Menu
What's new
By:
Filters

Audiomodeling String Sections

DANIELE said:
I didn't think to use the Reaper FX Chain feature
Click to expand...
I would be curious to learn how many SWAM parameters can be exposed as automation or track controls in Reaper, if you have a moment later. In Cubase, there is a ton. Using a few (and locking them) on a track in this example:

1705932816146.png

... and a different "category" in the Quick Controls, since I have the old CMC-QC controller.

1705932936943.png


Some of these parameters have MIDI CC assigned, but it's also possible to control several others that do not have assignable MIDI CC.

I think there is an opportunity here for me to improve my workflow and better allocate what's going to be handled with what - via CC, vs what I can control with QC or what will be be drawn automation. For String Sections specifically, I'm currently using automation/QC to control tone, panning and general instrument side of things, while the Note Expression and Exp Map is used for articulation and perfromance:

1705933416585.png

I'm pretty happy with this so far, but would be very interested to learn your worklfow (or anyone else's if willing to share) to see how I can improve.

Cheers!
 
youngpokie said:
I would be curious to learn how many SWAM parameters can be exposed as automation or track controls in Reaper, if you have a moment later. In Cubase, there is a ton. Using a few (and locking them) on a track in this example:

1705932816146.png

... and a different "category" in the Quick Controls, since I have the old CMC-QC controller.

1705932936943.png


Some of these parameters have MIDI CC assigned, but it's also possible to control several others that do not have assignable MIDI CC.

I think there is an opportunity here for me to improve my workflow and better allocate what's going to be handled with what - via CC, vs what I can control with QC or what will be be drawn automation. For String Sections specifically, I'm currently using automation/QC to control tone, panning and general instrument side of things, while the Note Expression and Exp Map is used for articulation and perfromance:

1705933416585.png

I'm pretty happy with this so far, but would be very interested to learn your worklfow (or anyone else's if willing to share) to see how I can improve.

Cheers!
Click to expand...
From what I see there is a ton of parameters automatable in Reaper, I think they are the same as the ones you can automate in Cubase.

I found the size parameter and it works as intended. I don't know how I didn't think about it before.

About the workflow it is pretty similar, in Reaper there aren't articulation maps for now so I a mix of Reaticulate and normal keyswitches (because these instruments often require continuous ks).

Then I use all the CCs I need (basically Dynamics, Vibrato, Position, Pressure, Pitch Bend if needed).

In the end I'll now use an external automation for the ensemble size.

My workflow is still on the go as I'll refine it while I'm learning to use the instruments.
 
youngpokie said:
I would be curious to learn how many SWAM parameters can be exposed as automation or track controls in Reaper, if you have a moment later. In Cubase, there is a ton. Using a few (and locking them) on a track in this example:

1705932816146.png

... and a different "category" in the Quick Controls, since I have the old CMC-QC controller.

1705932936943.png


Some of these parameters have MIDI CC assigned, but it's also possible to control several others that do not have assignable MIDI CC.

I think there is an opportunity here for me to improve my workflow and better allocate what's going to be handled with what - via CC, vs what I can control with QC or what will be be drawn automation. For String Sections specifically, I'm currently using automation/QC to control tone, panning and general instrument side of things, while the Note Expression and Exp Map is used for articulation and perfromance:

1705933416585.png

I'm pretty happy with this so far, but would be very interested to learn your worklfow (or anyone else's if willing to share) to see how I can improve.

Cheers!
Click to expand...
Would love to hear how this sounds
 
Phryq said:
Would love to hear how this sounds
Click to expand...
I'm getting close to finishing my "trial-and-error" track and then will post something here.

But I just want to ask @StefanoLucato and @lelepar - please please please allow users the option to disable the OSC communication between the instances and please please please find a way to get them to load faster. I use the external reverb and I automate the anti-divisi settings upfront in a template, but currently the plugins really slow everything to an unbearable crawl at start-up as they are trying to talk to each other even though there is no need for that at all. Thanks!

:emoji_pray:
 
youngpokie said:
I'm getting close to finishing my "trial-and-error" track and then will post something here.

But I just want to ask @StefanoLucato and @lelepar - please please please allow users the option to disable the OSC communication between the instances and please please please find a way to get them to load faster. I use the external reverb and I automate the anti-divisi settings upfront in a template, but currently the plugins really slow everything to an unbearable crawl at start-up as they are trying to talk to each other even though there is no need for that at all. Thanks!

