16 bit files almost unlistenable now...

[PeterSt]

12 hours ago, fas42 said:

As PeterSt would be happy to point out, the tiniest, tiniest variations in how the data is dealt with can be enough to make an audible difference

 

Frank, and @numlogfor that matter :

 

This is not really about tiniest variations. In order to understand that, we must go all the way back to how I designed the first NOS1 (which was and is nothing of any business as such) and which was almost all about and almost about nothing else than :

Compare Redbook with Hires. And to mention what my business as such could be about : proving that no Hires is any substantially better than Redbook.

What did I need to do ?

 

Create a DAC that behaved 100% the same for Redbook vs Hires. How to do that ? let all happen in software and let the DAC run at the same speed and bit depth always.

And so it happened (the software testifies of that up to date - see fx button).

 

But now the reason how this thinking came about :

 

Back at the time a used a FireFace800 (Firewire connected Pro device) and I wanted to compare Redbook with Hires. It bugged me because I could nit find "a one and only" setting that sounded best for Redbook and worked for Hires. How come ? the buffer setting best sounding for Redbook, was too small to let Hires (24/192) work without stuttering. I thus had to higher the buffer length for Redbook and now it did not sound good. Then Hires would win. But ...

But his is apples and oranges.

Btw, this was true for all Firewire interfaces, which I developed with myself, until the software department of those guys (which was a split company from the main chip manufacturer) gave up on my latency requests. This is also where I gave up on Firewire (good, because it died anyway, soon after).

 

The net result of the whole exercise back at the time was that I could work with buffer sizes as small as 1ms (round trip), sufficiently low to not compromise the Redbook playback as it was back at the time (must have been 2009) and way more high enough to let Hires play just the same. Now we had apples and oranges and the DAC always playing at the same rate and bit depth.

 

Still necessary to elaborate on the difference ? well, the electrical difference between the both situations - each tuned for its best performance and in the Firewire situation, was huge. Longer buffers, lesser transmissions, only because of 32 bits vs 16 bits. It is as logical as can be, *if* you only know that this difference in interface can exist. Firewire (FireFace800) could do it and the NOS1 also always has been able to do it (no matter the kind of interface and no matter the driver version). And all is determined by the selection of the driver (mode). So play a 16 bits file and the transfer is per 16 bits.

N.b.: I2S is always 32 bits, and that indeed is padded.

N.b.2: 24 bit interfaces also exist (as a pain) and they again work out differently.

 

ALL already differentiates in the playback software, once it is recognized as a possible interface difference. So the software plays differently for 16, 24 and 32 bits interface (and that all in Integer - not float).

Only when one does not believe this all can't imply an audible difference, then it doesn't make a difference.

 

PS: My business as such is to imply the best SQ for Redbook. So now you know.

 

 

[PeterSt]

3 minutes ago, fas42 said:

A piece of software that has "millions" of possible settings is really what I was pointing to - if digital replay is so well sorted, why is this necessary - and not actually asking you, Peter, .

 

Haha. All I did was trying to point out that ahead of the tiny variations, first the rough stuff must be understood. I mean, if we don't see that the selection of a 16 bits file vs one of 24 bits implies a 16 bit audio word stream vs a 32 bit (no typo) word audio stream, or requiring double the bandwidth and processing activity (possibly not only at the PC side but also at the DAC side), then the fine variations are a moot thing. And btw Frank, I did not try to teach you or anyone else - I was only pointing out priorities of interest (and of course how a 16 bit stream vs a 32 bit stream, the last 16 bits zero-padded, will sound different indeed).

Once we have that behind us, there's the "millions" of possibilities of e.g. controlling the various buffers, a.o. (!) the one we are literally talking about (the buffer which is filled by the playback software to the driver, which from there has its own control apart from driver buffer settings, if possible). FYI In XXHighEnd these are 3 settings main settings, for Kernel Streaming a bit differently set up than for WASAPI (which uses Kernel Streaming itself, internally). There is also the setting which eliminates half of the normally double FiFo buffers which setting by far sounds the best (it consumes half of the bandwidth in use otherwise in all components). This setting ran into forgetfulness because NOS1 users (like me) can not apply it.

Anyway, on top of this all is the relatively infinitely more rough (for implied setting) physical buffer size implied by the 16 bits vs 32 (or maybe 24) bits stream.

