Is there a modular audio environment like vcv/reaktor/max, where I could plug stuff together by typing text, instrad of mousing pips? I honestly tried to get into pd, but it felt like typing book using the character map.
Saw a good talk at PyCon a couple of years ago about FoxDot, which is a wrapper around SuperCollider. Bit foggy on the details now, but it seemed like a good place to start.
Yes, most audio programming languages allow you to create DSP graphs by connecting nodes together. This is often done with some kind of pipe operator (for example, ChucK has the "chuck operator", Faust has a bunch of operators for connecting batches of nodes in different ways).
My favorite approach is in Sporth: because it's concatenative, you don't need any operator, you just type the things you want to connect. Shameless plug, I made a playground for it: https://audiomasher.org/browse
I'm the author of Scheme for Max and Scheme for Pd, open source (sibling) projects for scripting, live coding, and doing algorithmic music and sequencing in Max/MSP, Ableton Live, and Pure Data in S7 Scheme. I work in them using an OSC bridge from Vim so that I have a REPL in my editor controlling the environment entirely with lisp.
Supercollider sounds like what you're after. You can even use vim :)
https://github.com/pjagielski/awesome-live-coding-music here's a list of related stuff