:emoji_pray:
Click to expand...
Ya, I could see myself freezing everything's dry tracks and using convolution reverbs.

Though reading about it makes it seem their built in verb is super fancy.
 
If I am reading this score page correctly...

1706988105273.png

... then the setup for this is as follows:

Violin I (assuming 20 players)
- div a: 2 desks (i.e. divisi size 4, SWAM instance #1)
- div b: 4 desks (i.e. divisi size 8 (i.e. 4+4), == instance #2, #3)
- div c: 4 desks (i.e. divisi seize 8, == instance #4, #5)
Violin II (assuming 18)
- div a: 5 desks (i.e. divisi sizes 5+4, SWAM instances #6, #7)
- div b: 5 desks (i.e. divisi sizes 5+4, SWAM instances #8, #9)
Viole (assuming 16)
- div a: 4 desks (4x4, SWAM #10, #11)
- div b: 4 desks (4x4, SWAM #12, #13)
Celli (assuming 14) - div a and b
D.basses (assuming 12) - div and b

and so on, until we arrive at grand total = 21 instances of SWAM Section Strings!

These setups are not the most common but they are by no means unusual. Perhaps my count is wrong (I'm really hopeless at math!), if so I would appreciate a correction. But if I am right, it seems like a lot of instances to reproduce this setup, considering how CPU hungry they are. So, while I really appreciate the work the AM team did on CPU optimization (because I can now run 4 instances in real time), the thing that I'm really struggling with is a workflow that involves live playing of parts - it's nearly unworkable because the edits on the section/ensemble level are a nightmare.

Here's the dilemma: part of me wants to see a couple of additional divisi section sizes (i.e. extending that row) so that fewer instances would be needed, perhaps half of the currently needed 21 instances (perhaps with a bigger timing accuracy range). But on the other hand, this granularity in section sizes is honestly amazing for spatialization and realism, so perhaps more CPU optimization is possible in the future.

@StefanoLucato @lelepar could you share some thoughts on this dilemma and how to improve this particular example (and workflow)? Thanks.

EDIT: I forgot to add - it takes an insane amount of time to load 21 instances.
 
Last edited:
youngpokie said:
EDIT: I forgot to add - it takes an insane amount of time to load 21 instances.
Click to expand...
Finally someone hit a problem I described early in this thread as problem since version 1.0.0. I am glad that I am not alone with this problem. I hope you contacted a support team. More person will write about this problem, more chance they fix it asap.
 
youngpokie said:
I'm getting close to finishing my "trial-and-error" track and then will post something here.
Click to expand...
Now I'm doubting this track is a good example. It's basically my "exercise" score for practicing TEController to play in all of the parts (@mbishop might remember that convo), because it's easy and short, but it's not very string-focused despite some intricacies. Anyway, here's where it's at right now (warning - loud headphone mix).

View attachment Queen of Spades - Introduction.mp3


I had some problems making this, which I didn't manage to solve:

- there are about 20 instances of SWAM section strings IIRC, because there isn't much choice of section sizes. This granularity is great for detail and spatializing, but it's not possible for me to pre-mix them using MIDI performances. They are extremely CPU hungry.

- I played everything in, starting with a blank click track and then using those MIDI recordings as subsequent references. A full track could be done quickly (depending on the complexity of the music), but with SWAM section strings this workflow is impossible. I had no choice but to bounce to audio after every 4 instances, which leads to all kinds of problems later on.

- This bouncing makes it quite inconvenient to go back and make any MIDI edits, it's simply too much hassle to freeze and unfreeze, so I gave up around the double basses (which I think it quite audible) and didn't do more than a couple of practice runs and minimal MIDI editing. Of course, I have to account for the fact that I'm not very fluent with the TEController, etc, etc - but still!

- My wish #1 - please work on additional CPU optimization and/or if possible provide additional options for larger section sizes (8 for violins, 6 for contrabbassi and so on.).

- The next problem is that once the audio is bounced, the tone is baked in (again, it's easy to hear in the mockup). There are only 2 eq controls provided, fullness and brightness. This is not enough, and certainly nothing like the 3 EQs plus formants and resonances for solo instruments. Not only is it harder to dial in the tone, I also have to go back and unfreeze MIDI once again to change anything when mixing. So again, it's too cumbersome and inconvenient and I gave up.

