Login / Register
 
AD/DA converters: lecture on what actually matters and how they affect the signal
New Reply
Subscribe
kraku
Thread Starter
#1
15th July 2013
Old 15th July 2013
  #1
Lives for gear
 
kraku's Avatar
 
Joined: Jun 2008
Location: Finland
Posts: 1,801
My Recordings/Credits

Thread Starter
kraku is offline
AD/DA converters: lecture on what actually matters and how they affect the signal

Here's an excellent explanation about digital audio converters and how they affect the signal, what actually matters and how.

Highly recommended for everyone since there's always lots of debate and argument over the audio converters on these forums.




It's also interesting to know that the old home tapes had something like 5-6 bit quality and the best studio reels in perfect conditions around 12-13 bit quality

It's also nice to know that dithering really doesn't matter with current bitdepts.
Quote
9
#2
15th July 2013
Old 15th July 2013
  #2
Lives for gear
 
Acid Mitch's Avatar
 
Joined: Aug 2010
Location: Space is the place
Posts: 2,954

Acid Mitch is offline
Quote
2
#3
15th July 2013
Old 15th July 2013
  #3
Gear addict
 
Joined: May 2011
Location: London
Posts: 439

Risc_Terilia is offline
That was really interesting thanks.
Quote
1
#4
15th July 2013
Old 15th July 2013
  #4
Lives for gear
 
Daisuk's Avatar
 
Joined: Oct 2012
Location: Oslo
Posts: 1,031
My Recordings/Credits

Daisuk is offline
I'm not even going to pretend that I understood all of that, but basically, regarding the first part - what he is saying is that D/A converters have no real effect? Or did I miss something? Hehe. :p
__________________
MarsMelons - netlabel for experimental music

esmone - my drawings and cover art
#5
15th July 2013
Old 15th July 2013
  #5
Lives for gear
 
mainesthai's Avatar
 
Joined: Jan 2013
Location: Armenia
Posts: 533

mainesthai is offline
__________________
Bad: If it measures better it must sound better.
Equally bad: Coitus science I have my ears.

All complex questions have simple answers.
They are all wrong.
Quote
3
#6
15th July 2013
Old 15th July 2013
  #6
Lives for gear
 
grumphh's Avatar
 
Joined: Dec 2009
Location: Denmark
Posts: 4,035

grumphh is offline
Quote:
Originally Posted by Acid Mitch View Post
The video has been posted on GS a few times.

D/A and A/D Wow!!

D/A and A/D | Digital Show and Tell (Monty Montgomery @ xiph.org)

http://www.gearslutz.com/board/newbi...gear-hype.html

I'm pretty sure it was posted in the EMI & EMP section also but I can't find a link.
The only thing i have seen about this is an article (presumably by the same guy) that was linked to.

So for me the video was a welcome addition to this forum - as i don't frequent any of the other fora here...



And considering the amount of anti digital stuff i have read here i do find it hysterically funny that he does recreate perfect sines with an old consumer grade soundcard
__________________
.

Gear list
Synth sounds



...Marimba trills!
#7
15th July 2013
Old 15th July 2013
  #7
Banned
 
Joined: Jun 2006
Posts: 1,695

3phase is offline
Quote:
Originally Posted by kraku View Post
It's also interesting to know that the old home tapes had something like 5-6 bit quality and the best studio reels in perfect conditions around 12-13 bit quality

It's also nice to know that dithering really doesn't matter with current bitdepts.
Since all of theese statemets are clearly wrong the usefulness of this video for audio producers education is questionable..its good for schoolkids thou since its explains some basics of digital audio but as we can see it provokes the wrong conclusions. like that a real tape would sound inferior than a 16 bit recording.. The guy hasnt used the word sound but used the word quality in polemik ways.. a 6 bit audio recording quality is dreadfull..that has nothing to with the audio quality of a tapedeck.

noisefloor is not defining sound quality and quantisation noise is not the same thing as tape noise.

correct conclusion is that a 16 bit recording has less noise than a wheeltape.. but thats it !

same fals conclusion about dithering..

the correct conclusion out of that waht is said in the video is that you dont need to dither for getting 1 k sine tones on a cd..

but that dont says that dither dont matters when getting 32 or 24 bit music recordings on a cd.. only deff or very unexperienced peole that never did cd mastering in theire live would say that.

