Toads For Supper

How toads actually work in practice and why your patches sound muddy

Toads for Supper is a Max for Live device by Peter Kuhlmann designed for granular and spectral processing of audio. You load it into Ableton Live, drop an audio track underneath it, and route your source material through. That's the basic signal path. The device takes short bursts of audio data — grains — and rearranges them in time and pitch independently of each other. It became popular in ambient and experimental electronic music circles around 2018-2019 and remained useful through the early 2020s. It's not a one-click solution for any genre. The controls require deliberate adjustment to sound intentional rather than messy.

What you need before you start

You need Ableton Live 10 or later with Max for Live installed. The device itself is available as a .amxd file. Peter Kuhlmann's website and the Max for Live library host it. Some users obtained early versions through community forums and GitHub repositories. I found the most stable version in the Max for Live bundled devices directory after Live 11 was released. Your source audio matters more than you might expect. Toads for Supper performs best with material that has clear transients and dynamic variation. A clean electric guitar pickup, field recordings with environmental noise, or a vocal phrase with natural articulation all work well. Heavily compressed drums and synthesized pads with long attack times struggle because the grains have nothing to latch onto.

The controls and what they actually do

The main section of the device contains a sample buffer, a playback position slider, and a grain density control. Below that are pitch shift parameters and a re-trigger rate. There's also a spectral freeze mode that I'll get to. Grain size is the first thing most people get wrong. By default the grains sit somewhere around 30 to 50 milliseconds. If you leave it there with high density, you get a wash of texture that loses all rhythmic definition. Dropping the grain size to 10 to 20 milliseconds gives you more articulate material that can still be pitched without sounding formant-distorted. I usually start at 15ms and adjust from there depending on the source. The re-trigger rate determines how often new grains are started. At low values it sounds like arpeggiated particles. At high values it becomes a continuous texture. This is where most people's patches go wrong — they crank the re-trigger too high and then wonder why the result sounds like white noise. Start at 1/16th note and work your way up in tempo divisions.

Practical workflow for building a usable patch

Here's the process I follow when starting a new patch. Load an audio clip into Ableton. Place Toads for Supper as the first effect on that track. Arm the track and play or loop a section of your source material. Adjust the buffer length so it captures roughly 2 to 4 seconds of audio. Set the grain size to 15ms. Set the re-trigger to 1/8th notes. Move the playback position slider slowly while listening for sections where the grain structure sounds interesting. Once you find a spot that works, save it as a preset. Then layer a second instance of the device on a separate track using a different portion of the same buffer. This gives you two independent granular voices from the same source material. The two tracks can be panned differently and given slightly different pitch offsets for width. Don't skip the output stage. Send the Toads output through a reverb. The granular texture alone often sounds dry and thin because you've broken the original phase relationships apart. A medium-length hall reverb with a decay around 2 to 3 seconds restores cohesion. I use a plate algorithm sometimes for more metallic character when the source material is warm.

Common Toads For Supper pitfalls and how to avoid them

The biggest issue I encountered was grain clustering at the start of the buffer. When the playback position sits near the beginning of the audio, the grains grab onto transients and produce an overwhelming burst of texture. Moving the position slider just 5 percent further into the buffer cleared this up immediately. There's no visual indicator of transient density in the device, so you learn to listen for it rather than see it. Another problem is pitch shifting with large grain sizes. If you shift pitch by more than a fifth with grains above 40ms, the formant smearing becomes noticeable on vocal sources. I learned this the hard way when a vocal sample turned into unintelligible sludge at +7 semitones. Dropping the grain size below 20ms fixed it without losing the pitch shift character. CPU usage scales poorly with grain count. Running three instances of Toads for Supper simultaneously on moderate hardware will push your CPU into the red if the grain sizes are small and the re-trigger rates are high. I typically run two instances max at 44.1kHz sample rate. If I need more voices, I bounce the output to audio and work with the rendered material instead.

Advanced technique: spectral freeze for ambient pads

The spectral freeze mode is the feature most people miss. When engaged, it captures the current spectral content of the audio and holds it. You can then sweep the playback position through the buffer without the texture changing — the frequency content stays locked while the spatial position moves. This produces sustained pad-like sounds from source material that has no sustained qualities. A plucked string, a percussive hit, a spoken word fragment — anything becomes a drone. The trick is combining freeze with gentle pitch modulation. Add a slow LFO to the pitch control at a depth of 2 to 5 cents. The frozen spectrum will shimmer without losing its character. This creates ambient beds that retain the harmonic signature of the original source while completely removing its rhythmic content.

Alternatives when Toads isn't the right tool

If you're working with time-sensitive material that needs to stay rhythmic, granular synthesis isn't always the answer. Samples Delay or the built-in Ableton Grain Delays handle rhythmic granular effects with less CPU cost. For pure spectral freezing without the granular complexity, Portal by Output does a cleaner job with lower latency, though it's not free. If you need convolution-based texturing instead, IR 16 by Audiothing gives you different results with the same source material and far less CPU overhead. Toads for Supper has specific strengths and they're real. It handles organic source material better than most granular devices because its grain envelope shaping is more natural. But it's not a general-purpose tool. If your workflow involves fast iteration and you need instant results, it will slow you down. The interface isn't the most intuitive either — some parameters require clicking and dragging rather than having clear slider ranges. I spent the first two weeks just learning which controls had the most impact on the sound. The device remains functional in current Ableton versions. Some users report slight buffer management issues when using it with Sample Rate higher than 48kHz, and I've seen crashes when the grain density reaches maximum with a very long buffer. Keep your project sample rate at 44.1kHz for stability. If you need 96kHz material, process at 44.1kHz and resample afterward.

There's no single correct way to use Toads for Supper. The controls interact in non-linear ways, and the results depend heavily on what you put into it. The best approach is to treat it as a texture generator rather than a performance instrument. Build your patches deliberately, save presets for configurations that work, and move on to mixing before you try to tweak endlessly. The device rewards patience but punishes aimless knob twisting.