Ferment — real-time sound processor with instrument profiling & AI presets — free license for beta testers

[Ciderka Labs]

AUTO-ADMIN: Non-MP3, WAV, OGG, SoundCloud, YouTube, Vimeo, Twitter and Facebook links in this post have been protected automatically. Once the member reaches 5 posts the links will function as normal.

Hi,

I'm Petr from Ciderka Labs (https://labs.ciderka.com/). I've been building Ferment (https://labs.ciderka.com/ferment/)— a real-time sound processor for musicians who need to quickly find a sound at rehearsal, fine-tune it, and then use it on stage. I'm looking for beta testers and offering a free license in return.




If you're interested: labs.ciderka.com/beta (http://labs.ciderka.com/beta)

---

Why Ferment exists

I play double bass and needed an effect processor that comes from the synth world — deep parameter control, modulation, sound shaping — but actually works for acoustic instruments like upright bass, violin, or voice. Guitar pedals and stomp boxes don't cut it when your source isn't a guitar. And most plugin effects give you a preset browser and hope for the best.

Ferment takes a different approach: use AI to sketch a sound quickly, then dial it in precisely with 92 exposed parameters. Save it, recall it on stage. The whole workflow is built around getting from "I have an idea" to "I have a usable preset" as fast as possible.

It's a real-time sound processor with one coherent signal path. Here's how it works:

• Machine — the full signal chain: waveshaping, chorus, octaver, delay, spatial, character, dual LFOs, envelope follower, post EQ. Everything on one screen, everything MIDI-mappable. No tabs, no hidden menus — you see all 92 parameters at once and tweak during performance.
• Material — this is the part that doesn't exist elsewhere. Play a few notes on your instrument and Ferment analyzes your specific sound — spectral content, dynamics, harmonics, attack. Then it optimizes the effect chain for that source. A violin and a guitar don't need the same processing, and Material knows the difference.
• Magic — describe a sound in words ("dark, rhythmic, with metallic texture") and AI generates a full preset across all 92 parameters. You hear three options, pick one, refine. Every generated value is visible and editable — if you don't like what AI did, just change it. And if you don't want AI at all, don't use it — the rest of Ferment works completely offline.
• Library — browse and preview presets with one click before loading.

The short version: Ferment is for musicians who want to explore sound on stage, with their own instrument, instead of scrolling through presets that were designed for something else.

---

Who should test this

• Musicians who run a laptop or desktop on stage — processing through a DAW during gigs or rehearsals
• Players of acoustic instruments (strings, voice, guitar, wind…) who've been frustrated by effects that don't really fit their sound
• Sound designers curious about AI-assisted preset creation and willing to push it until it breaks
• Anyone who can tell me honestly what works and what doesn't

 

---

What you get

A free Ferment license. Full app, all features, no limitations. You test the beta, you keep the product when it launches. That's the deal.

---

Specs

• Formats: VST3, AU, CLAP
• Platforms: macOS 10.15+ / Windows 10+ (64-bit)
• Version: 0.9.5 Beta

 

---

If you're interested: labs.ciderka.com/beta (https://labs.ciderka.com/beta)

Or write me at labs@ciderka.com (mailto:labs@ciderka.com)

A few testers have already jumped in and their feedback is shaping the next build. Thanks for reading.

Petr / Ciderka Labs (https://labs.ciderka.com)

[CinningBao]

There is a mild distaste for plugins like this here. Any mention of taking skills away from the musician and farming it off to an 'intelligence' tends to leave a sour taste in many of our mouths. Not everyone, mind you, but a lot of us.

This thread probably isn't the space for that discussion so let's see where the conversation goes.

I beta test, a lot, so most invitations to things like this are difficult to ignore. I don't know why but I have no interest in asking an AI to build me a chain of effects. I've turned away from a number of these projects recently and I don't regret it.

I can totally understand the path of thought which leads a developer to create these kinds of tools. But developers solve problems; they are thinking creatively but not in an effort to express their feelings and thoughts; it's to creatively solve the maths problems they find themselves in.

Ii pains me to see people's potential to learn stuff is being removed by tools like this. I mean, regardless of what the (eg) Material function is, or how it works, the user isn't learning by making mistakes (as we usually would) and it just being offered things the user (likely) won't know if it's any better or not; they are just having to 'trust' the AI.

Sorry, I had to say it. I have nothing against you as a human! But any efforts to remove people's ability to make an error with nothing but their own decisions, is wrong. I'm sure there's an audience for it, but it isn't me.

