32-bit integer / 64-bit float meaning

New 9.5.40.update says: It is now possible to record audio files with a 32-bit integer bit depth and also export audio in 64-bit float and 32-bit integer.

What does it mean exactly and what are benefits?
Should one record in 32 bit (float or integer) then dihter master to 16bit?


Normally when you do a recording you record the exact bit-pattern from your AD-converter. The formats used there 16 or 24 bits. AD converters of today is usually better 16 bit, none are better than 24. But then you start to do processing. In cubase the processing is done on floating point. And when you convert integer to floating point or vice versa you might lose information. So it is good to be able to save in a format that does not has any loss. The processed tracks can then be sent to a mixing engineer on a other site without any loss. The mixer do his job and produce a new floating point that is sent to a mastering, and that will then be converted to integer for the consumer media that usually still is 16 bit. And the difference is quite small, but there is today no reason not to use technology we have. The extra cost is small.

Thnx :slight_smile:
So, should I always record in 32 bit float point just to be sure I have the best quality overall (for processing etc.)
When I render individual track (or master) to 24 bit should I apply dithering, is there option to dither from 32 to 24 bit?

Technically 32bit float is still 24bit Audio Information with a lot more headroom.
The real benefit is being able to turn down the signal when there is signal over 0dB and still maintain the waveform.
No clipping, so to say when processing, you can turn the signal down and recover.
On the way In you can clip the AD converter, that will still chop of the top of a wave, no matter what format you use.
32bit int. And 64 bit float are are welcome additions, when working across different DAW’s

What’s the difference between 32bit (float or integer) and 64 bit?

has anybody tested the cpu usage differences when using 24 bit or 64 bit in cubase?

Ehh Cubase works either in 32bit float or in 64Bit double precision.
Performance depends on system and plugins (vst3 or vst2) used.
Personally I get a performance increase when using 64Bit double precision

One second:

Your CPU meter goes down when you switch to 64bit?

Wowser. Have to test it myself tonight

Yes, but again depends.
You should try yourself.
There is the theoretical improvement of the sound using 64Bit double precision.
As the name suggests, it is more precise. Fewer rounding errors.
Some claim that they can hear it, I don’t.

Read this thread, most interesting PG’s explanation and pro’s and con’s of a 64-bit float mix engine.