So this video is no contradiction to the general knowledge and practice of musicproduction and i wonder why peopel allways try to call high end audio a myth. Where are the benefits of declaring that no special measures are necessary to produce fine audio and that any cheap AD/DA will do?

Just because the highend converters are still so expensive dont makes them a bad thing. And one can only wish that they become the standard and not wiped from the market because some pseudo scientist declare all theire benefits as inaudible myth.
Quote
2
#8
15th July 2013
Old 15th July 2013
  #8
Banned
 
Joined: Jun 2006
Posts: 1,695

3phase is offline
Quote:
Originally Posted by grumphh View Post
i do find it hysterically funny that he does recreate perfect sines with an old consumer grade soundcard
? may i facepalm? digital audio is a standard since the early 80´s and was used to distribute recordings of symphonie orchestras from the start

Can you show me anybody that ever questioned that you can produce a sinetone with digital audio?`

remember the dx 7? that was all sine tones..even modulating each other..

sorry for you when this causes a nervous collaps for you..but the finest sine tones of a digital synth would not make them better sounding than the not so perfect sinetone of a minimoog..


thats all no contradiction since sound quality is not measured in terms of noisefloor and frequency response. Or mathematical perfect waveforms. These technical data only describe single aspects of a the technical quality of the signal, but dont tell you how good something sounds .. Actually only your ear really can tell you how "good" something sounds. Technical measurements are good to trace problems and ensure transparent operation..since not everything that sounds good is a transparent process.

When you only look at freq response or friendly measured noisefloor of unity gain situations on a single channel a behringer desk can have better technical data here than a big broadcast console.. But have you ever heard anybody claiming that a behringer desk will do as good as a telefunken valve console? even the biggest troll wouldnt dare to do such nonsense claims
Quote
3
#9
15th July 2013
Old 15th July 2013
  #9
Lives for gear
 
kirkelein's Avatar
 
Joined: Mar 2012
Location: Stockholm
Posts: 925
My Studio

kirkelein is offline
Interesting video. He almost seems to make the claim that higher bit depths are unimportant, as well as sampling rate. I don't think that's his point though, but he could have been clearer on that. Perhaps a disclaimer would have been in order.

All I know is that tape decks sound wonderful and no digital bit depth replicates that. All I know is there is audible difference between my 96/24 and 44.1/16 mixdowns.

Interesting video though, thanks. I finally understand how the basic process works, at least.
kraku
Thread Starter
#10
15th July 2013
Old 15th July 2013
  #10
Lives for gear
 
kraku's Avatar
 
Joined: Jun 2008
Location: Finland
Posts: 1,801
My Recordings/Credits

Thread Starter
kraku is offline
Quote:
Originally Posted by 3phase View Post
? may i facepalm? digital audio is a standard since the early 80´s and was used to distribute recordings of symphonie orchestras from the start

Can you show me anybody that ever questioned that you can produce a sinetone with digital audio?
You might be missing the point here. In the video the guy reproduces a perfect 20kHz (at the brink of human hearing) sine wave with 44.1kHz sample rate. This is something most people on these forums say is impossible without having at least 96kHz or 192kHz sample rate. This begs for the question if the higher sample rates are really that useful.

The video wasn't about which processing method is most pleasing to the ear.
The video demonstrated very clearly that the converters are no where as limited as many people like to think. They are really transparent and capable of excellent accuracy.

The video busted some myths and explained the pros and cons of digital audio. It explained where the quality goes when you change bit depth etc. It also demonstrated you can reproduce audible range signals accurately with fairly old and cheap audio interface which is running at 44.1kHz.
Quote
3
#11
15th July 2013
Old 15th July 2013
  #11
Banned
 
Joined: Jun 2006
Posts: 1,695

3phase is offline
Quote:
Originally Posted by kraku View Post
You might be missing the point here. In the video the guy reproduces a perfect 20kHz (at the brink of human hearing) sine wave with 44.1kHz sample rate. This is something most people on these forums say is impossible without having at least 96kHz or 192kHz sample rate. This begs for the question if the higher sample rates are really that useful.

The video wasn't about which processing method is most pleasing to the ear.
The video demonstrated very clearly that the converters are no where as limited as many people like to think. They are really transparent and capable of excellent accuracy.

