64 Bit FP Audio Engine?
-
- KVRer
- 14 posts since 7 Mar, 2010 from Aspen, CO
Personally I see no reason why this would be necessary. There is MORE than enough headroom in 32 bit floating point numbers. I'm not entirely sure which format of Floating Point numbers are typically used in audio applications, but I imagine the exponent portion of the number would allow you to represent the values closer to zero?
Either way, unless there was some computational benefit to using double precision floating point numbers in audio (which I do doubt). Typically double precision floating points have a longer latency bound for division/multiplication (form multiplication is 4 CPU cycles for single and 5 for double).
TL;DR: 32 bit offers sufficient headroom and the cons of 64bit (speed) likely make it not worth it to have more (inaudible) headroom.
Either way, unless there was some computational benefit to using double precision floating point numbers in audio (which I do doubt). Typically double precision floating points have a longer latency bound for division/multiplication (form multiplication is 4 CPU cycles for single and 5 for double).
TL;DR: 32 bit offers sufficient headroom and the cons of 64bit (speed) likely make it not worth it to have more (inaudible) headroom.
-
- KVRer
- 14 posts since 7 Mar, 2010 from Aspen, CO
Actually after some reading, it seems that using double precision floating points is pretty common in audio these days! I stand corrected!
-
qtheerearranger qtheerearranger https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=325452
- KVRian
- 681 posts since 26 Mar, 2014 from Denver, Co
Yeah I mean 32 Bit floating is phenomenal and to my knowledge there are only a few DAWs with 64 Bit Floating, including Pro Tools. I'm sure Bit wig will get there one day, but like if you care about 64 bit floating more than the creative freedom bit wig offers you, switch to a DAW that offers that until Bit Wig gets there.
• Logic Pro 10.8.1
• MacBook Pro 2023 - M2 MAX - 96 GB RAM
• Focusrite Red 8Line + UAD Satellite
• MacBook Pro 2023 - M2 MAX - 96 GB RAM
• Focusrite Red 8Line + UAD Satellite
- KVRAF
- 2938 posts since 9 Dec, 2011 from falling
Are you saying you know the audio engine is 32 bit floating point? I haven't managed to find it documented anywhere.qtheerearranger wrote:Yeah I mean 32 Bit floating is phenomenal and to my knowledge there are only a few DAWs with 64 Bit Floating, including Pro Tools. I'm sure Bit wig will get there one day, but like if you care about 64 bit floating more than the creative freedom bit wig offers you, switch to a DAW that offers that until Bit Wig gets there.
Bitwig Certified Trainer
-
qtheerearranger qtheerearranger https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=325452
- KVRian
- 681 posts since 26 Mar, 2014 from Denver, Co
I just assumed it was, haha. As I said though, I'm sure bit wig will get caught up one day, they just started though! It's kinda unusual for a DAW company to not even have one quote about their audio engine on their website however.
• Logic Pro 10.8.1
• MacBook Pro 2023 - M2 MAX - 96 GB RAM
• Focusrite Red 8Line + UAD Satellite
• MacBook Pro 2023 - M2 MAX - 96 GB RAM
• Focusrite Red 8Line + UAD Satellite
- KVRAF
- 2938 posts since 9 Dec, 2011 from falling
I didn't find it in the manual either.qtheerearranger wrote:I just assumed it was, haha. As I said though, I'm sure bit wig will get caught up one day, they just started though! It's kinda unusual for a DAW company to not even have one quote about their audio engine on their website however.
Bitwig Certified Trainer
-
- KVRian
- 653 posts since 4 Apr, 2010
The trouble with saying "64-bit" these days...well, I think most/all of the people here in the thread so far understand the difference between a 64-bit OS, and processors, which is about memory addressing, and 64-bit floating point audio, but even then...
There's 64-bit audio busses, and there's 64-bit computation. The latter is pretty trivia for host-based DSP, at least using the general-purpose floating point features of modern processors. And yes, there are plenty of times where 32-bit floating point isn't good enough for audio DSP algorithms (essentially 25 bits of precision, but multiplies and accumulations gets immediately truncated to that same precision—the resulting error within an iterative algorithm can degrade it noticeably).