 

Quote

if digital replay is so well sorted

 

It thus isn't much. Or maybe it is (as in bit perfect is super easy these days contrary to 12 years back) but we learned to control SQ with it anyway. 

[PeterSt]

3 hours ago, gmgraves said:

As I walked down that side of the street, passing establishment after establishment I could note to myself: there's live music playing here, but it's a PA system there, and here again is live music and here again is a PA system.

 

Of course. That is because each band is sitting right in the shop window (say etalage in French ). No band ? then it is PA.

haha

[PeterSt]

7 minutes ago, gmgraves said:

Do you believe that High-Fi can be perfect like Frank does ? 

 

George, yes. It is even my very example of the live bands, that once you can't differentiate the band from the stereo when walking on the street and listening from outside, the stereo must be the better one. I use it explicitly by means of walking outside (around to corners and then in the garden) to observe how "live band" I regard it.

I also have the toilet test, which is simpler but I think more down to the merit :

Go to the toilet, close the door which is good for the test and other reasons, also close the door to the living room (quite normal but again good for the test) and observe the still audible frequency spectrum. If things work out, you will here the same spectrum as in the living room (the stereo assumed there). Overweight on bass ? then things are not right. Hardly highs ? then not right. These both though, merely express the general merit of the system, like you referring to 1000W to be necessary for etc. When I read that, I thought to ask you for the math on my 118dB sensitive speakers and the knowledge about the 2x brigded amplifiers of 30W for the bass section.

And that this has been measured for 89dBSPL maximum THD of 2.7% (which is inaudible at 20Hz because you hear nothing indeed).

And that I out of everything could do the math on kick drum surface, pressure (mm of dent-in (excursion)) and reflect that on to woofer surface per speaker side (which is 3x15" in my case) again including the excursion of the woofers (which is depicted by the THD desire).

blablablabla

IOW, I sure try.

 

But I think you read my "report" about the drum kit being indistinguishable in all aspects ... this was in this very thread IIRC.

 

The toilet test is the real test because as you will notice (try it !!) that certain frequencies jump out. So now this is not about inherent merit like good sub bass or undistorted loud highs, but about how equal the performance is.

Did you try it already ?

Still not ?

OK, take your time with it. But if you do, you will notice that you have frequencies jumping on to you. And the biggest fun : you don't observe them audibly in the listening room. You are used to them ... (there).

The toilet hides nothing from you. 

All in other words : to a certain extent, if the toilet test succeeds (all frequencies come to you in even fashion) the "live band" will be close. No guarantee yet (because of rolled off bass and highs) but better yet. And the point is : we (our brain) work with such information. Buzzing ans standing waves and all what a stereo can do wrong ... it is not the real thing. Well, you know that.

 

[PeterSt]

Wat is sad, is that people can hide behind bad recordings, which in my view is the most weak "act" because it is so easy to diminish everything over it. The reason I'm saying this, is that I personally don't see bad recordings anywhere. I do see, however, many, many (many !) recordings which behave poorly just because my system apparently (!) can't deal with them. This is often related to "eras of recording" or types of music and instruments or a combination of both. Example of a last whole group that I could make my system suitable for, is stuff like Kiss, Boston, Foreigner, Toto, Journey - don't ask me how to exactly name/group that, but I suppose it is about how the fullness of the distortion guitar is applied (like full chords on them and then three of them together). So while this is music of my younger years, only recently (2-3 years ago) I could satisfactorily play it in my system (and at some decent level - which usually is crucial for something really working vs being painstaking).

 

At some stage, maybe 30 years ago, I had 800 or so CD's and I recall that I started a journey of having those CD's and a system, more than half being bad recordings, and how to improve on the playback of them. My CD rack was divided into areas with bad/poor recordings and just good ones. I could see how far I was with tweaking by the number of CD's in each department. This with the notice that I far more play(ed) the bad recordings because they need improvement. I ended up with FIVE bad recordings, at a moment I meanwhile collected 2000 CD's. Say that it took me 10 years to get there. Always tweaking. And btw, always the same system. One thing : my goal for audio life is making the poor sound good. And, of course be so ignorant that you think it can be done (contrary to the consensus of this thread, for some - no worries I am not accusing or anything).

