pieces of it for 4 years.

How much should it cost ?


Or, are you trying to explain to me that everything should be made like my Squeezebox, which, for £200 has a dac, two network connections, a color touch display, a remote, a network streamer, a processor and plenty of other things in just a few integrated components ?
Edit: What I mean here, is that the Rega dac, compared to my Squeezebox internal dac, costs 10 times more for no reason, exactly for the same reasons that you complain about the £££ dacs in a beautiful case, should I treat Rega as thieves ... ? Or do you seriously consider the Rega dac is better engineered than the Squeezebox dac ?

[magiccarpetride]

I really have the feeling that your "recipe" analogy is there to describe the recording process in general, and not so much the digital copy of it...

No, I was only focusing on the digital copy -- the copy that consists of nothing but a series of ones and zeros stored somewhere on the auxiliary medium.

Moreover, DSD digital is quite different from PCM.

Agreed, and good point, and for the purposes of my discussion I'd leave DSD out of consideration.

In my opinion, whatever the archiving media used (be it digital or analog) the reproduction problems remain complex, and, the reproduction technology used (be it digital or analog) does not guarantee an easier way to a more "faithful" result.

No objection there. I was strictly narrowing my attention to what happens between the stages when a 'bit perfect' blueprint of the music (i.e. the collection of ones and zeros stored on disk) gets read, transferred via ethernet cable into the Touch, and then from there streamed via spdif into an outboard DAC. My analysis stops at the point BEFORE the stream hits the DAC, nay, BEFORE it hits the spdif output.

Of course, once the signal enters the DAC, and then gets processed there and then gets forwarded to the preamp, power amp, and finally down to the speakers, all hell may break loose (and I'm not even dragging in the issue with cabling). But I'm not discussing that part.

If your aim was to demystify "Perfect Sound Forever", i agree with it. Otherwise, I must have missed the main point, probably.

Yes, 'perfect sound forever' is bullshit, on that we all fully agree.

[Tim]

Yes, 'perfect sound forever' is bullshit, on that we all fully agree.

Why? Are you saying a properly archived digital file with a corresponding MD5 hash value to ensure it's integrity might not sound the same in 15 years time, cos' if you are that's total rubbish.

[magiccarpetride]

Don't agree. The digital copy should remain bit perfect which means you should be able to read it a million times and it should be correct and identical every time. Think of the digital copy not as a recipie for making a pie. Think of the digital copy as an instruction set for a CNC machine. You feed the instructions to the machine and that machine will produce the same item a million times identically.

I agree with you that a digital artifact (i.e. the music blueprint in the form of a file that resides somewhere on the disk) should remain bit perfect. In other words, you should be able to read it million times and produce a copy that is 100% identical with the original. I've never contested that, as it is the starting point of my analysis.

This is similar to how you may hire an architect to make you a blueprint for your summer house. Once he's done, he is not going to give you the original blueprint, but a COPY of that blueprint. However, this copy is pretty much the same as the original, meaning you can safely take it to the contractors and they will build you a house off the copy of the original blueprint.

My point is this: if you give the same copy of a blueprint to five different crews, you'll get five different houses, not five identical houses. Same applies to digital transports -- you can feed identical FLAC file into five different digital transports, while everything else in your environment remains the same, you'll be hearing five different renditions of that same FLAC.

Where 99% of the deviation in playing digital music comes from is converting the digital to analogue. The DAC section. This is where the problem is. Every DAC does it slightly differently, and every DAC sounds slightly different (for better or worse).

I agree that DACs play important role, but I disagree with your lopsided over-evaluation -- DACs are not contributing 99% of the deviation. I don't know what the percentages are, but they certainly are not 99%.

[lurcher]

No objection there. I was strictly narrowing my attention to what happens between the stages when a 'bit perfect' blueprint of the music (i.e. the collection of ones and zeros stored on disk) gets read, transferred via ethernet cable into the Touch, and then from there streamed via spdif into an outboard DAC. My analysis stops at the point BEFORE the stream hits the DAC, nay, BEFORE it hits the spdif output.

Well up to that point I see no reason wht bit perfect can't be attained. It may not be so in reality at the moment, but the only reason I can see for that is crap software in the chain.

I had a thought the other week, that it should be possible to use a spdiff reciever and FPGA to make up a device that calculates and displays a checksum on a incoming audio stream. It would then be possible to play a audio file into it and see if what came out the spdiff reciever was actually the same as what started life on the hard disk.

If I could find a way of inserting extra hours into the day I would have a play with that idea.

I guess you could do the same with a sound card and PC that had a spdiff input.

[lurcher]

you can feed identical FLAC file into five different digital transports, while everything else in your environment remains the same, you'll be hearing five different renditions of that same FLAC.

Well, if by hearing you mean the data that is sent to that DAC, then I would say at least 4 of those transports are broken.

[Themis]

Well, if by hearing you mean the data that is sent to that DAC, then I would say at least 4 of those transports are broken.

Well I get what you mean, but let's make a test :

- We take ONE CD player and feed it with the same CD.
- We count the C1 errors (lets stick on the simplest error, for the example).

If we read the CD 10 times, we will get 10 different C1 error count totals.

So ? Is the CD player broken ?
(the answer is no, btw )

If we make the test with 10 different players, we get 100 different totals....


On the other hand, in the real-world digital transport,there is a difference between "data sent by transport" and "data processed by dac".