- My wish #2 would be for Audio Modeling to figure out some kind of solution to enable better tone shaping per each instance. I understand that's not easy with a plugin of a group of instruments. Perhaps one day I'll gain enough experience to make a fully pre-mixed template, but the reality so far is that my section counts change depending on the project, and so I'm doubtful it's going to fix anything.

- Btw, I've only used a small amount of EQ and a "glue" compressor preset on the master bus, nothing at all on the individual tracks (no automation of any kind, too). So the thinness of strings, the boominess, harshness and honkiness all come from the individual instruments as well my completely non-existent mixing skills.

- Problem #3 (maybe only for me) is that the current SWAM key switch system is not very good. There is a short section with orchestral bariolage in this mockup that requires several key switches simultaneously: polyphony, timing accuracy, bow lift, bow pressure, alternative fingering, etc. That's laughable when playing live - nobody will have enough fingers, arms and feet for that.

- What I would love is for Audio Modeling to reserve one or two "special" key switches that can work as combination switches and allow the user to map several CC controls to a single key switch and to freely choose it's MIDI note (not simply the octave).

Anyway. Appreciate any comments on this. I'm not going to touch it anymore, but would be helpful to know if anything is off that I didn't hear. Thanks.
 
The strings sound much better than I'd expected (I've been having a go with SM Strings). I really like the composition! The room has maybe more tail than my taste (you'd mentioned that it sounded boomy), but maybe the compression is bringing that out. It's making me much more interested in the AM String Sections than I was, but I'll probably wait to see how the GPU Audio integration changes the performance. Are you using the built-in reverbs? I'm curious as well, how did you map your TEC controller to the CC inputs? What's working for you?
 
youngpokie said:
Now I'm doubting this track is a good example. It's basically my "exercise" score for practicing TEController to play in all of the parts (@mbishop might remember that convo), because it's easy and short, but it's not very string-focused despite some intricacies. Anyway, here's where it's at right now (warning - loud headphone mix).

View attachment Queen of Spades - Introduction.mp3


I had some problems making this, which I didn't manage to solve:

- there are about 20 instances of SWAM section strings IIRC, because there isn't much choice of section sizes. This granularity is great for detail and spatializing, but it's not possible for me to pre-mix them using MIDI performances. They are extremely CPU hungry.

- I played everything in, starting with a blank click track and then using those MIDI recordings as subsequent references. A full track could be done quickly (depending on the complexity of the music), but with SWAM section strings this workflow is impossible. I had no choice but to bounce to audio after every 4 instances, which leads to all kinds of problems later on.

- This bouncing makes it quite inconvenient to go back and make any MIDI edits, it's simply too much hassle to freeze and unfreeze, so I gave up around the double basses (which I think it quite audible) and didn't do more than a couple of practice runs and minimal MIDI editing. Of course, I have to account for the fact that I'm not very fluent with the TEController, etc, etc - but still!

- My wish #1 - please work on additional CPU optimization and/or if possible provide additional options for larger section sizes (8 for violins, 6 for contrabbassi and so on.).

- The next problem is that once the audio is bounced, the tone is baked in (again, it's easy to hear in the mockup). There are only 2 eq controls provided, fullness and brightness. This is not enough, and certainly nothing like the 3 EQs plus formants and resonances for solo instruments. Not only is it harder to dial in the tone, I also have to go back and unfreeze MIDI once again to change anything when mixing. So again, it's too cumbersome and inconvenient and I gave up.

- My wish #2 would be for Audio Modeling to figure out some kind of solution to enable better tone shaping per each instance. I understand that's not easy with a plugin of a group of instruments. Perhaps one day I'll gain enough experience to make a fully pre-mixed template, but the reality so far is that my section counts change depending on the project, and so I'm doubtful it's going to fix anything.

- Btw, I've only used a small amount of EQ and a "glue" compressor preset on the master bus, nothing at all on the individual tracks (no automation of any kind, too). So the thinness of strings, the boominess, harshness and honkiness all come from the individual instruments as well my completely non-existent mixing skills.

- Problem #3 (maybe only for me) is that the current SWAM key switch system is not very good. There is a short section with orchestral bariolage in this mockup that requires several key switches simultaneously: polyphony, timing accuracy, bow lift, bow pressure, alternative fingering, etc. That's laughable when playing live - nobody will have enough fingers, arms and feet for that.