 

The better the system, actually the harder it gets. For example, with almost real level cymbals today, while quite a while back I was happy to perceive some reality of a hi-hat (which was the first major achievement were it about SQ in general - unrelated to poor recordings) - read : when you hardly hear a hi-hat which is distinct and hardly anything to distort because it is always showing at a fairly low level (and often the only instrument in the drum kit that needs amplification at a performance), there is no fear to have from normal cymbals, just because they are of a too low level, too grey and they can't hurt. But wait until they start to show real levels (which would be 110dBSPL at 1 m) ... then things get infinitely more difficult.

 

Today I have 54K CD's and although I will never be able to play them all, readily by heart I can not tell about baddies in there, or bad or poor recordings. However, I have a "demo" gallery which contains some which can be improved upon, and I guess this quest will never really end. Oh, Michael Jackson is in it, with Bad and Thriller and the like. They G-D one day will play fine in my system, although you can sense from any angle that they only perform on PA systems. Still I keep on trying. And mind you, a LOT is about the stupidly synthesized drums of the 80's which ... well, still can vastly improve. Recognize Enigma with its distinct grey sound in the highs ? ... it can turn into completely normal fresh highs. Really. ELO, same thing but more (improved).

Those who visited me know that I play Get Back of The Beatles to them, and without exception they are sure it is a cover band who recorded it a year ago. But it is not and it is the super grey sound with something which looked like cymbal hash back in the 60s, which now sounds like super real normal cymbal with the most beautiful recording technique. You just can't imagine and therefore you won't believe it either. OK, you don't believe Frank and I am easily with you. But claiming that he thus mesmerizes all into poor recording and therefore all sounding the same (and good to Frank's ears) ? no. Well, sort of No. Point is, I do it too, although maybe per different means.

All, OK, almost all hits from the end 60's and 70's, sound like brand new and from yesterday but better. A while ago I put up a list of it (stuff like Give up your Guns and 40 more) .

 

Moral : out of 1000 bad recordings as how they come to "you" (not addressing anyone in particular) the chance is very fair that NONE remain poor when you play them in my system. Most probably you can't even define poor, other than from some idea of wrong microphone placement, perceived wrong panning, flat as a pancake whatever sourced reason or plain shrill because of no-idea-how. It is all not true once you represent it in a quite infinitive "tweaked" means (but all formally ending up in commercially available means), which btw still is on-going but not anymore to improve recordings. To make them super special, yes. Some ^2 cable is the latest example of that, this making everything down-right spooky and every piece, disregarding the type, a thriller or "play" and theater. It turns out that we don't need Mahler or Stravinsky or Bartók for that, because actually all music makers are artists, and this shows. So don't think that e.g. Ian Dury or Elvis Costello and even Grandmaster Flash (and some furious five) and 1000s more are not artists. They play for you. And btw, they don't play for recordings really. They make theater.

 

Let's keep in mind : all I have been doing the past years is improve the 16 bits 44.1 stuff, hunting for one thing only : genuine representation of the real thing through loudspeakers. I don't even attempt 24 bits or higher sampling rates.

I am in the special opportunity of "being able to" control 100% of the playback chain; Playback software including Operating System tweaking, Audio PC, D/A converter, amplification and loudspeakers. Oh, and some gadgets like cables. Point is : many of you use one part of it only. Now, all will admit that it helps them, or otherwise it wasn't used. Now use two. Or three (like software, PC, DAC - or - USB cable, DAC, Interlinks). This is still far from all. But what you achieve with that will (hopefully) be beyond imagination. So moral :

 

Each element helps vastly, once it has been explicitly made for a purpose and was developed (coincidentally ?) in the midst of a consistent chain. Say that each element is a reference for the other, so we can build upon it.

If you stack all of these elements together, well, then ...

Then Get Back sounds like recorded yesterday but better ? What a great hobby this actually is ...

[PeterSt]

Frank @fas42, Let my rant be an example of how everybody can put up such an indeed rant from his own perspective. This is my second of the same in this thread. Possibly I did it similarly in maybe 5 occasions, one of them that thread of a year back where I put up that list of 40 or so hits of the 60s and 70s.

