I didn't try to look closely enough at this thread but...
...so called 24bit DSP integer chips like 56000 aren't just "24bits" at all.
56K can do 24x24 multiply and then add it to 56 bit (so 8 bits for an overhead) accumulator.
So what exactly it does mean? Well... it DEPENDS on a code.
And it should (imho) sum it all - it's all code it's something beyond just calling "an intenger" or "a float".
Integer is King? - the challenge
-
- KVRian
- 762 posts since 2 Sep, 2004 from Poland
-
- KVRAF
- 4737 posts since 20 Feb, 2004 from Gothenburg, Sweden
Sorry.Christian Budde wrote:Do you know how ABX tests work?stefancrs wrote:That's not what I was wondering. How many "new takes" did you make? Basically, what I'm wondering is if some of the "new takes" where fakes where the algorithm didn't change. The reason I'm wondering is just because I want to know if there where "blinds" involved in the test or not, or if the algorithm always were swapped for one of the others. Maybe I'm just daft though
Here's an example A = 'Algorithm X', B = 'Algorithm Y', X = ???
You now have to say if X is A or B. If after 15 trials every guess was correct, you can assume that you've heard a difference. What 'blind' shall I put in there? X can only be either A or B.
After I finished the test I swear I had no idea about the result.
Ok, lets just ignore my lousy test and say that it all was coincidence, ok?
I honestly did not know that this was how it worked, and I did not try to cast a shadow or prove your test lousy.
For some reason, I thought the test was like:
Randomly choose one source, then randomly choose again (where you could end up choosing the same as the first), say if you spotted a difference or not. This came from the sentence "11 times out of 15 times I heard a difference", which obviously in ABX tests means "chose the right source" (from A or B when listening to X). And I did not get that.
Again, I apologize. I should've just read up on the whole thing.
Stefan H Singer
https://dropshotaudio.com/
https://dropshotaudio.com/
-
Christian Budde Christian Budde https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=25572
- KVRAF
- Topic Starter
- 1538 posts since 14 May, 2004 from Europe
No problem. I was also wrong two times today. First the DA converter and then the 24bit thing.stefancrs wrote:Again, I apologize. I should've just read up on the whole thing.
But now we all know more and live happily ever after
-
- KVRAF
- 12615 posts since 7 Dec, 2004
nuffink, whyterabbyt, you do not seem to agree with what i stated about electronic audio equipment. what knowledge or experience do you have in this field, what equipment have you worked with?
consumer line level signals are at low levels, generally below 2vpp. this would be 1vp. do you understand the definition of vpp vs. vp ?
in consumer equipment, clipping generally occurs at 1vp, with the nominal level being at 0.778vp. if you'd like, get out your scope/dmm/fg and run a sine into your pc sound card or module. you'll find most likely that the clipping level is as i've mentioned at 1vp on a cheap card, while the 0db level is at 0dbu, 0.778vp.
if you own a higher quality card you should find that the results are different. the clipping level should be at 5vp, while the 0db level should be at 3.9vp. the higher quality card's inputs should have software controlled gain adjustment to +12db (sometimes +15db), and some lowest input level like -100db.
if you own a modular synthesizer (which i assume you do not) you should find that the signal levels you encounter are 10vpp nominal, and 20vpp or more (to 28-30) for clipping.
if you did have any experience working with these different classes of equipment you'd be aware of the differences due to the need for signal level adjustment between different units. a majority of even the low/mid grade performance equipment you should encounter will have signal level switches.
the reason it is important to have very regular voltage ranges in modular synthesizers is twofold. the first reason is that with most of the equipment calibrated to 1v/o, a nominal range of 10vpp will give you ten octaves of range which is acceptable for most uses. (eg. 25hz to 25khz)
the second reason is noise and especially noise from emi. if you've never designed electronic circuits you can not be expected to understand this. i do not expect you to understand the function of all possible noise sources in electronic equipment, or other issues related to the voltage/current ranges used. i will tell you though that a nominal signal level of 10vpp is fairly close to the optimum balanced range for all parameters involved.
one very important fact to take note of is that we use voltage controlled equipment, not current controlled. this means that different factors are in play. inside circuits like dacs for example there are conversions between current and voltage. you should study documents on voltage and current noise, emi, bandwidth, thermal noise and other issues to begin to understand.
consumer line level signals are at low levels, generally below 2vpp. this would be 1vp. do you understand the definition of vpp vs. vp ?