The video busted some myths and explained the pros and cons of digital audio. It explained where the quality goes when you change bit depth etc. It also demonstrated you can reproduce audible range signals accurately with fairly old and cheap audio interface which is running at 44.1kHz.
the video dont bust any myths .., when todays producers know crap about digital audio i call that sad. But beeing unliteral in audio technology cant be called a myth..

The guy explains some basic facts but that on a level way below on what a producer should know allready. If that not the case.. ok.. good for you that some gaps got filled but dont do the usual beginners mistake to learn something from a youtube video and straight jump into the the teachers position and start than to spread all wrong conclusions as facts.

Thats actually producing audio myths.

The discussions about stepping in the producer world are more related to automation issues inside daws.. nobody ever questioned that a niquist filter makes perfect sinewaves out of digital data.. The analog/digital discussion was beyond that point in the early 80´s already. Its other issues that came into focus during that ongoing discussion like mirro ferquencys on the samplerate or phase distortion thru the niquist filter that explained the shortcomings of cd audio and a highend digital format was established with 96K24 bit with good reasons.

And it is beneficial especially during production.. the end product can be maybe down scaled to lower specs..by using dither !

but inside a daw higher samplerates and bit depth deffenetly sound better !!

I however agree that 192khz is really not necessary.. while doing it all on 32bit float still brings quality benefits over 24 bit files..and 24 bit sounds clearly better than 16 bit files wether theese are audible on the single track is questionable..but like so often it well might sum up in a multitrack production. I wouldnt question peopel that claim to have benefits with that. While on 192k i rather would think that this is a more helps more attitude and that allready risk to catch low freq radio transmissions in your recordings.
Quote
5
#12
15th July 2013
Old 15th July 2013
  #12
Lives for gear
 
grumphh's Avatar
 
Joined: Dec 2009
Location: Denmark
Posts: 4,035

grumphh is offline
Quote:
Originally Posted by 3phase View Post
the video dont bust any myths ..
Yes it does - on this very forum i have read about the stepped nature of digital quite a few times.
Be it by qualified people or not, the myth does exist.
Quote
1
#13
15th July 2013
Old 15th July 2013
  #13
Banned
 
Joined: Jun 2006
Posts: 1,695

3phase is offline
Quote:
Originally Posted by grumphh View Post
Yes it does - on this very forum i have read about the stepped nature of digital quite a few times.
Be it by qualified people or not, the myth does exist.
digital has a stepped nature..thats a fakt and no myth.. that the niquist filter makes a perfect sine out of that happens during the DA conversion..so the steps are gone when the signal is analog again.. but inside the digital domain it is stepped.. A major problem for all automation tasks since a 16 bit resolution is not fine enough there to dont produce artefacts .

So there is no myth..just wrong interpretation of the facts and i dont think that this video makes that any better as we can see with the conclusions here..

The guy should have spared us with his polemiks that try to defend a digital superiority that is in question since this technology was established.
He should have just stick to the facts of digital audio and not used theese comparison with wheeltape "quality".. Sorry.. i hate propaganda films

You cant realy say that the audio that meets the ears of the consumers has a better sound quality today than in the pre digital times..
So having no noise issue anymore has´nt really helped..has´nt it?
Quote
4
kraku
Thread Starter
#14
15th July 2013
Old 15th July 2013
  #14
Lives for gear
 
kraku's Avatar
 
Joined: Jun 2008
Location: Finland
Posts: 1,801
My Recordings/Credits

Thread Starter
kraku is offline
Quote:
Originally Posted by 3phase View Post
the video dont bust any myths .., when todays producers know crap about digital audio i call that sad. But beeing unliteral in audio technology cant be called a myth..

The guy explains some basic facts but that on a level way below on what a producer should know allready. If that not the case.. ok.. good for you that some gaps got filled but dont do the usual beginners mistake to learn something from a youtube video and straight jump into the the teachers position and start than to spread all wrong conclusions as facts.

Thats actually producing audio myths.
I can openly admit that I wasn't aware that you can reproduce the sine wave accurately which is near nyquist frequency, especially when it's cycles are not perfectly aligned with the sample points (sub sample accuracy near nyquist, etc.). I knew that the DACs were doing some kind of interpolation between the samples, but from what I saw in the video, the interpolation was much more advanced than I had thought. It's able to recreate a perfect sine from a serious looking mess of samples. This probably means that the interpolation method is using several samples as coefficients or something.