If I would put up my act as often as you do (which is in every existing thread in CA), I would most certainly be banned after the 5th in a row in say a week of time, because of self advertising. That is my "downside" and that seems to hold me back from posting again and again. You have no limits and keep on spreading the word. But people can't utilize it. It is empty. Trust me.

You are right, especially in your own context (which is how audio is anyway). But you can't tell people to tweak. Then better turn it into a product, label is snake oil if you want and then people can decide for their own. People buy it or they don't. Now people can not buy anything because you don't sell anything. Sounds odd eh ?

If I had to tell what I all did in 10 years of "tweaking" to end up with only 5 poor recordings ... I wouldn't even know. I also know that it would be completely useless to tell about a first wire to connect from A to B for better grounding. It helps me, but it won't help you. You don't even have a A or B. 

 

Done. 

 

[PeterSt]

5 minutes ago, diecaster said:

Yes, you have the most wonderful system on the planet !? Which direction do I need to bow toward? ?

 

Make that East.

It is not about my beautiful whatever system. It is about bad recordings which don't exist really. OK ?

[PeterSt]

54 minutes ago, numlog said:

and listening now this harshness is actually valuable high frequency data that gives the music an immense organic character that you will rarely encounter anywhere else 

 

Exactly.

 

54 minutes ago, numlog said:

and the 'good recordings' are now a bit boring and predictable.

 

Even more Exactly.

 

55 minutes ago, numlog said:

the best example of a 'bad recording' I had encountered a few years ago as a beginner to hifi was a lot of recently produced pop music, its loud as hell

 

Oops. This is so far from my "genre" that I didn't even think about that. Yes, all this modern bill-sh*t-100 is total crap for the recording, style and "artist value". It is nothing (to me).

 

59 minutes ago, numlog said:

A lot of aphex twin's work has a homemade/underprocessed sound to it,  with sub optimal gear it tended to sound harsh and seemed like it could be an example of bad recordings

 

Although not really my style of music, I am mentioning it more often - try Bucephalus Bouncing Ball (from Come to Daddy). It's a genuine "system tester" with its square sounding bounding ball which ever bounces faster (closer to the ground) and the square frequency goes up. Try that on a "poor" sounding system and you wouldn't even understand the purpose, as with almost exclusively all "electronic" music.

 

[PeterSt]

10 minutes ago, gmgraves said:

Is that not how everybody responds to bad sounding recordings? 

 

Not me.

Edit : The "of that work" is a bit dangerous, I'll admit.

[PeterSt]

2 hours ago, Teresa said:

Most of us have flown in our sleep,

 

All I recall from dreams in this direction, is falling from a plane. I couldn't fly ...

[PeterSt]

1 hour ago, pkane2001 said:

Point me to any papers or research by Linkwitz that remotely suggest that proper sound reproduction requires a soldered connection at the speaker binding posts as the highest priority.

 

I don't know whether these papers exist from him or an other. I do know though that a poorly soldered connection acts as a crazy antenna implying noise you don't want to see measured. But since I measure and hunt down any spuriae which doesn't belong, I know.

I have built synthesizers with so many parts in them that one took a year of soldering (not for a day job), so I know how to solder.

Not.

Once I found and saw what my "perfect" soldering did to audio cable connections plus that by then I wasn't the guy who solders everything together of what we (Phasure) provide, I explicitly will never ever touch the iron again when it is about soldering for audio. I take it that one who solders up to 0402 SMD almost blindfolded, will make "perfect" larger joints just the same. And it shows massively (measurement).

 

Maybe someone should write a paper about when soldering more poorly on the outputs of DAC (like the interconnect binding posts or where the internal wires regarding this spring from (PCB)), that this shows as 2nd harmonic distortion.

Could be 1dB only, but whenever I see this, these connections must be resoldered and it always helps. Of course you may like to claim that this is inaudible anyway, but this then is up to you. In the end these matters are about "accuracy" as such, which thus surely can be measured (THD) at this level.

[PeterSt]

Just now, pkane2001 said:

I should also reflow the solder on all PCBs and all components in my system.

 

I was waiting for that one.

No. It only applies to wire/cable ends. It never ever happend that I could measure distortion, and that it was about some part of PCB trace. Never. This, while 100x++ more components exist on a PCB than wires are in order.

