VI/Logic as a composing tool driven by notation software

[michael.matthews]

I have been experimenting with the VI Solo Stings collection, working it into a few of my earlier pieces. The sounds and the flexibility are wonderful, and the matrix organization is very well thought out.

I do have a concern. I have always used VSL in conjuntion with Finale and Logic as a composing tool - a way to hear pieces as I work. I do all of my work in Finale, creating and tweaking the VSL patches in Logic as I need to. I send the Finale MIDI data through the IAC ports to Logic, with Finale acting as the playback engine.

I wonder whether with this release the VSL team is relegating people who use VSL in this manner to more of a back seat than we perhaps already have. I have a dual 2.5 G5 with 8 GB of RAM. With that amazing setup I can hardly run the 22 instances of solos strings (6 6 4 4 2) that I need for my current piece. I get lots of clicks, pops and other sound anomolies, as well as logic crashes when I try to edit the matrices.

I am attempting to load only a single matrix into each VI instance. I had started out trying to use the combi matrix, but gave up on that, and have constructed some really stripped down matrices, roughly equivalent to the "basic_all" combination of the VSL solo strings. Even with this configuration I have many problems.

I understood that there would be some performance hit with the VI, but I am somewhat shocked by how much. With my previous 4 GB of RAM, I was able to play back full orchestra pieces in real time (3333 4331 timp perc harp strings). I bumped my computer up to 8 GB of RAM for VI, and it is clear to me that, unless I am missing something, I will never be able to even remotely approach this with VI.

This brings me to the freeze function, and my earlier musing that users such as I are becoming more peripheral. The freeze function seems useless for my application (I hope I am mistaken here). I need to play back, change an note here, a rhythm there, add a measure, etc., and play back again. I would have to freeze and unfreeze every instrument every time.

I am asking the VSL team to make some comments about their support for this type of usage. I love the instruments and the company, and I really want to be able to use VI for my work. Am I misssing something in my preliminary assessment, or will something change? Are we notation software playback people (NSPP's) a factor, or are we the destined to be slowly relegated to the outer darkness, where we shall become lost, wandering souls?

Thanks,

Michael

[Jonathan Mitchell]

Thanks for raising this one Michael. I'm very interested to see the answer. I'm also in the 'composer' category, rather than a programmer/mixer. If I could fire Sibelius at VI and have my standard orchestra 2222, 322, timp, perc, normal strings + pizz DB, playback while I wrote then I would be pulling out my chequebook to upgrade to VI immediately. With the sample libraries, I can just about get away with that format with the VSL Pro Cube, Sib and Giga with about 120 samples on one 3.6Ghz with 2Gb mem and a pair of 250Gb fast drives on board.

[herb]

I think everything depends on your workflow and style.

When your are programming a piece where you need the full potential of an instrument, that means where you have previously used six to ten, maybe more EXS instances for one instrument, you are definetly saving a lot of CPU power, because now you can manage the programming in one VI instance.

On the other site, when you programm a piece where you only need a single articulation or a simple articulation combination which was previosly also manageable with one EXS instance your are loosing CPU power.

For my work the advantage is clear, what I've previolsy done on three differemt PCs, I can can do it now on one PC and I'm also five to ten times faster.

best
Herb

[michael.matthews]

Thanks for the reply, Herb. Obviously you have thought extensively about VI and I am just getting started with it. Any suggestions would be appreciated.

Take my current piece. It is for 22 solo strings (66442), each on a seperate staff (and midi channel). The instruments often have completely independent parts, with simultaneous multiple dynamic layers. Ideally I would like each, at a given time, to be able to switch between, stacc., detaché short and long, sustained, marcato, tremolo, sul pont sus. and tremolo, pizz., snap pizz, col legno, harmonics, fp, trill, for example. The L1 articulation combi covers most of this (excluding col legno), but I am simply unable to load 22 instances of this patch simultaneously (one combi matrix per instance). Even with 12-15 instances I get lots of clicks in the sound.. This is with a dual 2.5 G5 with 8 GB or RAM.

Is there a better way to do this?

Thanks,

Michael

[herb]

If you are writing for 22 different instruments, it's the best way to use 22 different instances, so that each instrument can be programmed complete independently.

I'm not the OSX guy here, but I think it should be possible to perform 22 instances on a dual 2.5 G5. Maybe you are overloading RAM as Nick described in a different topic, going over 2,7 GB could cause these problems.

Maybe the upcoming 1.05 update will be helpful for your situation.
It supports shared memory. So you can load different instances with the same patches/samples without rising up RAM. Especially for your piece where you need 12 different instances of the violins this could be a big RAM safer.

best
Herb

[michael.matthews]

Thanks for the info, Herb. The 1.05 update sounds like it might be a big help. Regarding the post from Nick, do you have a link or know the name of the thread. I tried a search but could not find the post.

Regarding 1.05, how far along is this in the development process?

Best,

Michael

[herb]

http://community.vsl.co.at/viewtopic.php?t=8395&postdays=0&postorder=asc&start=30&iframe=true

best
Herb

[magates]

Finale is a hog. You need to be runing finale on a seperate machine if possible. It will easily take up a full CPU. Look at sysetm usage while runing finale and logic.

[stevesong]

Hi Matthew:

I'm wondering what Logic's I/O buffer settings are in your setup. I'm using 512KB and not having problems with clicks and noises except when an instrument is played for the first time. Subsequent playback is fine. Also when running a setup with Finale playing back Logic's audio instruments it is extremely important that you turn off "Auto enable external sync" in Logic. (This is found under File Menu/Song Settings/Synchronization - in the ensuing dialog box uncheck "Auto enable external sync.") If this is not turned off, you will get all kinds of artifacts during playback.