- What I would love is for Audio Modeling to reserve one or two "special" key switches that can work as combination switches and allow the user to map several CC controls to a single key switch and to freely choose it's MIDI note (not simply the octave).

Anyway. Appreciate any comments on this. I'm not going to touch it anymore, but would be helpful to know if anything is off that I didn't hear. Thanks.
Click to expand...
I really REALLY like that. To me the strings sound very natural (a little bit thin for my taste). Especially the dynamic behaviour shines here. Would you mind to render an unmixed string only file?

What are the other instruments?
 
Major1981 said:
Would you mind to render an unmixed string only file?
Click to expand...
Do you mean a single track rendering of all string instances? I think that's not a problem, but there's going to be an issue with contrabasses, I basically gave up on them and there are only 2 tracks (versus 4 or 5 for each other section).

By the way, here is another piece from the same opera that's only strings (I mean only the beginning, not the voice) . Quite easy to play live and not very long. Is it worth trying to make a mockup, up to the vocal part?



mbishop said:
The room has maybe more tail than my taste
Click to expand...

Yes, I agree. I was using this Ozawa recording for the tempo track of the Introduction mockup but since it was recorded at Tanglewood (AFAIK) it sounded even drier to me than the average pit orchestra of the opera, so maybe I was trying to compensate for that and overdid it.

I used Sample Modeling for French Horns and Tuba, the rest is all Audio Modeling. The common criticism of SM horns (too tame) is very true and the same applies to SWAM horns. For both, adding the metallic resonance is at FF not possible using dynamics alone. I think other CCs have to be involved.


EDIT: OMG!!!! Seiji Ozawa died 3 days ago

1707514209310.png
 
Last edited:
I'm fully working on the track I mentioned a while ago and I'm trying to push these instruments to their limits. I have only a few hours in the evening so it is very energy consuming, I'm so tired...the track I'm writing is complex, it is mostly an action track that resemble a bit the John Williams style both harmonically and orchestration-wise.
I'm basically doing these things all at once:
  • Balancing the template I did in Mir Pro 3D;
  • Trying some complex harmony techniques;
  • Trying some more advanced orchestration techniques;
  • Testing AM String Sections.
About the last point I need some string player here to address a question regarding the bow pressure.

I'm writing all the track with the same bow pressure and I'm gonna tweak it later but I'd like to know how it behaves in real playing. I mean how should I draw the cc to made it sound more alive?
Of course in delicate passages the bow pressure will be more delicate and maybe even flautando in some parts but yesterday I wrote a fingered tremolo on a major 3rd and I struggled quite a bit to understand ho to perform it both in the speed, the attack (aka velocity) and the bow pressure.

I'm still not convinced about how it turned out, the minor/major 2nd trills/tremolos sound better, the larger intervals tremolos are more difficult to build. Basically you write a sustained note and then you wrote many short notes on the other tone of choice. It is a little buried in the track so it doesn't sound bad but I'd like for it to sound good by itself.
I managed to do it with medium velocity to have an attack neither too soft nor too hard. I also assume that the bow pressure is almost constant with the fingered tremolo as it is basically performed during a slow bow stroke. About a standard tremolo or a trill I think it is a bit different as the bow moves quickly.

In short, for the real players:
  1. How the bow pressure works in general?
  2. How the bow pressure works when you perform a tremolo, a trill or a fingered tremolo?
  3. At what speed can the articulation be performed and what note attack should I choose for it to behave realistically?

As a side note I'd like for AM to put some more info about the real playing advanced techniques and how to perform them in their tutorial.
They are already very useful, don't get me wrong. :2thumbs:

Sorry for the wall of text and thank you.
 
DANIELE said:
How the bow pressure works in general?
Click to expand...
I haven't played actively for a long time. You don't consciously use most techniques. But I sat down at my instrument again and observed what I was doing. And in most cases, the bow pressure correlates with the volume. In most cases, you can run expression and pressure in parallel. Expression corresponds roughly to the speed of the bow. It is interesting if you stroke the bow slowly with a lot of pressure close to the fingerboard. It then scratches nicely.

