KVR Audio

liam1895 · Post by **liam1895** » Mon Aug 13, 2012 5:41 am

I've been looking around and trying to really find out the difference between the two but im struggling finding any straight up answers.

Is there really any human audible difference?

hibidy · Post by **hibidy** » Mon Aug 13, 2012 6:12 am

manducator · Post by **manducator** » Mon Aug 13, 2012 8:45 am

Or do you mean 64-bit floating point?

liam1895 · Post by **liam1895** » Mon Aug 13, 2012 9:09 am

manducator wrote:Or do you mean 64-bit floating point?

floating point

BertKoor · Post by **BertKoor** » Mon Aug 13, 2012 9:49 am

liam1895 wrote:floating point

The only benefit I see of that is for e.g. convulsion, impulse response based calculations, some specific filter designs. Where loads of calculations are stacked and small rounding errors can become bigger. The plugin can do that internally in 64bit if needed, input and output can be in 32bits. For all other uses 32bit is more than sufficient.

I'd even go as far as stating that for most cases (simple mixing) 16bit audio storage is enough. In the final song the volume of one instrument track is lowered by something like 10dB, so only the top-most 14 bits are audible. That's why you can get away with AD-converters with a signal-noise ratio of around -100dB. With the track volume lowered that's around -110dB. CD audio with 16bit can only reproduce -96dB.

tapper mike · Post by **tapper mike** » Thu Aug 16, 2012 11:26 pm

Here we go again.

maxxxter · Post by **maxxxter** » Thu Aug 16, 2012 11:35 pm

DELETED

BertKoor · Post by **BertKoor** » Fri Aug 17, 2012 11:30 am

Nitpicking on my spelling/grammar... Got nothing better to do?? You understood what I was saying. I'd like to see you write in a foreign language.

mfb · Post by **mfb** » Sat Aug 18, 2012 8:31 pm

liam1895 wrote:I've been looking around and trying to really find out the difference between the two but im struggling finding any straight up answers.

Is there really any human audible difference?

Okay - by 64 and 32 bits you probably 64 and 32 bit processor (CPU) and software. This is just about the internal workings of your of how your CPU crunches bits, with 64-bit systems generally being a bit faster and able to address more memory than 32-bit systems. Note that 64-bit CPU's are capable of running either 32 or 64 bit software. These days pretty much all computers have 64-bit processors, but some people still
use 32-bit operating systems & software. Whether you process on a 32-bit or a 64-bit system/software should make
absolutely no difference to your audio if you are using the same program. It may, however, make a difference to which programs run at all, and how fast they run.

Now a different thing is the bit-depth of your audio.
Audio can be encoded so that each sample uses more or
fewer bits. The most common is 16 bits (normal CD audio),
other values frequently used are 8,12, 24 and 32 bits.
8 will sound decidedly lo-fi, 12 bits was used for many
early samplers and some people love the sound. These
days anything less than 16bits is uncommon. (The bit
depth of your audio is independent of whether you have
a 32 or a 64-bit processor)

Basically the higher the bit depth, the more accurately the
audio signal is represented, but the larger your file sizes,
and the higher your processing requirements. I'd be surprised
if anyone can hear the difference between 32 and 64 bits
(I'm not even sure there is much audio software
that uses a bit depth of 64 out there). Some people can
hear the difference between 16 and 32 bits, probably many
people cannot. If you're just playing back your audio, then
16-bit is probably fine. If you're processing it to hell
and back, for example extreme compression or EQ then it
might be worth going to a higher bit-depth.

re8 · Post by **re8** » Mon Aug 20, 2012 8:05 pm

mfb wrote:
liam1895 wrote:I've been looking around and trying to really find out the difference between the two but im struggling finding any straight up answers. Is there really any human audible difference?
Whether you process on a 32-bit or a 64-bit system/software should make absolutely no difference to your audio if you are using the same program. It may, however, make a difference to which programs run at all, and how fast they run.

Earlier versions of Abelton Live (v5 IIRC) had users complain that it "sounded digital" & was "unfit for (pre) mastering". v6 or v7 they added "Hi-Quality" mode which is the same effect in 64bit opcodes. Complaints died down after that; "digital sounding" complainers are told to use "Hi-Quality" mode.

asantoni · Post by **asantoni** » Fri Aug 24, 2012 6:59 pm

As BertKoor said, it's probably really only relevant if you're using loads of filters or things like convolutions where error could stack up into some audible distortion or noise. If you can't hear anything "sounding digital", then you don't have a problem. (Hey, we got this far with 32-bit electronic music, right?)

If you're not confused enough already, whether or not your plugins and DAW process audio as 64-bit has nothing to do with whether or not the plugins or DAW are 64-bit.

You could be running Windows 95 on a Pentium, and your DAW could be processing stuff as 64-bit.

What matters is the internal precision that the DAW and plugins are processing audio as. I think a handful of DAWs like FL Studio have been advertising internal 64-bit processing for quite a while, and this is why. Does it make a difference in terms of audio quality? 98% of the time, probably not. This is what the "high quality" mode in Ableton that re8 referred to does.

Where it is more likely to matter is speed. If you want to use a 64-bit processing engine, it will help if the host is 64-bit. Unfortunately, it's not always a clear performance boost to use a 64-bit host with 64-bit audio processing. The reason for this is two-fold:

1) Since even before 2004, all Intel CPUs had this feature called "SSE2", which accelerate 64-bit calculations, and it's often used to optimize the most performance-critical part of audio software, like filters. So the most CPU-intensive parts of the code could already have been fairly well optimized regardless of whether the program required a 64-bit processor or not.

2) In 64-bit land, you've got pointers that are now double the size, more registers, and basically, more of everything. There's more data for the CPU to move around, and that means the performance bottlenecks can shift. Memory bandwidth can be more of an issue.

What these factors means is that simply using the 64-bit version of a DAW doesn't mean you get better performance. One clear benefit is that on Windows, your DAW can use more than 4 GB of RAM, which can be great if you're planning a whole live set in something like Ableton Live. But anyways, don't feel bad about being confused, because this stuff will make your head spin. Use the 64-bit version of your host if there is one, but don't feel bad if you have to stay 32-bit for a while.

(Did I just do that "devs hijacking every thread" thing people are complaining about???)

docdued · Post by **docdued** » Sat Aug 25, 2012 9:33 am

It's really unfortunate that we have the 32 bit and the 64 bit option both in the audio bit depth and in memory/CPU layout... I really don't want to know how many people mix these up without knowing. Why couldn't they make 30 and 60 bit CPUs instead, for instance

? The world would be so much simpler...

32 vs 64 bit audio processing