 

We could try to ignorantly join George who seems to know everything better than we all together, but you can also invest some time in obtaining EMI meters of a few (radiation) sorts, and hold them close (no wait, a few ft will do already) to open ended cabling. What you see there is the ultimate of a poor connection. Make that connection, say, half-good, and half of the radiation disappears. Make it close to good and radiation is still there, which btw can be measured super easily anywhere in the chain (but take some audio output like from DAC or from (pre)amp and it will do). It will show noise of/in all frequencies.

 

So this can be measured and btw also heard as hum or noise or both. For the latter, all it takes is something like a 118dB sensitive speaker at full amplifier gain. So you see, I have quite different motives for avoiding noise. If I wouldn't attack these matters, noise would blast audibly through closed doors. And no, I don't like (to use) analogue attenuators.

Do the math for fun - I think you can do it. Take 30W (into 2-4 Ohm) for the amplifier.

 

Now on to all the sh*t we can not measure (not me) but which is ultimately audible as lesser quality sound. Whatever that exactly is. Go to Frank for that.

haha

 

 

[PeterSt]

6 minutes ago, PeterSt said:

Do the math for fun - I think you can do it. Take 30W (into 2-4 Ohm) for the amplifier.

 

OK, I think this would be a bit difficult without the gain figure.

20 on to 2.25VRMS DAC output.

 

But never mind because it will end up in what level of noise we will be able to hear in the first place.

[PeterSt]

Just now, pkane2001 said:

I've fixed (and built) many cables in my life

 

Paul, that's the problem. I did too.

In my case it apparently didn't do much for my soldering skills, while I soldered so many countless joints already.

Doesn't say a thing about your case, of course. If we could only agree a little that this does really matter (but measure it).

[PeterSt]

5 minutes ago, PeterSt said:

If we could only agree a little that this does really matter

 

I can tell you :

The only thing (really the only thing !) I physically make in this Phasure realm, is the Blaxius Interlink. I hate this part of my life as h*ll because it is such precise work and cut something 0.1mm wrongly and the cable is destroyed (while figuring that cut another 2cm to start over of one out of the pair gets noticed by the customer).

So say I am working on a pair for 20 minutes and then the coax cable needs to be soldered to the BNC pin. Such an easy job. But at that stage I hand the cable to the real skilled. When the pins are on, I continue with the cables myself again.

So that serious I am with this. Just because I know how it measures when doing it not the 100%. And btw it is not anything which would be visible (like dirt). It is about temperatures of surface and surface thickness which cools down the iron and length of time and ... well, don't ask me. You should see the row of flux materials we have here ...

It's about how two different materails (can) blend. Chemistry.

 

[PeterSt]

And how do we like the voice of Adele ?

Edit : in 16 bits of course.

[PeterSt]

8 hours ago, pkane2001 said:

1. A loudspeaker can never do better than to accurately convert electrical signals into acoustic signals. Thus the source material determines ultimately how well an illusion can be created.

 

Just in case people start listening to Frank, I can make it worse for you :

 

When one adheres the theory of the lower the THD(+N) the better it will be (as some absolute statement which can't be subjective to anything except for thinking that less distortion is better than more), all comes down to bringing down THD + Noise, until I can't measure it any more, but still can expect what will help the THD still. On a side note : less jitter is the other phenomenon but way less easy to measure (for me).

 

Now, what it makes it worse for you (up to insanity) is that I claim by empirical evidence (my own) that it hardly matters where you take out the distortion. I actually can stop right here, because *if* indeed the THD is the means to get anything right, then what's there further to explain. Distortion is everywhere, and taking it out of one place helps. Taking it out of another place, again helps (the aid adds up). Putting back the first will exhibit about the same quality as when both are switched around (and it doesn't matter really what aspect of quality we talk about, but with the notice that THAT is subjective - and it is already in relation to your system which is unique).

 

Some times I put forward the example of Jan Garbarek's crazy sax (with Adele's voice equally difficult) and how some visitor of a friend's system (also audio manufacturer) brought his opera singer wife who couldn't stand the sax and so it was tweaked by the owner of the system until the opera singer approved it (mind you, this is almost Frank's story). It took 2 hours or so throughput. I could clearly hear myself that the pain staking sax turned into something more likable. 

This is how the first "dial" in XXHighEnd came about and really one day later I could do exactly the same by software tweak. Since that time (which was even pre-officialXXHighEnd, say officium, haha) I can play that sax (ok ok, through speakers).

