How are Exp Filtertable and Wavefield Different?

One of the most common questions I get is: how are Exp Filtertable and Wavefield different? I’ll give you the short answer, and then we’ll dive into the DSP so you come away understanding a lot more about different spectral processing techniques.

Exp Filtertable and Wavefield are both wavetable-based spectral filters. If you’ve used a wavetable synth, you already know that a wavetable is just a stack of shapes, and you can glide, or morph, from one shape to the next. These filter plugins take that same idea and point the shapes at a filter instead of an oscillator that generates sound.

And when I say spectral filter, I just mean one that shapes the whole frequency spectrum at once. That’s unlike regular filters and EQs, which have a cutoff knob and fixed frequency bands. Ordinary shelf, bell, and notch filters have shapes that are each defined by a single equation and are limited in flexibility and cannot produce arbitrary shapes. Spectral filters can. And that’s something both plugins share.

On the surface they may look like they do the same job. But under the hood, each one does something completely different with the wavetable shape. And that single decision is what changes how they sound, what they’re good at, and the kind of processing you can get out of them.

The Short Answer: Exp Filtertable Convolves, Wavefield Maps Directly

So here’s the short answer: Exp Filtertable convolves with the wavetable, meaning it processes the wavetable shape as a time domain object. The shape itself isn’t the frequency response you hear; the plugin has to transform it to find out what frequencies come out. Wavefield instead maps the wavetable shape directly as a frequency domain curve. Those are two very different things and require completely different processing methods. Let’s get into the DSP.

The Problem Every Spectral Filter Has to Solve: FFT, STFT, and Block Artifacts

Spectral filters have a particular problem to solve. I’m going to show you the two main solutions, and each choice determines what the filter can do.

To shape sound by frequency, a plugin first has to see the sound as frequencies, rather than individual samples over time. The standard way is to take a quick frequency snapshot of the audio, many times a second. The math that does that snapshot is the FFT, the Fast Fourier Transform. You’ve probably seen its output a thousand times: it’s what’s running behind any spectrum analyzer. It takes a slice of audio, tells you how much of each frequency is in it, and thousands of frequency domain slices can be processed individually. Then the inverse FFT transform can turn that back into samples over time afterward. Keep the FFT in the back of your mind, because it shows up again in both plugins.

Because a sound is always changing, you can’t take just one snapshot. You have to slice the audio into short, overlapping chunks and snapshot each one. One slicing method is the short-time Fourier transform, or STFT.

Here’s where the problem I mentioned earlier comes in, and it’s entirely in whether those tiles get either processed individually and glued back together, or not. The spot where two chunks meet is a seam. If the gluing isn’t careful, the seam shows up in the sound. One way it goes wrong: filtering can push a little energy past the edge of a chunk, and instead of carrying over, it wraps back to the front of that same chunk as a faint artifact that was never in your audio.

It gets worse when the filter is moving. Two neighboring chunks get shaped slightly differently, so where they overlap they disagree. And that disagreement reads as a ripple or a roughness.

This is exactly what Exp Filtertable points to on their own site. They say they use convolution because misaligned STFT blocks can degrade the sound. And they’re right. It’s a real problem. The fundamental difference between Exp Filtertable and Wavefield is whether that problem gets solved or avoided.

Convolution: How Exp Filtertable Sidesteps Spectral Artifacts

Here’s the trick Exp Filtertable uses to sidestep those spectral artifacts completely. And you’ve almost certainly heard of it before.

Their fix is convolution. And you’ve probably already met this concept. It’s the technology behind convolution reverbs, the ones that capture the exact sound of a real cathedral or studio and let you play your audio like it’s inside that space. That’s the most well-known example, so let’s use it to pin down what convolution actually does.

For every single moment of your audio, convolution takes the input and adds a character to it according to a time-domain fingerprint to produce the output. It does that continuously, sample by sample, rolling smoothly across the whole signal. There’s no point where it stops, processes a chunk in isolation, and starts again. The filtering flows straight through, so there’s never a seam where two separately-processed pieces have to meet.

Exp Filtertable is not a reverb, but it’s using the same operation, and the fingerprint isn’t a room. It’s the wavetable shape, acting as a filter. Same machinery, totally different job. And that fingerprint has a name you’ve probably seen too: the impulse response, or IR. If you’ve ever loaded an IR into a reverb or a cab sim, that’s the same thing: a little capture that carries the whole character of what you’re applying. Here, the wavetable shape becomes that IR, but remember, it’s in the time domain, not the frequency domain. The spectral curve created by the wavetable as an IR is something you kind of find out after processing.

Because the filtering rolls continuously across the signal instead of being assembled from separate chunks, the seam problem from a minute ago simply never appears. Nothing has to be stitched, so nothing can be misaligned. It’s a genuinely clever, completely legitimate solution.

