Diferent Bits

O

omnipotent

New member
Now, I'm not sure if everyone has heard about it already, but I'm going to say it any way (i was exited to see it.)

http://www.steinberg.net/products/ps/cubase/pc/vst32/index.phtml?id=03010103&sid=05711114

Now think of the possibilities; move over 24 bit processing. No hardware (that i know of) supports 32bit yet, but i belive some software already does and more is sure to come.

Cubase is one of the first companies to alow processing at up to 32bit, which i belive is as detailed as the human ear can hear; this then means you could (in theory) process your music as many times as you liked, without and loss in sound quality.
 
In fact, 32 bit processing is far above what the human ear can hear. 0db is the weakest sound we can hear, and pain sets in at about 120db. That translates to 20 bits, so even with 24 bits processing you have 16 times the maximum detail of what a human can hear.

And, since it all ends up on 16 bit CD's anyway, all that detail is gone, gone, gone.

I can definitely see the point in 24 bits internal processing. I can definitely see the point in mixing to 24 bits (so you have a lower noise floor when compressing during the mastering). For most, 24 bit recording is overkill, but sure, it must be nice to have that extra margin sometimes. I do have a hard time seeing any case where 32 bit recording isn't total overkill, though.
 
Before you all get too excited....

I think you need to read a little bit deeper into this.

http://www.digido.com has some stuff about "supposed" 32 bit processing. Smoke and mirrors!

Basically, it comes down to this. 32 bit FLOATING POINT processing (pretty much the way Windows applications work...:) ) equals 24 bit FIXED processing.

Manufactures like to throw around big numbers, and the 32 bit thing is the latest marketing scam they are using.

Many, many, many plug in's, editors, mixers, etc...for PC use 32 bit Floating Point processing, and some of the ones that only claim 24 bit are 24 bit Fixed.

When we start talking about these 32 bit files that can be recorded, this don't make sense. The resolution is still only going to be at the very best whatever the highest resolution of your A/D/A converters are. If you are using a 24 bit soundcard (which most only have at best 21 or 22 actual bits of resolution, IF that....) you can save that file to some weird scheme made up by the manufacture, but the reality is that at best, it will be 24 bits in real life.

I have seen no 32 bit Fixed processing in computer app's or plugs yet. I HAVE seen double sampling rate processing, which is supposed to be like doubling the Fixed bits, but who can be sure.

About the only way you can get higher internal bit processing is to have a DSP chip integreted in the computer that provides this kind of power. Sonic Solutions has this, and I believe the newer ProTools has this, and I would have to check the info I have, but I believe Sadie provides this too. All of these systems are not only software, but a hardware box that provides powerfull real time DSP that a computer cannot provide because of the current restriction of 32 bit floating point internal processing. The other way to get more Fixed DSP is to have digital processors, external types, like Weiss digital eq's and compressors, or even the hardware version of the Waves L1 Ultramaximizer that have chip sets in them that provide for this. I think I seen where the L1 is 48 bit fixed! It would just plain offer better processing with this kind of power compared to it's PC plugin counterpart that is restricted to the equivelent of 24 bit fixed processing. There are digital processors that reputed to have as high as 72 bit fixed processing, and their price shows for it! :)

Anyway. I suppose that this is "improvement" for Cubase, but it is not exactly what they lead you to believe it is. Did you know that you can save a 32 bit file in Goldwave? (www.goldwave.com). This editor has been around for a while....:) Cool Edit Pro has a 32 bit mode thingy that appears to be the model Cubase is copying, or at least trying something similar.

Anyway, with any software/hardware that is digital, it is important to know what it's Fixed processing power is. A lot of stuff out their right now IS 24 bit Fixed, and the stuff that is 32 Floating Point isn't a bit better (no excuse for the "bit" of pun here....:) )

Sorry to rain on the picnic. Just wanted to clear up some finer points. Really though, you will have a very difficult time finding facts about REAL processing power of most software. With hardware, usually you can find that little "floating point" thing, but the reality is that it's actual DSP is no better then the bit depth of the A/D/A converters. This is much like when Panasonic came out with their first digital mixer, they claimed that it was 24 bit! In reality, it only provided 24 bit internal processing, and it would still only accept a 16 bit word. So it was a 16 bit capatible mixer, not a 24 bit. Imagine the surprise of some of the owners of that console when they tried to feed a 24 bit bit stream to that console and found that it didn't work! :)

Ed
 
Re: Before you all get too excited....

32 bit FLOATING POINT processing (pretty much the way Windows applications work...:) ) equals 24 bit FIXED processing.
Click to expand...

Hmm. Interesting. Floating point processing would give you a non-linear resolution. Ie, the difference in volume between two values near the zero region would be very little, but the difference between two values in the maximum levels would be very big. I don't know how that would affect sound. You would clearly have an EXTREMELY low internal noise floor (which is basically useless to have, since the recorded noise is so much bigger anyway), but you would get loose resolution at high levels. Iäm not sure if this is a good or bad thing... :)
 
Don't worry ED our picnic will be just fine...

I'm sure I've heard you say somthing similar to what you said here before. When I saw the site I wasn't in an instant rush to go out and buy Cubase VST/32 PC, I just thought it (with some of the other new features) looked like a big improvement.

But its nice to know theres someone out there putting an end to all the new marketing scams.:)
 
Well....

regebro though has an interesting point about values being non-linear. I suppose in theory this would improve what could be done to the audio. But.......

The point of having smaller divisions at the least significant bits is more or less worthless by my understanding for TRUE 24 bit audio.

If the idea that the upper end, the most significant bits, have bigger divisions, well, that I think could be a bad thing, because that would suggest that fine nueance in the audios amplitude would get "averaged" in some way because there is the big possibility that the audio could fall between values. These "averages" are what cause a very low distortion because the audio is not represented in a linear fashion. It won't sound like distortion per se, just that if there is enough of it, the audio will lose some clarity.

Anyway....The point is that the improvements in this "supposed" increased dynamic range would happen at the wrong end of the spectrum because at at about 21 or 22 bits, audio has reached what the human ear can possibly detect. This would be evident with reverb trail outs and fade outs.

The true measure of how good DSP works is in how good the code that is written for it is. But no matter how good the code is, the bits of resolution of the Post-DSP'ed file is going to be less then it was Before DSP. How much less is a sign of how good the code is that did the DSP. Good DSP usually will require more internal processing to execute. That is not to say though that a well written code could not be done with less processing, it could just be better thought out.

Anyhow, I still say smoke and mirrors here. When you see 32 bit on a application that runs in Windows, it is floating point, and is usually not a bit better (pun intended) then 24 bits.

Ed
 
Gees

regebro claims there is not and can't hear a different between different cable AND 16, 20, and 24-bit (other than noise floor)? What's next?

I think someone is stuck so much on specifications and it's clogging his ears.

regebro, I'm not trying to attack you. I just find that rather VERY interesting and putting a little humor behind it.
 
Re: Gees

Recording Engineer said:
regebro claims there is not and can't hear a different between different cable AND 16, 20, and 24-bit (other than noise floor)? What's next?

I think someone is stuck so much on specifications and it's clogging his ears.

regebro, I'm not trying to attack you. I just find that rather VERY interesting and putting a little humor behind it.
Click to expand...

Hehe. Well, point taken.
Still, I'm NOT stuck on specifications. If I were, I would think there was a practical difference between 24 and 32 bits. I don't. If YOU can hear the difference between a CD recorded in 24 bits and 32 bits, I can imagine why you can hear the difference between different cables too.
You are then definitely hearing things. :)
 
You must log in or register to reply here.
Back
Top