This AI Guitar Pedal Allows Custom Effects Creation
It’s hard to say if there was a demand for an AI guitar pedal, but it was only a matter of time before someone took the plunge. Enter Polyend, a respected name in the music gear scene known for crafting unique and niche devices. With a history of creating grooveboxes and multi-effect pedals, there was hope that if anyone could nail an AI-driven effect pedal, it would be them.
The Polyend Endless is a $299 programmable guitar pedal powered by an ARM processor. It works in tandem with Playground, a suite of interconnected AI agents that transform any text prompt into a usable guitar effect. If you have a specific idea in mind, you don’t have to wait for a manufacturer to create that pedal; you can simply prompt it. Whether you’re after a rare combination of effects like a ring modulator and auto-wah, this pedal aims to deliver. While I’m not entirely convinced this is what guitarists have been craving, it’s a commendable first step in merging effect pedals with AI technology.
To clarify, the AI isn’t housed within the pedal itself. Instead, Polyend has trained a custom language model to generate effects that you can load onto the pedal. You can also create effects using C++, but most users will likely download existing free “Plates” (as Polyend calls them) from the community site or generate them through Playground. For those who want a physical faceplate to match a downloaded effect, it’s available for $20.
Currently, the Plates gallery features around 60 effects, primarily developed by Polyend. These range from simple saturators to tape loop simulators and even guitar synths, with some self-playing drum machines thrown in for good measure. Standout effects include Grunt (a lo-fi octave down), Infinite Hall reverb, and Tessera (a granular pitch-shifting reverb). Stardust is another highlight, offering a massive-sounding granular delay, reverb, and tremolo combo that’s hard to find elsewhere.
Polyend is also opening the gallery for third-party contributions, allowing users to create effects in Playground and submit them for consideration.
Playground is likely the main draw for those interested in the Endless. It serves as a web interface for several AI agents that work together, trained on Polyend’s effects library. These AI components interpret prompts, select effect algorithms, generate code, and validate it to ensure it runs smoothly without causing any audio mishaps.
If you’ve used a chatbot, Playground will feel familiar. You describe the effect you want and its controls (you have three knobs and short/long presses on the footswitch), and it will return a few options to turn your idea into something tangible.
Typically, you’ll receive three options. You can choose one and let Playground handle the rest, or you can tweak your idea before it starts generating code, which is crucial since generating effects costs tokens. The pedal comes with 2,000 tokens, and you can purchase more at $20 for an additional 2,000. This should cover a few effects, but the system can be a bit unclear; the complexity of an effect and the number of iterations needed will determine the token cost. A simple fuzz might only set you back 20 tokens, while a granular looper with rhythmic glitches could run you 500.
What tends to drain your tokens isn’t the initial generation but the iterations needed to refine your desired effect. Trying to get Playground to understand the right resonance levels for a bandpass filtered delay or the amount of ring modulation for a gritty fuzz can be a challenge.
Polyend provided me with 10,000 tokens ($100 worth) for this review. I ended up using just over 3,500 tokens to create three effects I liked, along with a few that didn’t hit the mark. If I were paying for tokens myself, I might have been less inclined to keep experimenting. Additionally, Playground can be slow to generate code; depending on the effect’s complexity, it could take anywhere from five to over ten minutes each time. Often, I found myself losing patience after several attempts if I wasn’t getting the results I wanted.
The effect that gave me the most trouble was one the chatbot dubbed “Resonant Taps.” I asked for “a clean multi-tap digital delay with narrow, highly resonant bandpass filters and subtle modulation.” The first version lacked control over the delay feedback, and when combined with the resonant filters, it was on the brink of self-oscillation—definitely not pleasant. Replacing the resonance control with a feedback knob didn’t improve things much.
After six attempts, I decided to describe the sound I wanted instead of specifying the effect. I requested “gently modulating resonant pings that follow the pitch of my guitar, almost like a string section.” What I got sounded more like a ’70s synth having a meltdown. It was entertaining, but not what I was after. After spending about $7 worth of tokens, I shifted my focus to something more straightforward, like a fuzzy, glitchy CD-skipping effect.
Even that required some coaxing. My initial prompt for “random stuttering and glitching” with “tap tempo” and a blend knob for a ring modulator didn’t deliver. The stutters felt more like machine gun fire than rhythmic. It took six generations and specifically asking for the stutters to sync to 16th, 8th, or quarter notes to finally achieve what I wanted. Despite requesting an increase in the ring mod level each time, it still felt a bit too subtle. However, I was pleased with the final result—it’s wonderfully chaotic and just dirty enough.
There are a few quirks to note. Endless can only load one effect at a time. Loading new effects is straightforward; you connect it via USB, and it appears as an external drive for file transfers. However, even though the pedal reboots automatically after loading a new effect, many don’t function properly until you manually power cycle it by unplugging it. You might also need to eject Endless and reconnect it to your computer before loading a new effect, which can slow down the iteration process.
The sounds I’m trying to create are complex, with many variables at play. But that’s part of the appeal. There are countless digital delay and fuzz pedals that can outperform the Endless in those areas. If you’re after standard effects, you’d be better off building a collection of dedicated pedals rather than relying on AI. The main reason to get Endless is to explore and create effects that don’t exist yet.
It’s essential to manage your expectations. The effects generated by Polyend’s Playground won’t rival the meticulous programming of professional audio developers or dedicated hobbyists. Endless excels at quickly experimenting with ideas that you can further develop, either by programming directly in C++ for the Endless or using other platforms like Max MSP or Pure Data. It’s unfortunate that Polyend doesn’t provide access to the raw code of effects created in Playground for further tweaking.
You might find it more beneficial to explore these concepts on a modular guitar pedal like the Poly Effects Beebo, Empress Effects ZOIA, or Eventide H90, which allow you to assemble effects patches from building blocks. While these devices are pricier—$449, $549, and $899 respectively—they come with active communities, stable firmware, and high-quality effects. I’d much rather spend an hour crafting an effect on my ZOIA than wrestling with an AI.
Lastly, the AI aspect may deter some users. However, Polyend appears to be making a genuine effort to use AI responsibly. They build code solely from effects developed in-house or from open-source components and are working to minimize their environmental impact. Some Playground functions utilize external AI agents via API, which is unavoidable. Still, founder Piotr Raczyński mentions that the Playground servers are on-site, not in a third-party data center, and are “almost 100% energy self-sufficient thanks to solar power and heat pumps.” It’s essentially the “free-range, pasture-raised” version of LLMs.
If you’re staunchly anti-AI, this won’t change your perspective on Endless. This might explain why Polyend avoided using the term “AI” during its announcement. However, if you’re more open to AI or have embraced LLMs, this could alleviate any concerns. The primary benefit of the Polyend Endless is its affordability and accessibility for creating custom effects. If a vibrant community develops around it, similar to the one for the ZOIA, it could become a fantastic source for experimental effects. But given the music community’s general skepticism towards AI, that’s a significant “if.”