in consumer equipment, clipping generally occurs at 1vp, with the nominal level being at 0.778vp. if you'd like, get out your scope/dmm/fg and run a sine into your pc sound card or module. you'll find most likely that the clipping level is as i've mentioned at 1vp on a cheap card, while the 0db level is at 0dbu, 0.778vp.
if you own a higher quality card you should find that the results are different. the clipping level should be at 5vp, while the 0db level should be at 3.9vp. the higher quality card's inputs should have software controlled gain adjustment to +12db (sometimes +15db), and some lowest input level like -100db.
if you own a modular synthesizer (which i assume you do not) you should find that the signal levels you encounter are 10vpp nominal, and 20vpp or more (to 28-30) for clipping.
if you did have any experience working with these different classes of equipment you'd be aware of the differences due to the need for signal level adjustment between different units. a majority of even the low/mid grade performance equipment you should encounter will have signal level switches.
the reason it is important to have very regular voltage ranges in modular synthesizers is twofold. the first reason is that with most of the equipment calibrated to 1v/o, a nominal range of 10vpp will give you ten octaves of range which is acceptable for most uses. (eg. 25hz to 25khz)
the second reason is noise and especially noise from emi. if you've never designed electronic circuits you can not be expected to understand this. i do not expect you to understand the function of all possible noise sources in electronic equipment, or other issues related to the voltage/current ranges used. i will tell you though that a nominal signal level of 10vpp is fairly close to the optimum balanced range for all parameters involved.
one very important fact to take note of is that we use voltage controlled equipment, not current controlled. this means that different factors are in play. inside circuits like dacs for example there are conversions between current and voltage. you should study documents on voltage and current noise, emi, bandwidth, thermal noise and other issues to begin to understand.
-
- KVRist
- 214 posts since 29 Dec, 2006
Hey, I was wrong when correcting you -- forgot to count the implied msb...Christian Budde wrote:I was also wrong two times today. First the DA converter and then the 24bit thing.
-
- Beware the Quoth
- 35435 posts since 4 Sep, 2001 from R'lyeh Oceanic Amusement Park and Funfair
try thinking a bit harder about what I responded to. Which wasnt your comment in the first place.aciddose wrote:nuffink, whyterabbyt, you do not seem to agree with what i stated about electronic audio equipment. what knowledge or experience do you have in this field, what equipment have you worked with?
consumer line level signals are at low levels, generally below 2vpp. this would be 1vp. do you understand the definition of vpp vs. vp ?
Since when were 'signals' the voltages gear runs on?
Did you actually read my question? Because it was a set of questions in the first place, not directed at you at all, and unanswered except with some charts about signal levels.
The original post I requested clarification of was
Note that emphasis... 'runs on'. Not 'uses higher signal levels'. Do you understand the difference, and why I was querying one in terms of the other? Are you saying something handling signals of 2Vpp can't be running off a 15V transformer, for example?camsr wrote:Why do you think studio equipment runs o a higher voltage than consumer grade equipment?
And when the response was to a set of signal level tables, did you notice my response to that, asking exactly what a set of signal level tables had to do with the OP's claim regarding 'studio' gear versus 'consumer' gear. You know, when I asked
(btw, see that little '?' thingy at the end of the sentence there.... it means something is a question. As in, Im looking for more information or clarification. So your cock-waving over how much supposed insider knowledge of 'equipment worked with' is just that... 'i say its true so it is true' cockwaving instead of a useful f**king provision of validated information. Try the latter some time, without the former, it'll be a whole new experience for you)me wrote:and at what point does that involve the gear itself running at 'higher voltages'?
An idiot on Set Theory:
"In some cases there is an object called red that contains everything that is red. In much the same way a pot is a plate."
"In some cases there is an object called red that contains everything that is red. In much the same way a pot is a plate."
-
- KVRian
- 770 posts since 2 Apr, 2003
Yes, that's what a low pass filter does.living sounds wrote:Apply a low pass filter and see how the envelope of the sample that went through the plugin has lost contour, definition. It looks blurred. It certainly sounds blurred as well.
Your reasoning has more holes in it than Swiss Cheese... and doesn't taste as good with some bacon in a croissant.
