sure! Which is why we all love the Beatles so much!
It's all about fidelity!
Switching from Cubase to Logic
-
-
You'll find that all up-to-date modern DAWs use 32-bit floating point processing internally. It's a common misconception that DAWs that offer support for 32-bit wav/aiff files have more headroom or a lower noise floor (the files, of themselves, technically do, but not so that the internal processing sees any actual benefit). It's practically impossible to clip a DAW internally, as shown in a series of studies by my old studio neighbour Holger Lagerfeldt. The problems arise when a plugin hasn't been coded optimally, at the D/A stage or of course if you overload the A/D converters while recording to disk. As for more dynamic range, 24-bit already offers a range that exceeds any real-world practical limits, from the point where your ears start to bleed to where sound will fall below ambient background noise levels. Of course, we all want better and more pristine audio, but we hit the point of vastly diminishing returns some time ago.
Thank you for Clarifying because when I read the Cubase Manual back in the day, they made it look like it was something exclusive to Cubase/Nuendo. I also have to agree about hitting the point of diminishing return some time ago.
-
The problems arise when a plugin hasn't been coded optimally, at the D/A stage or of course if you overload the A/D converters while recording to disk.
One other scenario springs to mind, Sammy:
If you allow a plugin's output to enter the next in a processing chain (channel inserts) at too hot a level, the latter's input could clip.
EDIT: This might only be the case if, as Sam said, a "plugin hasn't been coded optimally".
-
One other scenario springs to mind, Sammy:
If you allow a plugin's output to enter the next in a processing chain (channel inserts) at too hot a level, the latter's input could clip.My understanding is if the file is 32 bit wave, you can't clip it anywhere except in the physical input ( AD conversion)
-
I've heard of it happening, Dean, but it could be down to what Sam referred to when he stated that "...problems arise when a plugin hasn't been coded optimally".
I'll add the caveat to my post now. Thanks bro'.
-
I've heard of it happening, Dean, but it could be down to what Sam referred to when he stated that "...problems arise when a plugin hasn't been coded optimally".
I'll add the caveat to my post now. Thanks bro'.
Indeed, Nicky. One of the tests that Holger ran was exactly this one; he purposefully stacked the stock plugins in Logic and overloaded the input of the next in the chain, to see if they could be clipped. He found that as long as the output of the last plugin in the chain was reduced back to 'safe' levels (or the master fader was pulled down to achieve the same), the audio was fine. However, not all plugins behave that way. I believe there were a few freebies that exhibited the expected behaviour of doing nasty things to the audio when the gain-staging was set up to clip the plugins input.
Of course, analogue emulations can be programmed to exhibit some clipping artefacts as they try to mimic their real-world counterparts with varying degrees of success... -
Indeed, Nicky. One of the tests that Holger ran was exactly this one; he purposefully stacked the stock plugins in Logic and overloaded the input of the next in the chain, to see if they could be clipped. He found that as long as the output of the last plugin in the chain was reduced back to 'safe' levels (or the master fader was pulled down to achieve the same), the audio was fine. However, not all plugins behave that way. I believe there were a few freebies that exhibited the expected behaviour of doing nasty things to the audio when the gain-staging was set up to clip the plugins input.Of course, analogue emulations can be programmed to exhibit some clipping artefacts as they try to mimic their real-world counterparts with varying degrees of success...
Also, as soon as you start putting "analogue modellin plugins" all over your +10dbfs tracks you're WAY outside the intended working area of those plugins. Keep in mind that in the old days, the equipment was calibrated differently. To hit the "old" zero, you should probably shoot for -18dbfs on the daw meters... iirc...
-
Exactly, Michael. Slate preaches this often when suggesting how to set up the VMS in that huge BeerGutz thread. I think he recommends -10dB...
Indeed, Nicky. One of the tests that Holger ran was exactly this one; he purposefully stacked the stock plugins in Logic and overloaded the input of the next in the chain, to see if they could be clipped. He found that as long as the output of the last plugin in the chain was reduced back to 'safe' levels (or the master fader was pulled down to achieve the same), the audio was fine. However, not all plugins behave that way. I believe there were a few freebies that exhibited the expected behaviour of doing nasty things to the audio when the gain-staging was set up to clip the plugins input.
Of course, analogue emulations can be programmed to exhibit some clipping artefacts as they try to mimic their real-world counterparts with varying degrees of success...Thank you, Sam. I remember where I read it now. Folks did the same thing with Digital Performer, which at the time claimed, I think, around 1500 or 1600dB dynamic range.
It behaved the same way you described for Logic, but some 3rd-party plugs clipped the bejeezus out of the signal. The analogue-modelled plugins' tendency to exhibit overdriven characteristics is obviously not what we're talking about here, for any who may be a little confused. We're talkin' about serious, soul-destroying digital clipping - nasty stuff.
-
Also, as soon as you start putting "analogue modellin plugins" all over your +10dbfs tracks you're WAY outside the intended working area of those plugins. Keep in mind that in the old days, the equipment was calibrated differently. To hit the "old" zero, you should probably shoot for -18dbfs on the daw meters... iirc...
i usually don't participate in those discussion but let me just add something from my private point of view: when you model an old analog component and your desire is to copy it perfectly you will have to model the input stage in a way that it overloads when the input signal is to loud. or at least it should do _something_ so others can rave on the "character and coloring".
many plug-ins simply have sweet spots and and they are designed to operate in a certain range. what you always should do is level the input/output stage in a way that a plug-in doesn't change the volume of the channel in a drastic fashion. you can do easily by hitting bypass and adjust the output so the levels stay the same. that's good for other reasons as well, for instance you can compare how the sound changes much easier when the level stays the same. if a recording has not enough volume, boost the volume with a Gain plugin before you start processing. otherwise the plug-ins don't operate at the sweet spot.the important part about floating point is the mixer, not the channel. and that's what all the advertising is about. basically it is an attempt to crush fixed point solutions.
from a today's view this is not that exciting anymore but back in the early 2k years many digital mixers were 16bit (yamaha 02r, older protools systems etc) and 24 bits (newer DSP based Protools systems) were more or less state of the art for hardware solutions. and of course the amount of channels make a difference. if you use 30 tracks, you don't have to worry about your headroom - even with a 24bit system. if you you run 150, things get more interesting. so keep you channels at e.g. -16dB when you start mixing and you'll end up with a better mix. because even in a floating point environment, to hot (sum up channels which are to close to 0dB) means it will clip. -
the important part about floating point is the mixer, not the channel. and that's what all the advertising is about. basically it is an attempt to crush fixed point solutions.
Indeed, which is why every major DAW today processes internally with 32-bit floating point, as you hinted at ("[in] today's view this is not that exciting anymore"). Also good advice with starting a mix with channels peaking at -16dBFS (whichever point I start mixing a track from, be it vocal, bass, drums or guitar, depending on the song, I always reach a stage where I have my drum buss peaking around -12dBFS and balance everything else accordingly). As I wrote earlier, Holger Lagerfeldt did some tests to see if he could clip Logic and Pro Tools internally, but I'm not sure how many tracks or how big the test projects/sessions were, so your statement is very interesting to me. Think I might open a session here and experiment, hehe...