Since you seem to know how it's done / how it actually works, would you mind sharing your knowledge on this topic with us?



Quote:
nobody ever questioned that a niquist filter makes perfect sinewaves out of digital data..
I did, under certain conditions. See above.

EDIT:
Now that I read your response again, why did you use the words "nyquist filter" in the context of DAC?

Quote:
Its other issues that came into focus during that ongoing discussion like mirro ferquencys on the samplerate or phase distortion thru the niquist filter that explained the shortcomings of cd audio and a highend digital format was established with 96K24 bit with good reasons.
Mirror frequencies aren't a problem. The analog signals are always band limited before they hit the ADC.

Also phase distortions probably should be overcome using phase linear filters? (My DSP knowledge is very limited but I think this should be the case here.)
#15
15th July 2013
Old 15th July 2013
  #15
Banned
 
Joined: Jun 2006
Posts: 1,695

3phase is offline
Quote:
Originally Posted by kraku View Post
I can openly admit that I wasn't aware that you can reproduce the sine wave accurately which is near nyquist frequency, especially when it's cycles are not perfectly aligned with the sample points (sub sample accuracy near nyquist, etc.). I knew that the DACs were doing some kind of interpolation between the samples, but from what I saw in the video, the interpolation was much more advanced than I had thought. It's able to recreate a perfect sine from a serious looking mess of samples. This probably means that the interpolation method is using several samples as coefficients or something.

Since you seem to know how it's done / how it actually works, would you mind sharing your knowledge on this topic with us?





I did, under certain conditions. See above.

EDIT:
Now that I read your response again, why did you use the words "nyquist filter" in the context of DAC?



Mirror frequencies aren't a problem. The analog signals are always band limited before they hit the ADC.

Also phase distortions probably should be overcome using phase linear filters? (My DSP knowledge is very limited but I think this should be the case here.)
mirror frequencys are no problem? thats good to know..thank you master..

However there are guy that claim that the hottness in the highs of digital audio are a psychoacoustic fx caused by mirror frequencys of the samplerate around 8 khz..

i actually found the guys claiming that more trustworthy than a guy that dont even knows that digital audio can reproduce 20 k sinewaves..

however ther might be other things that go wrong in the DA conversion.. you were wright that the acurate reproduction of waveforms near the niquest frequency is problematic since the theory is based on an ideal filter that doesnt exsist in nature.
therefore you wont have an ideal reproduction at theese high frequencys.. it maybe can cope with a clean sinus wave ..but thats it..

the higher 96 k samplrate was introduced to let this filter work above listening range at 48 k..so any artefacts caused by a less than ideal filter happen above the listening range and not within it.

sounds like a good plan..or?

in nature a pure sine wave is rather rare.
Quote
2
kraku
Thread Starter
#16
15th July 2013
Old 15th July 2013
  #16
Lives for gear
 
kraku's Avatar
 
Joined: Jun 2008
Location: Finland
Posts: 1,801
My Recordings/Credits

Thread Starter
kraku is offline
Not sure if you're serious or not...

Quote:
Originally Posted by 3phase View Post
mirror frequencys are no problem? thats good to know..thank you master..
You're welcome.

A simple question, just to make sure we're in the same universe here:
Can you explain to me in two short sentences what a mirror frequency is? Don't quote / paste URL as your response. Use your own words.


Quote:
i actually found the guys claiming that more trustworthy than a guy that dont even knows that digital audio can reproduce 20 k sinewaves..
No-one claimed such a thing, atleast not in this thread if you read through the responses.


Quote:
however ther might be other things that go wrong in the DA conversion.. you were wright that the acurate reproduction of waveforms near the niquest frequency is problematic since the theory is based on an ideal filter that doesnt exsist in nature.
therefore you wont have an ideal reproduction at theese high frequencys.. it maybe can cope with a clean sinus wave ..but thats it..
I know nothing about ideal filters vs. nature. All I know is that the video demonstrated clearly that even the old DACs use a bit more advanced interpolation of samples and they are able to perfectly recreate the sine wave. This also means that they should be able to recreate any other band limited waveform.