But look closely at what it’s solving: a problem that only exists in STFT if the chunks aren’t glued back together correctly in the first place. The way Exp explains “misaligned blocks” might give one the idea that artifacts are an inevitable cost of doing STFT. Sure, it takes a bit of work to properly reconstruct the blocks, but it is very possible, and Wavefield is proof of that. This fact, and the resulting freedom of processing unlocked by directly using the wavetable frames in the frequency domain, is the core of how Exp Filtertable and Wavefield are different.

Convolution Doesn’t Even Avoid the FFT

But plot twist: Filtertable and Wavefield both split the audio into chunks. So why does one have to reconstruct and the other doesn’t?

This is where it gets interesting. There is another method of slicing audio into blocks and using FFT besides STFT that I haven’t covered yet. I just told you convolution rolls along continuously, and conceptually it does. But doing that math directly, sample by sample, is far too slow for real time. So in practice, plugins use a shortcut called fast convolution. And that shortcut chops the audio into chunks and runs the FFT on each one. Just like the blocks and FFT in the STFT method.

So now you’re probably thinking: wait, if convolution also works in chunks, doesn’t it have the same block-seam problem? And here’s the core difference. Fast convolution chops into chunks, but the algorithm itself is built so those chunks always recombine into the exact right answer. Any overflow that runs past the edge of one chunk is able to be carried into the next and added back in easily. It’s exact by construction. The chunk boundaries are real, but they’re invisible to the process, because the math handles them perfectly.

The raw STFT method we saw earlier uses the same chunks and the same FFT, but it shapes each chunk a little independently and just overlaps them back together, with no guarantee they line up. That’s the difference. Same blocks, but one process reconstructs exactly by nature and one only approximately. So the artifacts were never caused by the FFT, and never caused by working in chunks. Both approaches do both. Artifacts come from one thing only: chunks reconstructed approximately instead of exactly.

And that’s the whole reveal. If chunk artifacts come from sloppy reconstruction, not from chunks themselves, then you don’t actually need convolution to avoid them. Reconstruct everything from STFT correctly, and it’s just as clean. Convolution gets there one way; correct reconstruction gets there another. Same clean destination, but very different results in processing.

How Wavefield Is Different from Exp Filtertable

This is the basic difference between Exp Filtertable and Wavefield. And it comes down to one decision about what a wavetable shape even means.

Each plugin uses wavetables. The difference is what the plugin does with a frame. And a frame is just one snapshot shape from that table. Morphing, again, is only sliding smoothly from one frame to the next, which is how these plugins put the sound in motion.

In a convolution tool like Exp Filtertable, that frame is treated as a shape in time. It becomes the fingerprint, the impulse response, that gets stamped across your audio. To find out what that actually does to your frequencies, the plugin runs it through the FFT, and the resulting filter shape is whatever that frame’s frequencies happen to add up to. In other words: you pick a wavetable shape, and the actual frequency response comes out as some other shape, decided indirectly, by the math.

Wavefield reads the exact same frame a different way: directly as a frequency curve. Every point along the frame is simply a volume level for each frequency, applied straight to the spectrum. There’s no fingerprint and no in-between transform. The shape you see on screen is the filter you hear: boost where it’s high, cut where it’s low, with the ability to draw and edit the frequency curves, and your filter shapes are applied exactly as drawn.

Both plugins have the same wavetable starting point, but Filtertable converts the shape into a filter, and Wavefield reads the shape as the filter. That’s the whole difference in a sentence. And that difference has consequences you can hear the second you start moving the wavetable position.

Why the Motion Sounds So Different

This is why some spectral filters make wild, sweeping motion and others just kind of… shimmer in place.

This part isn’t about what shapes each plugin can make. Both can make detailed shapes. It’s about what happens when you morph: when you slide through the wavetable while it’s playing.

In Wavefield, because the frame is read directly as frequencies, position along the shape lines up with position in the spectrum. The left is your lows, the right is your highs. So when you morph, peaks and cuts in the shape can arbitrarily slide across the frequency range. That’s what you hear as a sweep: formants shifting, energy moving up and down, different parts of the spectrum going different directions at the same time. Dramatic, obvious motion.

Using convolution, morphing the wavetable position on its own behaves differently. Because the shape is used as a time-domain fingerprint and then transformed, there’s no neat left-to-right line between the shape and the spectrum. So morphing tends to make the character fluctuate in place. It ripples and shifts, but the fact that the time domain IR affects all frequencies simultaneously limits its ability to create wild sweeping. Its effect is more “evolving texture.”

