i was mentioning that i thought the admins should implement ai chat assistance at a fair and democratized price for the site's benefit, so, i plugged this question into Claude 3.7 Sonnet to show off a demo of why i thought this would be a good idea, in general, ... ... ...
Claude:
I'd be happy to help you with your eurorack patching question!
Yes, you can definitely run CV outputs from your Ornament & Crime (O_C) into the Herzlich Labs MMI and then back into the O_C for sample and hold functions to trigger and modulate your Braids. This kind of feedback patching is absolutely possible in a eurorack system.
Here's how you could set this up:
- Use one of your O_C outputs to send CV to the MMI inputs
- Take the CV output from your MMI (which will now be influenced by your gestural control)
- Route that back into an input on your O_C
- Configure one of your O_C apps (likely Hemispheres) to use its S&H functionality
- Use the resulting modulated output to control parameters on your Braids
Your setup is perfectly capable of this kind of patching. The MMI is designed specifically for these kinds of performance-oriented signal manipulations. Since you have two uO_C modules, you could even use one for generating the initial CV and the other for the S&H processing.
The buffered multiples in your rack will also come in handy here, as you might want to split some of these signals to multiple destinations.
As you mentioned, the ZOIA could accomplish this as well - it's extremely flexible for this kind of signal routing and processing. But using the O_C as you described would work great and give you a more hands-on, hardware-focused approach to the patch.
Is there anything specific about this patch flow you're having trouble with?
...
(and if this model was further fine tuned and continuously retrained on more new and annotated data, pretty soon it would know more than anyone on this forum.)
((peace ))