For most DAWs, the audio busses/paths (the buffers between plug-ins and mixer channels) remain 32-bit floats. That's really not a problem, since the extra precision is mainly needed during intermediate calculations—inside the plug-ins, for instance.
So while some of the latest DAWs touts 64-bit audio paths throughout, I don't see that at a serious improvement, but more of a marketing gimmick. And in the end, it's going to be output to a 24-bit DAC, and your hardware and your ears can't even resolve that much. I don't see enough error accumulating to matter.
There's 64-bit audio busses, and there's 64-bit computation. The latter is pretty trivia for host-based DSP, at least using the general-purpose floating point features of modern processors. And yes, there are plenty of times where 32-bit floating point isn't good enough for audio DSP algorithms (essentially 25 bits of precision, but multiplies and accumulations gets immediately truncated to that same precision—the resulting error within an iterative algorithm can degrade it noticeably).
For most DAWs, the audio busses/paths (the buffers between plug-ins and mixer channels) remain 32-bit floats. That's really not a problem, since the extra precision is mainly needed during intermediate calculations—inside the plug-ins, for instance.
So while some of the latest DAWs touts 64-bit audio paths throughout, I don't see that at a serious improvement, but more of a marketing gimmick. And in the end, it's going to be output to a 24-bit DAC, and your hardware and your ears can't even resolve that much. I don't see enough error accumulating to matter.
My audio DSP blog: earlevel.com
- KVRAF
- 2938 posts since 9 Dec, 2011 from falling
It would be cool if one of the wonderful Bitwig folks would answer this one for us.
Bitwig Certified Trainer
- KVRist
- 301 posts since 28 May, 2013 from Utah
No point in having a 64 bit audio engine until Bitwig implements true pan knobs and mono tracks. If you can't get the proper audio from source it doesn't make to much of a difference.
- KVRAF
- 2938 posts since 9 Dec, 2011 from falling
Has this question re: the Bitwig audio engine been answered anywhere?
Also, are there plans for 32-bit float recording?
It would be really cool if someone from Bitwig could clear this all up.
Also, are there plans for 32-bit float recording?
It would be really cool if someone from Bitwig could clear this all up.
Bitwig Certified Trainer
- KVRAF
- 6305 posts since 9 Dec, 2008 from Berlin
It's 32 Bit floating point since that is what makes the most sense at this point in time.
Almost all audio interfaces are 24 Bit non-floatingpoint AFAIK, so that is why recording is using the same I would assume.
But if you have a case where it would make real sense to do it differently, write to tech-support and tell them about it. The devs are very down to earth but open to sensible ideas - just not into marketing fluff
Cheers,
Tom
Almost all audio interfaces are 24 Bit non-floatingpoint AFAIK, so that is why recording is using the same I would assume.
But if you have a case where it would make real sense to do it differently, write to tech-support and tell them about it. The devs are very down to earth but open to sensible ideas - just not into marketing fluff
Cheers,
Tom
"Out beyond the ideas of wrongdoing and rightdoing, there is a field. I’ll meet you there." - Rumi
ScreenDream Instagram Mastodon
ScreenDream Instagram Mastodon
- KVRAF
- 2938 posts since 9 Dec, 2011 from falling
https://www.youtube.com/watch?v=Qt-EJhDDHUIThomasHelzle wrote:It's 32 Bit floating point since that is what makes the most sense at this point in time.
Almost all audio interfaces are 24 Bit non-floatingpoint AFAIK, so that is why recording is using the same I would assume.
But if you have a case where it would make real sense to do it differently, write to tech-support and tell them about it. The devs are very down to earth but open to sensible ideas - just not into marketing fluff
Cheers,
Tom
This is the tip of the iceberg. The more processing you do on a file, the more useful 32-bit floating point becomes.
Bitwig Certified Trainer