Convolution tools know this, and they add controls to compensate. Exp Filtertable has some actually very cool frequency FX like fold and warp, plus a shift control that works a lot like a cutoff, transposing the whole shape up and down, as one block. It relocates the character; it doesn’t make the spectrum move piece by piece the way direct-frequency morphing does.

Wavefield has its own set of sound shaping controls, all built around this “the frame you see is the filter you get” concept, including built-in LFO modulation through the wavetable, the ability to edit the shapes, step sequence and gate them, change playback direction, skew shaping along the spectrum, split shaping between left and right channels for a one-of-a-kind stereo width that’s mono compatible, and a lot more.

So in Wavefield, dramatic spectral motion is built into wavetable position morphing. It comes straight out of reading the shape as frequencies and doesn’t require additional compensation. In a convolution tool you build that motion up with extra modifiers, and it has a different feel. If hands-off, “just hit play” sweeping movement is the sound you’re chasing, Wavefield’s main modes accomplish this. And there’s still a third “Spectral IR” mode if you want that convolution flavor.

Exp Filtertable vs Wavefield: Which Should You Reach For?

So between Filtertable and Wavefield, which one should you actually reach for? It depends entirely on the sound you’re after.

Exp Filtertable’s convolution gives you smooth character initially, and its frequency FX - fold, warp, scale, shift - let you push it into bigger movement when you want it. Wavefield’s direct approach gives you dramatic, sweeping spectral motion straight from wavetable position morphing, and because it only touches how loud each frequency is, not the timing or phase, your signal’s phase stays intact while you carve. It also has a convolution-style mode of its own when you want that flavor. Neither one is objectively “better.” They’re different tools that get you to different sounds.

And one quick note on phase, since I mentioned it. Phase is just the timing relationship between your frequencies. It’s the thing that, when it gets knocked around, smears your transients or thins a sound out. A lot of spectral processors nudge it as a side effect without telling you. Wavefield’s magnitude mode leaves it completely alone. This protects the integrity of your source material while applying any degree of frequency warping. You can drop a phase response meter on it and watch it stay clean.

But don’t take my word for any of this. Both plugins have free demos. Load them both up, morph a wavetable through each, and just listen. The difference is obvious the second you hear it. That’s the only comparison that really counts.

Exp Filtertable vs Wavefield at a Glance

Frequently Asked Questions

What is Exp Filtertable?

Exp Filtertable is a wavetable-based spectral filter plugin for sound design. It uses the frames of a wavetable to shape audio across the frequency spectrum and lets you morph through frames to animate the filtering over time. It applies the wavetable using a convolution-based approach rather than the STFT method.

What is convolution in audio, in plain terms?

Convolution applies one sound’s “fingerprint,” its impulse response, continuously across another sound. It’s the technology behind convolution reverbs that capture real spaces. Because it slides that fingerprint across the whole signal smoothly, there are no chunk boundaries, which is why it avoids certain spectral artifacts.

Is Exp Filtertable a convolution or STFT plugin?

Exp Filtertable uses a convolution-based approach. According to its own product description, it chose convolution over the typical STFT method to avoid artifacts caused by misaligned analysis blocks. Both approaches actually use the FFT internally; the difference is in how the processed audio is reconstructed.

What’s the difference between Exp Filtertable and Wavefield?

Both are wavetable spectral filters. Exp Filtertable convolves with the wavetable, so the frequency response is the FFT of the frame, determined indirectly. Wavefield reads the wavetable directly as a frequency curve, mapping each frame onto the spectrum as per-frequency gains, so the shape you see is the filter applied.

Does STFT always degrade sound quality?

No. STFT artifacts come from chunks that are reconstructed incorrectly, not from spectral processing itself. With correct windowing and overlap-add reconstruction, an STFT-based filter can be clean. Convolution is one way to avoid the artifacts; correct reconstruction is another way to reach the same clean result.

Does convolution use the FFT?

Yes. Real-time convolution is normally done with fast convolution, which chops audio into chunks and uses the FFT, the same building blocks as an STFT effect. So convolution-based versus STFT-based isn’t a question of FFT or no FFT; it’s a question of how the chunks are reconstructed.

Why does morphing a wavetable sound more dramatic in some spectral filters than others?

It depends on how the frame is read. When the frame maps directly to the frequency spectrum, morphing sweeps spectral features across frequencies independently, giving dramatic multi-directional motion. When the frame is convolved as a time-domain fingerprint, the raw position morph evolves in place; convolution tools add modifiers like fold, warp, and shift to introduce bigger movement, but those tend to move the shape globally rather than per-frequency.

Is Wavefield an alternative to Exp Filtertable?

Wavefield is a wavetable-based spectral filter plugin that can serve a similar creative role with a different approach: it reads wavetable frames directly as frequency curves and includes a Magnitude mode that shapes the spectrum without altering phase, plus a convolution-style mode for different character.