Quote:
in nature a pure sine wave is rather rare.
Depends on how you look at it. You can recreate any signal using sine waves. So you can think of all the sounds you hear as combination of sine waves.
Quote
2
#17
15th July 2013
Old 15th July 2013
  #17
Banned
 
Joined: Jun 2006
Posts: 1,695

3phase is offline
you are a fountain of pure wisdom, so i rest my case since its useless to fight certain tendencies in electronic music production.
Quote
2
#18
16th July 2013
Old 16th July 2013
  #18
Lives for gear
 
antto's Avatar
 
Joined: Dec 2012
Location: 127.0.0.1

antto is offline
i've never seen anyone call it "nyquist filter"
it's true that the goal of this filter is to remove everything above nyquist and pass everything below, which is an ideal filter which doesn't quite exist in the real world

the perfect reconstruction filter is the so-called sinc() interpolation, which is perfect BUT infinitely long, thus it's not practical for realtime DSP
the solution to the infiniteness of the sinc() filter is to "trim" it, further, a window is usually applied
trimming the perfect reconstruction filter makes it less perfect - the filter will no longer be "brick-wall" but will have a smoother slope

your (analog) sine-wave, when filtered with a non-ideal filter, and then sampled, will become quieter when it goes very close to nyquist (due to smooth slope) and then, when it goes a bit above nyquist - will alias (aka bounce back down from nyquist) and continue to be quieter and quieter

sampled signals are sampled signals, the confusion with stairsteps comes because there are many ways to represent them visually
the video did a great job demonstrating that there is no "signal" between the samples
the easiest way to display the signal is to draw straight lines between the samples - stairs
but the proper way is to "reconstruct" the signal, because displaying the digital signal onto the screen (which displays a raster graphics) is basically resampling, yes, resampling

3phase: what he meant with the tapes is that the noise-floor of digital signals can be compared to the noise-floor of tape
that is just a comparison of the level of noise, not about how it sounds like, whether it's nicer or uglier
Quote
2
Oli
#19
16th July 2013
Old 16th July 2013
  #19
Oli
Lives for gear
 
Oli's Avatar
 
Joined: Apr 2009
Location: Perth
Posts: 2,590

Oli is offline
The presentation, and most like it, don't cover all the relevant issues with digital audio, it's conversions, and processing. Don't read too much into it. Worse, as 3phase mentioned, it's easy for people to walk away with false impressions. I have designed cheap converters (not for music), and have dealt with some of these issues.

It's very common for cheap interfaces to have poorly performing filters (at both ends). There are various other pertinent factors involved as well (don't have time to go into them here), but poor filters are often quite noticeable.

There was a handy converter comparison thread here previously (it was pulled), which demonstrated some performance issues with various converters. The thread didn't go into technical explanations, though it did serve as a handy demonstration.

One important aspect of sample rate, is processing of a signal. Unless you just record and playback, anti aliasing filters at your ADC stage are not going to prevent aliasing in your final signal.

btw, a single sin wave is the easiest signal to represent, process, and reconstruct.
Quote
3
#20
16th July 2013
Old 16th July 2013
  #20
Lives for gear
 
JSt0rm's Avatar
 
Joined: Apr 2005
Location: Los Angeles
Posts: 3,788

JSt0rm is offline
#21
16th July 2013
Old 16th July 2013
  #21
Lives for gear
 
antto's Avatar
 
Joined: Dec 2012
Location: 127.0.0.1

antto is offline
AD/DA converters are unaware of whether you feed them a single sine wave or complex signal, they just convert

a single sinewave is good enough for many tests, because it easily shows if there is any distortion, quantization and/or other noise, yet only takes a very narrow portion on the freq analyzer

JSt0rm: who was that directed at, if i may ask?
#22
16th July 2013
Old 16th July 2013
  #22
Banned
 
Joined: Jun 2006
Posts: 1,695

3phase is offline
Quote:
Originally Posted by antto View Post
3phase: what he meant with the tapes is that the noise-floor of digital signals can be compared to the noise-floor of tape
that is just a comparison of the level of noise, not about how it sounds like, whether it's nicer or uglier
i know that he ment that but he didnt say that.. he calls it a 6 bit quality what tape does which suggest that it sounds as bad as a 6 bit signal.. Thats pure polemic propaganda because he didnt done that by accident.. the beard should have been a warning..