-
- KVRAF
- 12615 posts since 7 Dec, 2004
"Since when were 'signals' the voltages gear runs on?"
the clipping level is determined mostly by the supply voltage levels. gear with a supply of 15v for example has a clipping range of about 25vpp if you're lucky (12.5vp) the transistors used in a standard opamp die generally have a 40v or less c-e breakdown voltage meaning that for most practical applications the voltage range must be limited to 30v, which happens to be 15*2. with 15v supplies, many opamp circuits will clip at 12.5vp, so the standard clipping level in high grade audio equipment is set to 10vpp. half of that is the usual nominal signal level giving 6db of headroom. analog electronics are similar to integer processing in that the nominal levels are adjusted based upon requirements for headroom. inside the circuits it is common to see 18db of headroom rather than just 6db. 18db is 1/8 or 5/8=0.625 which is the "line level" transmission standard. it is fairly variable between pieces of equipment exactly how much headroom is allowed for, this is why most equipment gives you that nifty volume pot on the output. usual configuration is center = 0.625, 100% = 5.0 and the pot generally is exponential taper. when i find equipment which does not obey this it is frustrating.
when talking about equipment which uses lower nominal/clipping levels i'm talking not just about the whole circuit, but also about the inputs. the circuit may be capable of driving a full 10vpp signal, however it isnt very useful if the input and internal configuration will clip below this. when i said "high grade" equipment, i meant equipment that accepts 10vpp signals without clipping and can also drive these signals.
http://xhip.cjb.net/temp/public/nobach4u.wav.mp3
the clipping level is determined mostly by the supply voltage levels. gear with a supply of 15v for example has a clipping range of about 25vpp if you're lucky (12.5vp) the transistors used in a standard opamp die generally have a 40v or less c-e breakdown voltage meaning that for most practical applications the voltage range must be limited to 30v, which happens to be 15*2. with 15v supplies, many opamp circuits will clip at 12.5vp, so the standard clipping level in high grade audio equipment is set to 10vpp. half of that is the usual nominal signal level giving 6db of headroom. analog electronics are similar to integer processing in that the nominal levels are adjusted based upon requirements for headroom. inside the circuits it is common to see 18db of headroom rather than just 6db. 18db is 1/8 or 5/8=0.625 which is the "line level" transmission standard. it is fairly variable between pieces of equipment exactly how much headroom is allowed for, this is why most equipment gives you that nifty volume pot on the output. usual configuration is center = 0.625, 100% = 5.0 and the pot generally is exponential taper. when i find equipment which does not obey this it is frustrating.
when talking about equipment which uses lower nominal/clipping levels i'm talking not just about the whole circuit, but also about the inputs. the circuit may be capable of driving a full 10vpp signal, however it isnt very useful if the input and internal configuration will clip below this. when i said "high grade" equipment, i meant equipment that accepts 10vpp signals without clipping and can also drive these signals.
http://xhip.cjb.net/temp/public/nobach4u.wav.mp3
Last edited by aciddose on Wed Jun 06, 2007 11:24 am, edited 1 time in total.
-
- KVRAF
- 6519 posts since 13 Mar, 2002 from UK
I can't speak for rabbyt, obviously, but personally I built the Large Hadron Conductor and the Internet from a Heathkit in 1972.aciddose wrote:nuffink, whyterabbyt, you do not seem to agree with what i stated about electronic audio equipment. what knowledge or experience do you have in this field, what equipment have you worked with?
That's the thing about the internet we can all pretend to be experts, though in your case you might need a bit more practice. For instance if I were to redo that feeble attempt at patronisation you tried I'd just jump over to the ESP page on noise... http://sound.westhost.com/noise.htm ...and paraphrase that. Rod Elliot does a much better job than you, probably because he knows what he's talking about. Plus he isn't attempting to tie it in to some spurious rubbish about "high grade, mid grade and consumer grade equipment" (how comes you missed Lou Grade?) like you are.
Which, of course, was what made me laugh at the time.
Now with improved MIDI jitter!
-
- Beware the Quoth
- 35435 posts since 4 Sep, 2001 from R'lyeh Oceanic Amusement Park and Funfair
doesnt really answer my question. thats analogous to saying 'a car which uses high performance fuel can go faster' when someone queries a claim that 'only sports cars use high performance fuel'. Yes, sports cars are indeed generally designed to go faster; but that's not relevant to the claim. A moped could be using high performance fuel.aciddose wrote:"Since when were 'signals' the voltages gear runs on?"
the clipping level is determined mostly by the supply voltage levels.
Like I said, are you saying something designed to be handling signals of 2Vpp can't be running off a 15V transformer?
An idiot on Set Theory:
"In some cases there is an object called red that contains everything that is red. In much the same way a pot is a plate."
