Subject: Re: [linux-audio-dev] EVO status...
From: Ruben van Royen (rvroyen_AT_guidedbees.com)
Date: Thu Jan 17 2002 - 17:46:57 EET
On Wed, Jan 16, 2002 at 03:28:53PM +0100, Tobias Ulbricht wrote:
>
> On Tue, 15 Jan 2002, Paul Davis wrote:
>
> > >Thats actually a good point... I guess the answer is if 2 GB is
> > >enough RAM to have enough channels to overload the CPU and IO
> > >bandwith of the host. Things like GigaSampler allow you to layer and
> > >a bunch of instruments into one channel.
> > >
> > >It would still limit you when using things like GigaPiano which has a
> > >_huge_ sample size for each piano.
>
> What about dynamic sampling?
> i.e. different samples for different velocity ranges?
> Is that already included in *one* .gig-sample?
> I've never played with GigaSampler, so I dunno how they look like.
>
> At least I can imagine, that you will always be able to increase the
> throughput proportional to the "sound quality" (by higher quality samples,
> layering of multiple samples, and younameit.), so I'd like to see the
> cached HD sampling.
Yes, this is done, there are different velocity layers possible. Actually,
you can have lots of layers in different dimensions, for example creating
samples with both the sustain pedal up and down. Actually, I know of at
least one project where a piano was sampled and the result didn't fit in a
.gig, because of the 2GB filesize limitation imposed by windows. I believe
8 velocity layers with and without sustain were used.
>
> >
> > thanks for reminding me of the other 2 reasons why "just use lots of
> > RAM" doesn't work in the general case.
> >
> > --p
> >
>
This archive was generated by hypermail 2b28 : Thu Jan 17 2002 - 17:44:08 EET