KVR Audio

hibidy · Post by **hibidy** » Sat Aug 16, 2008 8:58 am

I still think it's funny how all the "mags" are reviewing version 4.......I wonder how many people are buying it without knowing it's disco'ed.......?

bduffy · Post by **bduffy** » Sat Aug 16, 2008 8:59 am

It's gone disco?

egbert · Post by **egbert** » Sat Aug 16, 2008 10:14 am

bduffy wrote: I wonder how Kontakt 3 stands up (probably shite; seems like everything I buy has the worst aliasing/quality/etc)?

Kontakt 2.x brought in three levels of quality for the resampling engine. SFZ has a similar sliding scale - just with more steps. The lowest level in either case maximises voices at the expense of quality and the higher levels and use many more interpolation points and therefore use much more CPU.

The lowest level in K2 is probably comparable to K 1.x quality and similar to EXS24 and Halion etc - loads of aliasing for large pitch shifts. (Build some patches out of hardware synth samples in the early version of any of these samplers - say two per octave - and play some four of five voice chords: the aliasing is awful.)

Bear in mind that none of this matters if you are not actually re-pitching samples (either transposing by some number of semitones up or down or correcting mistuned notes by a few cents). If you have samples for every note and the samples are tuned already then there will only be differences if you are using the filters and envelopes. Simple playback of unprocessed drum samples for example should be more or less identical across all the major samplers.

If you are using the higher quality settings in K2 or K3 you are not losing anything to Giga in terms of playback quality. Giga had the most efficient disk streaming engine though.

bduffy · Post by **bduffy** » Sat Aug 16, 2008 10:20 am

Thanks for the further explanation. I was aware the difference only comes with re-pitching or processing. I do wish someone would update those tests with the latest versions, though.

aMUSEd · Post by **aMUSEd** » Sat Aug 16, 2008 10:30 am

lol I just read the SOS review. Not very good really.

One thing that got my attention though is that the late v4 can load and play VSTi's and since it uses kernel streaming GSIF2 drivers latency is really low. It made me wonder why a GSIF based host has never been develped and why GSIF has never been promoted as a competitor to ASIO and WDM. If it can achieve much lower latency with less CPU overhead you'd have thought it would be ideal for a softsynth focussed host.

bduffy · Post by **bduffy** » Sat Aug 16, 2008 10:31 am

I don't know: is it proven to be faster? I don't recall very good performance from our Giga machine at work (in fact, it sucked).

kara · Post by **kara** » Sat Aug 16, 2008 10:33 am

hibidy wrote:I still think it's funny how all the "mags" are reviewing version 4.......I wonder how many people are buying it without knowing it's disco'ed.......?

True, I was highly surprised that there was a review in the latest CM....

k

aMUSEd · Post by **aMUSEd** » Sat Aug 16, 2008 10:41 am

I bet all the mags with multi page reviews are pretty pissed about having devoted all that space and time to reviewing a product that was dead already.

aMUSEd · Post by **aMUSEd** » Sat Aug 16, 2008 10:43 am

bduffy wrote:I don't know: is it proven to be faster? I don't recall very good performance from our Giga machine at work (in fact, it sucked).

Well I always thought that was one of the reasons so many stuck with Giga - the responsiveness and streaming was better (apparently). I've heard of people dedicating whole PC's to it although I believe it's also picky about what it will work with but maybe if development had been more focussed on the drivers that could be overcome. Surely a kernel level driver has to be faster?

bduffy · Post by **bduffy** » Sat Aug 16, 2008 10:50 am

aMUSEd wrote:
bduffy wrote:I don't know: is it proven to be faster? I don't recall very good performance from our Giga machine at work (in fact, it sucked).
Well I always thought that was one of the reasons so many stuck with Giga - the responsiveness and streaming was better (apparently). I've heard of people dedicating whole PC's to it although I believe it's also picky about what it will work with but maybe if development had been more focussed on the drivers that could be overcome. Surely a kernel level driver has to be faster?

I don't know. It sounds like it should be faster, and it's a kind a cheat if you dedicate the whole machine to one instrument, when you think about it. In my case, the GSIF machine was nothing like the Nuendo machine, with matching hardware, replete with RME interface. But if anyone's got proof, I'm all ears.

aMUSEd · Post by **aMUSEd** » Sat Aug 16, 2008 11:04 am

bduffy wrote:
aMUSEd wrote:
bduffy wrote:I don't know: is it proven to be faster? I don't recall very good performance from our Giga machine at work (in fact, it sucked).
Well I always thought that was one of the reasons so many stuck with Giga - the responsiveness and streaming was better (apparently). I've heard of people dedicating whole PC's to it although I believe it's also picky about what it will work with but maybe if development had been more focussed on the drivers that could be overcome. Surely a kernel level driver has to be faster?
and it's a kind a cheat if you dedicate the whole machine to one instrument, when you think about it.

Sure but dedicating a machine to a software instrument host wouldn't be so bad. I've done that myself with my laptop already and so has Receptor. I just wondered if you could develop such a thing based on the GSIF2 driver framework and market it as an ultra low latency players host.

bduffy · Post by **bduffy** » Sat Aug 16, 2008 11:18 am

aMUSEd wrote:
bduffy wrote:
aMUSEd wrote:
bduffy wrote:I don't know: is it proven to be faster? I don't recall very good performance from our Giga machine at work (in fact, it sucked).
Well I always thought that was one of the reasons so many stuck with Giga - the responsiveness and streaming was better (apparently). I've heard of people dedicating whole PC's to it although I believe it's also picky about what it will work with but maybe if development had been more focussed on the drivers that could be overcome. Surely a kernel level driver has to be faster?
and it's a kind a cheat if you dedicate the whole machine to one instrument, when you think about it.
Sure but dedicating a machine to a software instrument host wouldn't be so bad. I've done that myself with my laptop already and so has Receptor. I just wondered if you could develop such a thing based on the GSIF2 driver framework and market it as an ultra low latency players host.

We'll probably never know...

aMUSEd · Post by **aMUSEd** » Sat Aug 16, 2008 11:59 am

Well just as an experiment I thought I'd try this kernel streaming thing. Foobar has a kernel streaming mode driver that seems to work on my system and there is a VST host plugin for Foobar - I've already got it loading up EnergyXT to use as an audio host and will see how well this works. No midi unfortunately but I could use this for fx and audio (might be cool for live guitar processing for example and I could record the output using eXT or my Scope sound recorder)

bduffy · Post by **bduffy** » Sat Aug 16, 2008 5:23 pm

aMUSEd wrote:Well just as an experiment I thought I'd try this kernel streaming thing. Foobar has a kernel streaming mode driver that seems to work on my system and there is a VST host plugin for Foobar - I've already got it loading up EnergyXT to use as an audio host and will see how well this works. No midi unfortunately but I could use this for fx and audio (might be cool for live guitar processing for example and I could record the output using eXT or my Scope sound recorder)

Oh...crazy! Funny, Foobar has always been a little unstable on my system, playback-wise. It doesn't seem to like my EMU card...

egbert · Post by **egbert** » Sat Aug 16, 2008 5:59 pm

I gather that the trend is running in the other direction - audio hardware and video card drivers are moving more layers away from the kernel. The idea is to make the machine/OS more crash proof - stability at the expense of performance.

I don't use Vista but I gather you can shut down an application using a crashed soundcard driver without rebooting. When your soundcard driver runs in kernel mode and the app crashes you generally need to reboot - this has certainly been my experience (eg with with Cubase and RME ASIO drivers).

Giga Studio is no MORE! At least with Tascam!