"In some cases there is an object called red that contains everything that is red. In much the same way a pot is a plate."
-
- KVRAF
- 1940 posts since 16 Aug, 2004 from Vienna, Austria
The hardwareness can be measured - it's the additional distortion coming from noisy output amplifier stages in the devices, cable, and A/D converter when it reaches the computer. Hardwareness has a name, and it's Hissnuffink wrote:I suspect he could hear the essential hardwareness of it.
"Until you spread your wings, you'll have no idea how far you can walk." 
-
- KVRAF
- 2208 posts since 13 May, 2005
Once again someone hasn't even bothered to understand the post. I was talking of the lowpassed signal that went through the software reverb vs. the one that passed through the hardware.
JonHodgson wrote:Yes, that's what a low pass filter does.living sounds wrote:Apply a low pass filter and see how the envelope of the sample that went through the plugin has lost contour, definition. It looks blurred. It certainly sounds blurred as well.
Your reasoning has more holes in it than Swiss Cheese... and doesn't taste as good with some bacon in a croissant.
-
- KVRian
- 770 posts since 2 Apr, 2003
In another post you claim that convolution smears transients, seemingly as if this is some global rule of convolution... is this your claim?living sounds wrote:Once again someone hasn't even bothered to understand the post. I was talking of the lowpassed signal that went through the software reverb vs. the one that passed through the hardware.
JonHodgson wrote:Yes, that's what a low pass filter does.living sounds wrote:Apply a low pass filter and see how the envelope of the sample that went through the plugin has lost contour, definition. It looks blurred. It certainly sounds blurred as well.
Your reasoning has more holes in it than Swiss Cheese... and doesn't taste as good with some bacon in a croissant.
We're dealing with two processors performing mathematical calculations on a sequence of numbers, there is NO magical difference between hardware and software units in this respect, in fact if desired you can generate BIT EXACT equality between the two (This is not theory or conjecture, it is fact, I've done it when working with hardware DSPs)
[edited because I mistyped equality]
-
- KVRAF
- 3080 posts since 17 Apr, 2005 from S.E. TN
There is a lot of good-performing equipment nowadays that uses surprisingly low signal levels.
I'm not trying to defend whatever aciddose's premise is, but in old 'classic' studios, the aim of 'line level' signals was to have very large voltages driving very low impedances. It was pretty expensive to build gadgets like that, and is one reason a studio EQ or Compressor costed a lot more than a home-stereo component which was happiest running at about 0.5 V p-p.
Back then, if I would drag in some 'consumer' or 'musician's stage' FX box into a 'real studio' it was a pain in the arse, because for one thing patching into their patch bay, the levels from the studio would just swamp my 'non pro' device's inputs, and on the back end, it was hard to add enough makeup gain going back into the inserts, to get good level out of my cheapie FX box. Usually entailing either raising a return knob on the console so high as to get awful gain-staging hiss, or just having to patch the stupid box thru direct boxes into a mic/line main channel in the board (which would typically also sound like crap).
There are folks who still think you need to be using levels like that before it is 'pro'. So aciddose isn't speaking complete nonsense in the terminology. If one is careful, nowadays one can get very good noise performance without the old-style 'pro' high voltages and low impedances, but I don't think the terminology has completely gone away.
Another place the low impedance and high signal levels are actually a necessity, is for instance if you have a radio or TV studio close to the transmitter. Nowadays the transmitters are frequently remote from the studios, but in the past that wasn't the case. If you have a little studio shack sitting in the shadow of a 100,000 watt transmitter, you dam well BETTER have really hot line levels, and REALLY good shielding (GRIN).
====
On the 'analogness' issue, beneficial 'pleasant' distortion is mentioned. Perhaps a couple of other cues--
I tried to find a link on this, but can't. Working from memory, long ago Bob Carver claimed that if you add just a leetle bit of highest-octave hiss to a stereo feed, most listeners like it better. Stereophiles report that is has a 'more open soundstage' and similar gobbledygook.
Mastering engineer Glenn Meadows, reported taking loops from lo-freq turntable rumble and tracking noise, and mixing it (at a very low level) with squeaky-clean digital recordings.
At the dawn of CD, a lot of his high-end recording star customers complained that CD's are too 'sterile'. Glenn claimed that mixing in a little bit of turntable rumble made em happy with the sound. "Yeah, thats what its sposed to sound like." Dunno if he actually did that to any masters. I think it was just a little psycho-acoustic experiment on his customers.