Thing with tapedecks is that they actually can sound very good and the nature of tapenoiuse allows musical information below or within the noisefloor..opposite to quantisation noise where is no usefull information below the quantisation level and the noise has a real nasty frequency spectrum. so its total bullshit to compare qualitys there when you cant even compare the absolute noise levels really.

the guy should have focused on the technical facts of reconstruction filtering ( yes i did know that you can call it like that..) instead embedding a manipulative opinion. Especially when its the typical technocratic false opinion that when the theory is perfect all realworld artefacts must be inaudible. the "what you cant measure dont exsists" dogma.. Which is only true when you know what to measure and how to measure it.. And audio quality is complex to measure. level, distortion, phase.. and than there seems to be still another yet undefined psycho acoustic parameter that only shows in the proportion between the other measurements. Since systems with almost identical measurements can sound significantly different.

op amps are a good example..according to the technocrats they should sound within the audiorange all the same.. but ..they do not.

there is a famous german audio engineers word.

Wer misst misst Mist. (who measures measures crap)
#23
16th July 2013
Old 16th July 2013
  #23
Lives for gear
 
lain2097's Avatar
 
Joined: Jun 2009
Location: French quarter of Canada
Posts: 2,101
My Recordings/Credits

lain2097 is offline
My addition to this subject is that yes probably it's not super-critical with modern converters but I did notice a difference going from Echo AF12 to Apogee Rosetta 800. A-D is where the differences lie I find. It just seems to sound all "shiny" and have a nice top end response. Psychosomatic or not...

I mean just look at sheer amount of passives on a Rosetta 800, even just that should be doing something?:


Compare to the sparse Echo AF12, also note it's shared channel design whereas in the Rosetta each in and out is totally separate. I'd wager just that should really reduce the crosstalk. Yes I realise this is like comparing a Civic to a Mercedes but hey the AF12 never stopped me from making a pissed off chiptune album!
__________________
Check out my latest album http://lain2097.bandcamp.com/album/p...real-intention
#24
16th July 2013
Old 16th July 2013
  #24
Lives for gear
 
antto's Avatar
 
Joined: Dec 2012
Location: 127.0.0.1

antto is offline
Quote:
Originally Posted by 3phase
digital has a stepped nature..thats a fakt and no myth.. that the niquist filter makes a perfect sine out of that happens during the DA conversion..so the steps are gone when the signal is analog again.. but inside the digital domain it is stepped.. A major problem for all automation tasks since a 16 bit resolution is not fine enough there to dont produce artefacts .
digital signals are not "stepped", they only have actual values at regular intervals, and there is nothing in the "gaps"

you could call them stepped if they were resampled poorly, or if your DA converter has something like a sample-n-hold instead of a filter

how exactly do you put "automation" and "16 bit resolution" into the context above? because i'm trying hard and i can't make sense of that statement
Quote
4
#25
16th July 2013
Old 16th July 2013
  #25
Lives for gear
 
JSt0rm's Avatar
 
Joined: Apr 2005
Location: Los Angeles
Posts: 3,788

JSt0rm is offline
Quote:
Originally Posted by antto View Post
AD/DA converters are unaware of whether you feed them a single sine wave or complex signal, they just convert

a single sinewave is good enough for many tests, because it easily shows if there is any distortion, quantization and/or other noise, yet only takes a very narrow portion on the freq analyzer

JSt0rm: who was that directed at, if i may ask?
mostly kraku but its for everyone
#26
16th July 2013
Old 16th July 2013
  #26
Lives for gear
 
Joined: Dec 2010
Posts: 7,099

djugel is offline
seems to be a lot of hot-air in that video... kinda just makes me care less about something I could care less about..
#27
16th July 2013
Old 16th July 2013
  #27
Banned
 
Joined: Jun 2006
Posts: 1,695

3phase is offline
Quote:
Originally Posted by antto View Post
digital signals are not "stepped", they only have actual values at regular intervals, and there is nothing in the "gaps"
And what is that supposed to say? which gaps? there are no gaps between the digital data but steps between the values.. there are no digital signals just digital data, one sample value follows the next. but there is missing information inbetween the sample steps that needs to be reconstructed .. its like a list of values and between the values you have steps when you read them out..

There is nothing false about the steped visualisation except that people might get the idea that the analog outcome still has theese steps..

