Get an Impulse, Create a Convolution

When you want to capture an acoustical “sonic signature” (e.g., the natural reverberation of a hall) for use in a convolution processor, you will need to get its Impulse Response (IR) — essentially, a “definition file” of the acoustic space, or a piece of hardware. Once you’ve done that, you can apply the natural reverb of the church around the corner to your vocal tracks, or “implant” the behavior of your favorite hardware processor or mic into a plug-in.

There are several IR-based reverbs in the market, like Audio Ease’s Altiverb, Waves IR-1, TASCAM’s GigaPulse or Wizoo’s WizooVerb W2 (Figure 1). IRs of hardware are often used to create software counterparts of hardware peripherals or synths, but Waves Q-Clone (Figure 2) is a good example of an open-architecture convolution device where you can grab your favorite hardware EQ’s IR, then create virtual versions of that EQ.

There are two methods to obtain impulse responses: “Spiking” and “sweeping” will obtain the raw materials needed for getting convoluted signals, which are a result of multiplying audio signals in the frequency domain. But getting a high quality Impulse Response requires high quality recording equipment — mics, cables, audio recorder, and digital converters with good bit resolution and a high sample rate. As you want accurate, neutral IRs, you need to select carefully where to place the mics when recording IRs of acoustic spaces, and really know any hardware equipment you plan to “sample.”


Convolution reverbs are based on “spiking” IRs. The most common way to obtain a IR by spiking is to record a sound with the most possible frequencies, loud enough to excite the room and fire the most possible reflections, yet short enough not to be mixed in with the actual reverberation sound. A really loud hand clap, a balloon popping, or a starter pistol are good sound sources for this technique. You will likely use condenser mics, as they are generally more sensitive to high-frequency content. Highs are critical to capture accurate reverb tails.

The most natural IRs are taken in stereo as a mono IR sounds less “spacey,” but you can even record surround IRs. If the plosive source is too close to the mics it will lead to distortion, so a lot of pre-production must be done to consider proper mic positioning. It will also help to block other noise sources (doors, windows, air vents, etc.), and decide on a suitable recording technique. A cardioid pattern mic is recommended for small places (that way their patterns will not overlap when placed together), or when you need more detail over certain areas of the hall. Omni patterns are better to get a more general image of a space.


“Sweeping” is a bit more complex, but is suitable for purposes other than reverb tails. It consists of playing a sine tone that sweeps from 20Hz to 20kHz in the space, then recording it with your choice of mics and miking technique. There’s the need to take an additional step, as the resulting signal must be “deconvolved” (the reverse process of convolution, done by a piece of software) to separate the actual IR from the source sine tone. Note that the speakers playing back the sine wave should be as flat as possible, otherwise you’ll have to EQ.

Sweeping, given its nature, covers the audible frequency range and yields better IRs than just spiking. Furthermore, sweeping can obtain the response not only from acoustic spaces but from circuits by sending the sine tone into a hardware device, then recording its output to finally deconvolve it. This lets you create a software processor with the characteristics of the original hardware.

Sure, you can get IRs from the internet or use the presets from your favorite convolution plug-in; but grabbing your own IRs will give you a unique signature sound. It’s worth the effort, and by using IRs creatively, you will obtain sonic textures not available in any processor on the market.