> Side question for the folks at CCRMA: I'm also planning to do one for ChucK but are people at CCRMA working on one?
Not that I'm aware of, but I am not active in that group. Ge Wang is the main guy, as you probably know. On Tue, Feb 17, 2026 at 9:51 AM M. Edward (Ed) Borasky < [email protected]> wrote: > This is excellent news!! I'm currently in the process of building up a > local-model AI tool collection for music creation that's container-based > and a Faust container was something I had on my to-do list. So I'll be > testing this out soon. Side question for the folks at CCRMA: I'm also > planning to do one for ChucK but are people at CCRMA working on one? Since > the Faust image exists I'll probably just add ChucK to it now. > > I'm exclusively using local models; I have a five year old NVIDIA GeForce > RTX 3090 with 24 GB of VRAM that's been sitting mostly idle for the time > since I got it. It's not much of a space heater but it's smarter than the > other ones. ;-) > > -- > How many people can stand on the shoulders of a giant before the giant > collapses? > > Sent with Proton Mail <https://proton.me/mail/home> secure email. > > On Tuesday, February 17th, 2026 at 5:07 AM, yann orlarey < > [email protected]> wrote: > > Hi Julius, Michael, and Stephane, > > As mentioned by Stéphane, we are currently developing this kind of > environment, in particular FaustForge: > > https://github.com/orlarey/faustforge > > The core objective is to reduce friction between AI-assisted code > generation and actual DSP validation: write or refine Faust code, hear > results, inspect generated artifacts, and iterate quickly : > > - Fully Dockerized: setup is reproducible and isolated, so you can run the > same environment on any machine with minimal host dependencies. > - Live workspace mode: `.dsp` files are watched automatically; creating or > editing a file immediately updates the active live session. > - Continuous recompilation: changes are picked up in near real-time while > staying in the same working view. > - Multi-view synchronization: Regular UI, Orbit UI, and remote control all > reflect the DSP parameter state consistently. > - C++ inspection workflow: in the C++ view, you can now choose Faust > compilation flags using reusable presets, making it easy to compare > generated code under different compiler options. > > AI-friendly documentation is key in this type of project... > > Cheers, > > Yann > > Le mar. 17 févr. 2026 à 14:00, Stéphane Letz <[email protected]> a écrit : > >> We already have: https://github.com/grame-cncm/faustbrowser-mcp (my >> try..), and Yann more sophisticated version: >> https://github.com/orlarey/faustforge >> >> More to come probably... >> >> Stéphane >> >> >> > Le 17 févr. 2026 à 13:29, Michael Chourdakis <[email protected]> a >> écrit : >> > >> > I'm currently experimenting with Copilot to see If I can create an >> AI-driven faust editor >> > MICHAEL CHOURDAKIS, PhD >> > DSP & ARTIFICIAL INTELLIGENSE >> > SOFTWARE ENGINEER >> > www.turbo-play.com >> > >> > >> > >> > >> > Στις Τρί 17 Φεβ 2026 στις 2:21 μ.μ., ο/η Julius Smith < >> [email protected]> έγραψε: >> > Hi All, >> > >> > Interesting exchange with Claude just now: >> > >> > Me: ❯ Do you already know Faust well enough to write it, or should we >> make a Claude Code "skill" for more Faust expertise? >> > >> > Claude Opus 4.6 [Thinking]: >> > >> > ⏺ I know Faust reasonably well from training -- the composition >> operators, standard libraries, how compute() loops are structured in >> generated C++, >> > etc. But I'd be more reliable with a skill, for two reasons: >> > >> > 1. Subtle semantics -- The interaction between ~ feedback, <: split >> fan-out rules, and partial application can trip me up on non-trivial >> programs. >> > Having the primer in context would catch those. >> > >> > 2. Project-specific Faust→C++ patterns -- Your faust_cpp/ directory has >> a specific style: JGFaustLib.h utilities, hand-optimized biquads, the way >> > mydsp::compute() structure maps to native classes. A skill could load >> both the language primer and your project conventions so translations stay >> > consistent with what's already in faust_cpp/. >> > >> > My recommendation: make a skill. It would be lightweight -- just load >> FAUST_PRIMER.md plus a project-conventions section drawn from >> > faust/JGFaustLib.h and an existing example like JGCabinet. Then any >> time you ask me to write Faust, /faust would put the right context in the >> window. >> > >> > [I attach my FAUST_PRIMER.md, ~12.8kB, written by Claude from my >> teaching tutorial, but note that it is NOT COMPLETE! It's just what I >> regularly use. A better primer would condense the Faust Quick Reference >> manual. (Claude, Codex, or Gemini would all do a good job on that).] >> > >> > Cheers, >> > Julius >> > >> > P.S. I've written ~0 code in 2026, but I've managed a LOT >> > >> > -- >> > AI has taken my job, but only I know what it is. >> > _______________________________________________ >> > Faudiostream-devel mailing list >> > [email protected] >> > https://lists.sourceforge.net/lists/listinfo/faudiostream-devel >> > _______________________________________________ >> > Faudiostream-devel mailing list >> > [email protected] >> > https://lists.sourceforge.net/lists/listinfo/faudiostream-devel >> >> >> >> _______________________________________________ >> Faudiostream-devel mailing list >> [email protected] >> https://lists.sourceforge.net/lists/listinfo/faudiostream-devel >> > > _______________________________________________ > Faudiostream-devel mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/faudiostream-devel > -- AI has taken my job, but only I know what it is.
_______________________________________________ Faudiostream-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/faudiostream-devel