Also, what version of Finale are you using? I've found Finale 2005 far more reliable for playback than Finale 2006. I've never gotten Finale 2006 to playback external MIDI devices without lots of stuck notes unless I interpose a MIDI Patchbay configuration between Finale 2006's virtual outputs and the external device (rather than directing - - as one normally would - - Finale's outputs to the intended AMS object). Also no existing version of Finale is written to utilize both processors in dual processor machines which is one reason why Activity Monitor shows it to be taking up about 80% of CPU cycles during playback. I'm not convinced that this is a correct reading. For example, I have 48 channels of EXS24 and VI instruments running in Logic and Activity Monitor reports 58% of CPU cycles are in use during playback. (Logic, of course, makes use of both processors in dual processor Macs). Since there will soon be lots of dual core Windows machines, there may be some, however faint, hope that the next version of Finale will be able to make use of more than one processor.

[JWL]

@michael.matthews said:

I have a dual 2.5 G5 with 8 GB of RAM. With that amazing setup I can hardly run the 22 instances of solos strings (6 6 4 4 2) that I need for my current piece. I get lots of clicks, pops and other sound anomolies, as well as logic crashes when I try to edit the matrices.

...

DAW's in general have a 3-4 GB limit. Have you tried running the standalone version of VI along side the plugin version? I've not done this yet (am upgrading from 4.5 to 8 GB RAM this week). But I do recall someone here proclaiming success with this method in getting more instances to load.

I've got a 2.5 Ghz Dual, and with 4.5 GB RAM I've gotten up to 12 instances going (4+ matrices each)...

[michael.matthews]

@magates said:

Finale is a hog. You need to be runing finale on a seperate machine if possible. It will easily take up a full CPU. Look at sysetm usage while runing finale and logic.

Thanks, Matt.

Yes, I watch the CPU metre when Finale is running. I am not sure why it takes up so much processing power.

Michael

[michael.matthews]

@stevesong said:

Hi Matthew:

I'm wondering what Logic's I/O buffer settings are in your setup. I'm using 512KB and not having problems with clicks and noises except when an instrument is played for the first time. Subsequent playback is fine. Also when running a setup with Finale playing back Logic's audio instruments it is extremely important that you turn off "Auto enable external sync" in Logic. (This is found under File Menu/Song Settings/Synchronization - in the ensuing dialog box uncheck "Auto enable external sync.") If this is not turned off, you will get all kinds of artifacts during playback.

Also, what version of Finale are you using? I've found Finale 2005 far more reliable for playback than Finale 2006. I've never gotten Finale 2006 to playback external MIDI devices without lots of stuck notes unless I interpose a MIDI Patchbay configuration between Finale 2006's virtual outputs and the external device (rather than directing - - as one normally would - - Finale's outputs to the intended AMS object). Also no existing version of Finale is written to utilize both processors in dual processor machines which is one reason why Activity Monitor shows it to be taking up about 80% of CPU cycles during playback. I'm not convinced that this is a correct reading. For example, I have 48 channels of EXS24 and VI instruments running in Logic and Activity Monitor reports 58% of CPU cycles are in use during playback. (Logic, of course, makes use of both processors in dual processor Macs). Since there will soon be lots of dual core Windows machines, there may be some, however faint, hope that the next version of Finale will be able to make use of more than one processor.

Thanks, Steve. Yes, there are certainly fewer clicks at 512 than at 256. I am pretty sure that I have "Auto enable external sync" turned off in Logic. I am not at home right now but will confirm when I return. Thanks for the tip.

You are correct - Finale 2006 is more problematic for playback than 2005. I also use MIDI Patchbay.

The bulk of my current problems are, I believe, CPU related. having to do with how much processing power Finale and Logic/VI consume between them. When I use Finale to drive Logic with the 22 instance VI setup that I describes in a previous post, I get serious clicks at the onset of multiple simultaneous notes. If on the other hand, I generate a SMF of the same piece, import it into Logic and play it back, the clicks virtually disappear.

I also note the the memory sharing feature of the VI 1.05 update has been a big help with RAM overload.

I, too, share your home that Finale 2007 will be more processor efficient.

Michael

[michael.matthews]

@JWL said:

DAW's in general have a 3-4 GB limit. Have you tried running the standalone version of VI along side the plugin version? I've not done this yet (am upgrading from 4.5 to 8 GB RAM this week). But I do recall someone here proclaiming success with this method in getting more instances to load.

I've got a 2.5 Ghz Dual, and with 4.5 GB RAM I've gotten up to 12 instances going (4+ matrices each)...

Thanks. I have not tried this yet. It is next on my list, though I have not quite figured out how to do it.

Michael

[stevesong]

Matthew:

One last thing you might try increasing Logic's I/O buffer size to 1024. This increases latency, of course, but if you're playing back from Finale, that won't matter much. Again I'm not convinced that Activity Monitor is giving an accurate reading of Finale's CPU usage.As I mentioned I get clicks from VI instruments only when I first play an instrument. Once an instrument has been played a little, the clicks go away and usually don't return.

[michael.matthews]

@stevesong said:

Matthew:

One last thing you might try increasing Logic's I/O buffer size to 1024. This increases latency, of course, but if you're playing back from Finale, that won't matter much. Again I'm not convinced that Activity Monitor is giving an accurate reading of Finale's CPU usage.As I mentioned I get clicks from VI instruments only when I first play an instrument. Once an instrument has been played a little, the clicks go away and usually don't return.

Thanks for this tip. I will give it a shot.

Michael