Digital Performer audio engine resolution!

[Tritonemusic]

Since the MW Limiter processes audio at 64-bit floating point, what does that mean, exactly, when working in DP?

[Kubi]

Well, if I understand this correctly, seems like the issue here is audio processing - where higher resolution makes a whole lot of sense (just like it makes a lot of sense to work in 24 bits when you want to end up with a 16 bit CD, or to process using 32bit float when you want to print 24bit fixed.) So it doesn't surprise me that folks implement it in summing, or finishing limiters etc. I'm sure it has its merits.

Also, just as an aside - ProTools (non-native) I believe was 48bit fixed internally, at least a few years ago, which is definitely better than 32bit float if you watch your levels. But not sure what it runs at now - and native ProTools is definitely 32bit float, as is almost everything else.

PS: Anyone who claims that 24bits is overkill should maybe re-visit their monitoring environment...?

[James Steele]

Exactly, I think this is why the Ableton software guys were somewhat miffed at 'fixing' their summing engine by using 64 bits. The two "issues" they dealt with from forum/end users were CPU performance and claims that their audio engine summing bus sounded bad.

I wish more developers would have the courage to take a stand and release a statement calmly stating an objective, scientific argument debunking the need for a feature and, as diplomatically as possible, say "Sorry, we're not going to waste valuable programmer hours on this non-issue." This will be the DAW world's equivalent of the $10,000 boutique speaker cable.

[Michael Canavan]

Exactly, I think this is why the Ableton software guys were somewhat miffed at 'fixing' their summing engine by using 64 bits. The two "issues" they dealt with from forum/end users were CPU performance and claims that their audio engine summing bus sounded bad.

I wish more developers would have the courage to take a stand and release a statement calmly stating an objective, scientific argument debunking the need for a feature and, as diplomatically as possible, say "Sorry, we're not going to waste valuable programmer hours on this non-issue." This will be the DAW world's equivalent of the $10,000 boutique speaker cable.

Yeah I get why they did it, and arguably there is some future proofing benefits to it, but it was a concession they made to the masses, who understand just enough about the theory and applications of the math to demand things they don't need.

You see that in this thread, how many of us really understand the science behind, the computational or even audio related reasons for floating point VS fixed? 32 bit VS 64? and really understand the reasons why a developer would choose to use them? Yet we do jump to conclusions about it based on hearsay.

The only things that I can say with certainty based on developers talking about them are that 32 bit internal and summing engines are all that's needed, 64 bit OS and applications lets us access more RAM than we need to, and recoding in 24 bit reduces any artifacts created in the summing and bit reduction process to inaudible levels.

The truth of the matter is these days the tools are good enough, the improvements to the quality of the output are now on the onus of the person behind the computer keyboard.

[James Steele]

The truth of the matter is these days the tools are good enough, the improvements to the quality of the output are now on the onus of the person behind the computer keyboard.

Perfectly said.

[Shooshie]

Also, just as an aside - ProTools (non-native) I believe was 48bit fixed internally, at least a few years ago, which is definitely better than 32bit float if you watch your levels. But not sure what it runs at now - and native ProTools is definitely 32bit float, as is almost everything else.

Many years ago (2000-2001 era) Pro Tools was set at 24 bits resolution. I'm sure it's been upgraded since then, and it would make no sense for it to process internally at less than 32bits float. But is 48 bits fixed actually better than 32 bits float? I'd have to see them side by side, comparing the details of each, before I could make that call. 32 bits float covers a LOT of territory, and it does so with less CPU overhead than 48 bits fixed.

If non-native PT is set to 48 bits fixed, I'd wonder if it has to do with some limitation imposed by the external hardware which won't permit a floating point DSP. Now, I don't keep up with Pro Tools, so I am just saying what concerns I'd want to check into before making statements like that. And after checking into them, I'd be happy with whatever the results are. So, I'm just asking if you've actually checked into that, and if you know that 48 bits fixed gives Pro Tools better resolution than 32 bits float.

Shooshie

[HCMarkus]

The truth of the matter is these days the tools are good enough, the improvements to the quality of the output are now on the onus of the person behind the computer keyboard.

Perfectly said.

I'll pile on with a big +1.

[Klaus]

Also, just as an aside - ProTools (non-native) I believe was 48bit fixed internally, at least a few years ago, which is definitely better than 32bit float if you watch your levels. But not sure what it runs at now - and native ProTools is definitely 32bit float, as is almost everything else.

Many years ago (2000-2001 era) Pro Tools was set at 24 bits resolution. I'm sure it's been upgraded since then, and it would make no sense for it to process internally at less than 32bits float. But is 48 bits fixed actually better than 32 bits float? I'd have to see them side by side, comparing the details of each, before I could make that call. 32 bits float covers a LOT of territory, and it does so with less CPU overhead than 48 bits fixed.

If non-native PT is set to 48 bits fixed, I'd wonder if it has to do with some limitation imposed by the external hardware which won't permit a floating point DSP. Now, I don't keep up with Pro Tools, so I am just saying what concerns I'd want to check into before making statements like that. And after checking into them, I'd be happy with whatever the results are. So, I'm just asking if you've actually checked into that, and if you know that 48 bits fixed gives Pro Tools better resolution than 32 bits float.

Shooshie

Hi everybody, Klaus here ( mostly reading )
Since I am mixing with X32 ( Behringer ), I am impressed how it sounds, and that is 48bit fixed. Truly a difference, but I don't know if it is technically better, or ....what ?
I just like it. And the preamps and converters are probably doing their magic too

Cheers

Klaus

[EMRR]

I have tracked at 24 bit for a long time, and I've been trying to remember consistently to switch to 32fp at mix, since it sort of gives you 'no penalty' gain shifting at the mastering stage. And really, why not record the full system resolution.

64fp audio sounds like a pointless exercise. 64 fixed would surely be enough if you could even find a system that would capture that resolution. As stated plenty here, 32fp audio files surpass the laws of nature that we need to be concerned with.