I have discovered something else. Faster articulations in lower registers are better played if you don't play the string directly with a lot of pressure. This can also be achieved by using velocity, which seems to simulate the pressure when the string is first played. Fast and loud staccato passages are best played with bow pressure in the centre, velocity low and expression high.

This lady's videos are a great resource for understanding string instruments:

 
Last edited:
Major1981 said:
I haven't played actively for a long time. You don't consciously use most techniques. But I sat down at my instrument again and observed what I was doing. And in most cases, the bow pressure correlates with the volume. In most cases, you can run expression and pressure in parallel. Expression corresponds roughly to the speed of the bow. It is interesting if you stroke the bow slowly with a lot of pressure close to the fingerboard. It then scratches nicely.

I have discovered something else. Faster articulations in lower registers are better played if you don't play the string directly with a lot of pressure. This can also be achieved by using velocity, which seems to simulate the pressure when the string is first played. Fast and loud staccato passages are best played with bow pressure in the centre, velocity low and expression high.

This lady's videos are a great resource for understanding string instruments:

Click to expand...

Thank you so much. I assumed this but I don't have any string instrument at home and I cannot try it.

About the last statement I also noticed it recently. In the first place I was thinking fast passages (spiccato, staccato etc...) = high bow pressure, then I thinked about arm fatigue and I understood that the player cannot mantain high pressure while doing fast passages. And also it doesn't sound good and this makes sense.

I'm trying to apply the same principle with the fingered tremolo but there you have to deal with the attack and as the bow stays on the string all the time you should have a low velocity, but velocity too low triggers a portamento resulting in an unpleasant glissando sound that doesn't make sense.

The attack of the notes should be pretty fast as one finger stays on the string and the other one moves fast on and off of it. I think here it is a matter of note frequency and it is a matter of understan how quick the finger should move.
I'd like to have some insights from AM devs here. There's nothing about it on the manual.
 
@DANIELE Could I ask you about your load times?

In Cubase 12 and 13, it takes over 20 minutes to fully load a project with 20 instances of SWAM strings and then when an indiviudal instance is opened it hangs on this screen and sometimes it just freezes the entire DAW. I cannot save the project and restart without losing the work. Meanwhile, the Task Manager in Windows says Cubase is running fine.

1707580497126.png

So tired and angry with this mess, but wondering if it's any different in VEPro
 
youngpokie said:
@DANIELE Could I ask you about your load times?

In Cubase 12 and 13, it takes over 20 minutes to fully load a project with 20 instances of SWAM strings and then when an indiviudal instance is opened it hangs on this screen and sometimes it just freezes the entire DAW. I cannot save the project and restart without losing the work. Meanwhile, the Task Manager in Windows says Cubase is running fine.

1707580497126.png

So tired and angry with this mess, but wondering if it's any different in VEPro
Click to expand...
Sure, my actual project uses 10 instances actually and it took me 1 minute and 20 seconds to load. I must say I have the internal space disabled as I use Mir Pro 3D but I don't think it would change a lot in the loading time.

You could have some installation issue, did you try to uninstall the plugin and install it again?
 
youngpokie said:
@DANIELE Could I ask you about your load times?

In Cubase 12 and 13, it takes over 20 minutes to fully load a project with 20 instances of SWAM strings and then when an indiviudal instance is opened it hangs on this screen and sometimes it just freezes the entire DAW. I cannot save the project and restart without losing the work. Meanwhile, the Task Manager in Windows says Cubase is running fine.

1707580497126.png

So tired and angry with this mess, but wondering if it's any different in VEPro
Click to expand...
I strongly suspect the two issues are related. SWAM sections tries to synchronise all the instances as they load and it's using network requests (just to the local machine) to do it. Your OS looks as though it's blocking those messages, possibly intermittently, so the instances may be spending a lot of time retrying network requests before the template fully loads – and then they get blocked on playback, which generates the network-permissions window.

Are you running some kind of network monitor software similar to Little Snitch? If so, you may need to go into the settings to at least allow messages to localhost (though I don't know for sure which port AM is using for these multicast messages so that could be a tech-support request if there's no blanket setting).

The other possibility is that it's a permissions issue: the software needs to be reinstalled (as Daniele suggested) but with admin-level permissions.

It's possible there's a Windows setting for local network access but I don't know Windows well enough to be able to point to which admin command does that.
 
You must log in or register to reply here.
Top Bottom