Turntable rumble/low freq tracking noise, is pretty random stuff, and its not necessarily correlated between the two channels (as I recall Glenn explaining it). I haven't ever analyzed turntable rumble, so he may have been wrong as far as I know.
Anyway, if somebody wants to make a magic integer plugin, maybe they should also add a little high-freq hiss and a little uncorrelated rumble in there. Put em on a 'sounds better' button.
I'm not trying to defend whatever aciddose's premise is, but in old 'classic' studios, the aim of 'line level' signals was to have very large voltages driving very low impedances. It was pretty expensive to build gadgets like that, and is one reason a studio EQ or Compressor costed a lot more than a home-stereo component which was happiest running at about 0.5 V p-p.
Back then, if I would drag in some 'consumer' or 'musician's stage' FX box into a 'real studio' it was a pain in the arse, because for one thing patching into their patch bay, the levels from the studio would just swamp my 'non pro' device's inputs, and on the back end, it was hard to add enough makeup gain going back into the inserts, to get good level out of my cheapie FX box. Usually entailing either raising a return knob on the console so high as to get awful gain-staging hiss, or just having to patch the stupid box thru direct boxes into a mic/line main channel in the board (which would typically also sound like crap).
There are folks who still think you need to be using levels like that before it is 'pro'. So aciddose isn't speaking complete nonsense in the terminology. If one is careful, nowadays one can get very good noise performance without the old-style 'pro' high voltages and low impedances, but I don't think the terminology has completely gone away.
Another place the low impedance and high signal levels are actually a necessity, is for instance if you have a radio or TV studio close to the transmitter. Nowadays the transmitters are frequently remote from the studios, but in the past that wasn't the case. If you have a little studio shack sitting in the shadow of a 100,000 watt transmitter, you dam well BETTER have really hot line levels, and REALLY good shielding (GRIN).
====
On the 'analogness' issue, beneficial 'pleasant' distortion is mentioned. Perhaps a couple of other cues--
I tried to find a link on this, but can't. Working from memory, long ago Bob Carver claimed that if you add just a leetle bit of highest-octave hiss to a stereo feed, most listeners like it better. Stereophiles report that is has a 'more open soundstage' and similar gobbledygook.
Mastering engineer Glenn Meadows, reported taking loops from lo-freq turntable rumble and tracking noise, and mixing it (at a very low level) with squeaky-clean digital recordings.
At the dawn of CD, a lot of his high-end recording star customers complained that CD's are too 'sterile'. Glenn claimed that mixing in a little bit of turntable rumble made em happy with the sound. "Yeah, thats what its sposed to sound like." Dunno if he actually did that to any masters. I think it was just a little psycho-acoustic experiment on his customers.
Turntable rumble/low freq tracking noise, is pretty random stuff, and its not necessarily correlated between the two channels (as I recall Glenn explaining it). I haven't ever analyzed turntable rumble, so he may have been wrong as far as I know.
Anyway, if somebody wants to make a magic integer plugin, maybe they should also add a little high-freq hiss and a little uncorrelated rumble in there. Put em on a 'sounds better' button.
-
- KVRAF
- 2208 posts since 13 May, 2005
Early digital was suffering most from aliasing and jitter. If you add sufficient noise you'd probably mask those artefacts... 
But rumble?
I think there's nothing wrong with digital recording (provided you use great converters). IMO problems occur in the processing area...
But rumble? I think there's nothing wrong with digital recording (provided you use great converters). IMO problems occur in the processing area...
JCJR wrote:
Mastering engineer Glenn Meadows, reported taking loops from lo-freq turntable rumble and tracking noise, and mixing it (at a very low level) with squeaky-clean digital recordings.
At the dawn of CD, a lot of his high-end recording star customers complained that CD's are too 'sterile'. Glenn claimed that mixing in a little bit of turntable rumble made em happy with the sound. "Yeah, thats what its sposed to sound like." Dunno if he actually did that to any masters. I think it was just a little psycho-acoustic experiment on his customers.
Turntable rumble/low freq tracking noise, is pretty random stuff, and its not necessarily correlated between the two channels (as I recall Glenn explaining it). I haven't ever analyzed turntable rumble, so he may have been wrong as far as I know.
Anyway, if somebody wants to make a magic integer plugin, maybe they should also add a little high-freq hiss and a little uncorrelated rumble in there. Put em on a 'sounds better' button.