geddonn

Hi guys, so I've been working on a modular audio performance environment for a while now, mostly in planning but also trying out a variety of methods and feasibility tests.

Until today the project was going to consist of a global patch who used js scripting to generate patch windows which would then load whichever module I wanted (reverb, echo, grain etc.).

I chose to use separate windows for each process first to make things visibly clear but I also found that splitting the DSP across multiple windows meant I could squeeze more out of my machine.

The issues I had with this method was first that scripting in JS can be a bit buggy, and the initial groundwork was taking a lot of time.

Then today I came up with a new approach. I've attached a project that illustrates this approach.

What I thought of doing was using a single multi-thread poly for all DSP and have each voice with a nested poly that I could refer to different modules.

In the attached project I've just got a couple of test tone generators to show what I mean.

Now, I guess my question is, which method would be most stable/efficient?

Will I run into any problems with nesting polys inside polys inside polys (assuming each audio module can have it's own nested polys too)?

Is the multi window technique more efficient than multithread poly?

Until today the project was going to consist of a global patch who used js scripting to generate patch windows which would then load whichever module I wanted (reverb, echo, grain etc.).
I chose to use separate windows for each process first to make things visibly clear but I also found that splitting the DSP across multiple windows meant I could squeeze more out of my machine.
The issues I had with this method was first that scripting in JS can be a bit buggy, and the initial groundwork was taking a lot of time.

Then today I came up with a new approach. I've attached a project that illustrates this approach.
What I thought of doing was using a single multi-thread poly for all DSP and have each voice with a nested poly that I could refer to different modules.
In the attached project I've just got a couple of test tone generators to show what I mean. 

Now, I guess my question is, which method would be most stable/efficient?
Will I run into any problems with nesting polys inside polys inside polys (assuming each audio module can have it's own nested polys too)?
Is the multi window technique more efficient than multithread poly?

Any feedback and advice is very welcome.


looking-for-some-advice-on-which-method-to-persue

nesting polys can at some point let you loose overview about your system and it means long load times. but in theory there is no limit afaik.

I think I'm going to go down the nested poly~ road. I've also noticed that using [parallel 1] on a poly~ incurs a DSP overhead even when everything is muted so having one of these as the container for all DSP in the patch should keep that overhead to a minimum.

I saw that there was a point where people were having problems using send~ and receive~ in poly~. I haven't come across any issues myself yet so if anyone wants to warn me on any issues they have come into using them that would be most welcome.

I think I'm going to go down the nested poly~ road. I've also noticed that using [parallel 1] on a poly~ incurs a DSP overhead even when everything is muted so having one of these as the container for all DSP in the patch should keep that overhead to a minimum. 

I saw that there was a point where people were having problems using send~ and receive~ in poly~. I haven't come across any issues myself yet so if anyone wants to warn me on any issues they have come into using them that would be most welcome.


Looking for some advice on which method to pursue - Large Modular DSP setup.