[jules99]

I wish you all the best with your vibe-coded plugin, but do reconsider the UI. It's just dark, hard to read, a seah of knobs. It screams AI slop.

If you actually want to appeal to musicians using this on stage, I'd work with exactly those and figure out a workflow and UI that are much more intuitive. Because prompting a signal chain only for having a sea of similar-looking knobs to be even more confusing might not be the road to take.

[Ciderka Labs]

Thanks to both of you — really appreciate the honest feedback.

Some context: I've been playing for over 20 years, mostly guitar and bass, and lately more double bass. When I started exploring ambient textures and arco techniques, classic effects I'd used for years just didn't work well acoustically. That's where this started — a specific personal problem, not a desire to automate creativity.

The idea is simple: one place to quickly sketch a sound with the band, then refine it. I am proud of Harmonic shaping (not EQ), spatial effects, controllable tremolo — and then take that same sound to a live gig. I use MainStage and built my own 3-axis expression pedal, so I can shape the sound with my foot while playing with more effects. For me, one plugin actually reduces cognitive load on stage compared to chaining five or ten effects — and helps with latency too, which matters live.

I play with a group of musicians around me — violin, cellos, synths, jazz, classical, some of them play in philharmonic orchestras, some in bands, and we're all experimenting with mixing live electronics and acoustic instruments. Technically we're all amateurs when it comes to audio software. But that's exactly the point: the workflow of sketch → see what you're adjusting → refine it actually helps people learn. A friend and I are currently trying to turn a single violin with an expression pedal into something like a string section — usable live, with respect to the performance energy and original sound. Chorus or harmonization combination, sound character, and how do you quickly dial it in on a specific PA before you go on? That's the kind of problem this is trying to solve. And keep to having fun

On the AI part — I'd rather be upfront about it than pretend it's not there. AI role here is not the sound generator - it a translator. It helps me sketch faster, and honestly it's helped me learn — when I hear something I like, I can dig into how it's built and tune it myself. I never use any AI output as a final sound. And I have no interest in AI having control over the code or the audio.

I work in software development — 15 years, from developer to manager — so I have a lot of respect for human craft. Ciderka Labs is just for fun and to share, because I love playing a performing live (each performance is different, full of energy and mistakes , but this is, what I love)

On the UI — fair point, and I'll think about it. Honestly I put most of my energy into the functions rather than the visuals. I also have a bit philosophical issue with software trying to look like physical hardware — it's still software. On stage I actually prefer a dark, clean layout — easier to read under lights. And writing frontend in C++ is its own kind of pain, so clarity won over beauty. And what comes with modern UX - trying to make thing "as simple as possible", that every body have to understand all in seconds, makes them sometimes limited (Also like synth vs. guitar effect - works on the same background, effect is optimized and easier to understand, but works for guitar only...) . But the comment is noted and thanks for that.

Primarily I just wanted to share something I've been building and using for a while. Happy if more people want to get involved and give feedback.

[Ciderka Labs]

AUTO-ADMIN: Non-MP3, WAV, OGG, SoundCloud, YouTube, Vimeo, Twitter and Facebook links in this post have been protected automatically. Once the member reaches 5 posts the links will function as normal.

Quick update — I'm happy to say a few people have signed up for testing and I've started receiving feedback. Based on what's coming in, there may be some bigger changes between v0.9 and v1.0 than I originally planned.

If anyone else would like to join the test, now is the ideal time — I'd like to close this iteration within the next few weeks. After that, I'll be heads-down working on the changes before the v1.0 release.
To everyone already involved — thank you.

If you have any questions or feedback, feel free to reach out via private message here or email me at labs@ciderka.com (mailto:labs@ciderka.com).

[Zeisner]

I don't know why but I have no interest in asking an AI to build me a chain of effects.

Would it be different if you trained the AI yourself with your own data?

[Ciderka Labs]

I think we've drifted a bit here, so a quick clarification: Ferment is not an AI-driven plugin. The DSP and preset logic are all built into the module itself. Gemini's role is minor — mostly a translation layer. The "Magic" feature is just a shortcut to help new users get oriented faster, nothing more.

The "AI slop" comment surprised me more, honestly — especially from two screenshots. There's a lot of careful work under the hood.

Also want to thanks here to CinningBao, who shared some great insights with me privately — really appreciated that perspective.

As for a fully custom trained model: I explored it, but the data volume required to make it worthwhile is a real barrier. Direct LLM use for preset generation also produced too many errors, which is why I landed on AI augmentation rather than the core engine.