EXS 24 Logic Sampler Users Group

Bob Vandiver wrote: 

>
> Does one get the benefit of 32 bit float after samples have been imported
> from, say, gig format? Or does the benefit only occur if the samples are
> imported with 32 bit mode active?


As far as Gig conversion goes, all the conversion is doing is extracting the 16
bit wavs from the gig archive and storing them as 16 bit wavs on disk.
Gigasampler does not yet support 24 bit audio formats.

The 32 bit preference thing is merely a CPU saver.  All audio passing through
the audio engine is in 32 bit float format. So, if you have not activated 32
bit float setting in the EXS24 prefs,  the samples load into RAM in 16 or 24
bit format -- whatever they are in on hard-disk. 

Before each voice can be processed by the EXS24 it has to be converted on the
fly to 32 bit float -- this process consumes CPU cycles. If you activate the 32
bit setting in the prefs this conversion takes place during loading of the
instrument to RAM and you save on CPU when the instrument is playing.

This benefit is probably negated if you have disk steaming active as the
conversion of samples streaming up from the hard disk has to happen on the fly.
With streaming active only the initial attack of samples is stored in RAM --
the rest streams up from disk as needed. This might explain the limitation on
the sample start modulation (set it all the way up and you are still only a
little way into the sample) -- if it was otherwise the whole sample would need
to be in RAM for instant response.


Regards,
Murray