However..the analog outcome still deals with theese steps since the niquist filtering is in reality far from ideal and within the daw we apply all sorts of stepped modulation to the steped values that cause a whole lot of artefacts. zippery noises for example.. So its missleading to give the image that this stepping dont has its problems and thats no issue to be dealt with..Thats based on the superiority myth of digital audio. The most evil of all audio myth.
#28
16th July 2013
Old 16th July 2013
  #28
Lives for gear
 
TheBrightSide's Avatar
 
Joined: Jan 2012
Posts: 3,766

TheBrightSide is online now
I tried watching the video but it made me sleepy.
So to simply matters for me, would you spend a large amount on A/D + D/A, or would you say that it is not worth the expense. (I am considering an upgrade and would appreciate any input)
#29
16th July 2013
Old 16th July 2013
  #29
Lives for gear
 
antto's Avatar
 
Joined: Dec 2012
Location: 127.0.0.1

antto is offline
Quote:
Originally Posted by 3phase
And what is that supposed to say? which gaps? there are no gaps between the digital data but steps between the values.. there are no digital signals just digital data, one sample value follows the next. but there is missing information inbetween the sample steps that needs to be reconstructed .. its like a list of values and between the values you have steps when you read them out..
it means that the a digital signal itself has no "values" between the samples, but only at the samples, there are no steps, it's just values at individual moments in time, there are no "straight lines" connecting the samples, that's something YOU're introducing

Quote:
There is nothing false about the steped visualisation except that people might get the idea that the analog outcome still has theese steps..

However..the analog outcome still deals with theese steps since the niquist filtering is in reality far from ideal and within the daw we apply all sorts of stepped modulation to the steped values that cause a whole lot of artefacts. zippery noises for example.. So its missleading to give the image that this stepping dont has its problems and thats no issue to be dealt with..Thats based on the superiority myth of digital audio. The most evil of all audio myth.
why are you putting daw automation into this discussion at all?
that's totally up to the DAW, the signals are already digital and it's up to whoever designed it to either process all modulation stuff per sample or in some periods, to save cpu (which is not nice but in any case - totally not related to AD/DA converting)

if you're so worried about all of that, simply run your DAW at a higher rate
Quote
1
#30
16th July 2013
Old 16th July 2013
  #30
Lives for gear
 
Joined: Dec 2008
Posts: 1,283

W-W-Int is offline
just a quick chime in here $.02 and all...

there are no steps at d/a conversion. this is a myth. what happens is there are small analog values of constant change between one sample point's playback and the next. that d/a converter output stage doesn't sit there and magically shut off the current or maintain the exact same current of the last sample point it starts to taper downward slightly then ramp up or down to the next value as the output stage processes and outputs the next sample value.

and a/d conversion is a little tricky. how you design the converter and your choices in component selection and matching play a much bigger role in performance as to accuracy than you'd think vs. the bit depth or sample rate.

read a great interview in a crap hi-fi magazine back in the mid 80's with the man who lead the sony side of the design team for the cd standard. he also had previously designed the 8 track 10 bit digital recorder that was built by sony and used on countless classical albums of the day. i'm paraphrasing cause it's been a long time ago but...

he stated that that 10 bit machine had better sound reproduction than almost anything he'd since run across because of the way they designed and built it. they ordered thousands of components for each piece needed and batch tested them to see which were the closest in spec to each other for the pieces they needed. he said they used a cascade test where they would weed out the parts by running them in batches of ten and measuring and swapping out individual pieces and retesting based on the ones that got swapped out that had the least variance. this means they'd test a series of say ten capacitors and swap one out. the one swapped out would go into a grading system where if the next one swapped in caused the ten to have a higher variance than the one just taken out they'd toss out the new one, then move the one just put back and swap the next one down in line and do the comparison till they got the parts that had the least variance with each other and then use those in the final build. his conclusions were that because of that component matching it had a better converter stage than anything else at the time with 18 bit resolution.


and of course that was his subjective opinion but he was the guy responsible for a lot of the high end converters of the day for the serious recording gear in high end studios so his opinion should carry some weight.
New Reply Submit Thread to Facebook Facebook  Submit Thread to Twitter Twitter  Submit Thread to LinkedIn LinkedIn  Submit Thread to Google+ Google+ 
 
Topic:
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump

SEO by vBSEO ©2011, Crawlability, Inc.