Today the same sax (yes in that same church) is spell bound. Mystique.

 

 

[PeterSt]

4 minutes ago, pkane2001 said:

Your claim that every single distortion is critical and must be taken out is patently false.

 

So we let it all be, especially for you. Or am I overdoing it now with the "all" ? 

Patently false eh ? wow.

 

Do you like or love or require distortion for some reason I can't guess ?

[PeterSt]

6 minutes ago, pkane2001 said:

THD+N or RMS jitter or any other single number isn’t going to properly characterize a complex nonlinear system.

 

I would agree with that. But it brings us nothing.

Or does it ?

[PeterSt]

Just now, pkane2001 said:

Yes.

 

Do you claim that all with a THD+N below -120dB sounds the same ?

[PeterSt]

2 minutes ago, pkane2001 said:

If all distortion is below -120dB it will sound the same when playing a regular recording as you reduce distortion+noise level to -130dB, -150dB or -200dB.

 

Quote

As I said, THD+N is a single number and does not describe fully a complex system.

 

Should it, for some reason ? I understand, it would be convenient. But ...

 

This goes nowhere now. Blame me. 

 

PS: Not that it went anywhere before.

[PeterSt]

3 hours ago, diecaster said:

4 hours ago, fas42 said:

My experience is even more "extreme" - highly capable playback takes the room entirely out of the equation, subjectively - the acoustics of the listening space cease to exist, as a meaningful part of what you're hearing.

 

Wow. You actually think your "highly capable" playback system can defy physics and make the acoustics of the listening space irrelevant.

 

Take up your guts and come over.

If necessary for a beer only. 

[PeterSt]

5 hours ago, fas42 said:

How much the activity of one component impacts another in the chain, negatively, via any mechanism

 

That.

[PeterSt]

2 hours ago, fas42 said:

There's country and there's country ... this album blows me away - a peak listening experience, every song does it for me ...

 

Frank, FWIW, Country as (I suppose you too over there) and me know it, is not how the American knows it. We (thus at least over here) dedicate all "Country" while in the USA there is a sub-genre in that - I tend to call it the low-level (cheap) poor man's music. It is hard for me to come up with examples because I wouldn't be able to differentiate easily, so one of the USA guys could maybe do it.

What I generally hear is that there's Folk and Country, which for me can't do it either because I think Folk is much broader (these days merely named the Sing a Song Writer genre ?). 

Btw, we actually don't know "Folk" (so go figure).

 

Btw, what the Americans think of this will be the correct stance - just saying; "we" just don't know about it much.

If I am not speaking for "you" then apologies. I most certainly speak for me in the Netherlands.

[PeterSt]

2 hours ago, fas42 said:

When you play a recording, normally what you hear is the sound from the speakers interacting with the room - the room sound variation dominates. What happens when the SQ is sufficiently high is that the recording sound, as it occurred in front of the microphones, or was manipulated in the studio, becomes "louder" - the message it conveys masks what the room modified version is - and the latter disappears from your awareness, as in the live example above.

 

Frank, I will make it worse for myself (no credits left already) by stating that your claim is not even the thing what is happening; It would be correct that at some stage the one is going to be profound over the other and thus the other now is masked, but :

 

There should be much more emphasis on how the quality of the reproduction controls the music waves in space (they become tighter, more aimed, more condensed for their own frequency). I mean, it is so super easy to demonstrate how a room full with standing waves - low frequency but higher frequency just the same) can be turned into a room without any audible standing waves (mind the audible please because this is the masking part) but which also can easily be measured for the energy (walk around with microphone and FFT on-screen for frequencies you like).

 

I will keep it at this, so you can all have a good chuckle. If you only know that this is my means of measurement and as soon as even one corner of choice shows standing waves behavior (like profound lower frequency) I know I tweaked something for the wrong.

And hey, I don't only play Get Back to people over here, I also let them find tanding waves spots, just for gags. Nobody is capable. And obviously I wouldn't dare to write about it if it weren't true, right ?

Room is untreated but takes into account audio measures. Point is : if I leave out the SQ tweaks regarding this, it buzzes all over, requires bass traps and PEQ's on top of it. I was there ... (same room).