We all know how important it is that the ‘space’ within our mixes is carefully managed. King of the spatial effect category is reverb and there is a dizzying array of reverb tools available to producers in both hardware and software form. Broadly, reverb can be added to a project via two means; artificial reverbs and convolution reverbs. The differences between these approaches are somewhat comparable to the difference between a synthesiser and a sampler reproducing, for instance, a string sound. A synthesiser would use artificial means to appropriate the sound of the strings, using waveforms whose harmonic content was a match for those played by stringed instruments, alongside tone and envelope-shaping tools to get sounds even closer to reality. A sampler, however, would use real recordings of the stringed instruments. The same is true when it comes to shaping reverbs; artificial reverb units synthesise an imaginary space, providing parameters to control its size, reverb time, diffusion amount and plenty more besides, while convolution reverbs load a sample - called an ‘impulse response’ - which is a recording of the spatial behaviour of a physical space or object. Through the following pages, we’re focusing our attentions on this second approach.

How does that work?!

Hang on, though. How do you make a recording of a space? Surely for that space to ‘come alive and make a noise’, you need to trigger a sound in it. And how do you keep that particular sound out of the sound sources you do actually want to pass through your convolution reverb? In this sense, the use of a sample used at the heart of a convolution reverb differs slightly from a sample used for playback in a regular sampler. A sample recorded to be used within a convolution reverb is called an impulse response and the way it’s used by the reverb unit itself involves quite complex maths. The earliest convolution reverbs required this process to be done manually but these days, plugins like AudioEase’s industry-standard Altiverb and Logic Pro X’s own Space Designer handle the maths for you. It goes like this: The reverb host will analyse the recorded impulse response file and ‘remove’ the source trigger sound from it in a process akin to phase inversion, effectively removing from it the source sound while preserving the reverberant response (not present in the source sound alone). What this leaves is a reverb tail ready to be applied to any new source sound though, of course, that reverb tail’s ‘success’ is still subject to the qualities of the source recording. Ideally, an impulse response needs to capture a full range of frequencies to be of maximum value and this is only possible if the trigger sound, to generate the impulse response, is rich in frequency content itself. Generally, anyone taking the process of recording impulse responses seriously (either because they’re developing their own convolution plugin, or looking to sell impulse responses as ‘reverb sample packs’) would look to record a sine wave sweep in their chosen space, starting with ultra-low frequencies, sweeping up to super-high ones and stimulating a space’s response to all of the frequencies in between in the process. This isn’t always practical, however, particularly if you have a great-sounding space and are limited to what you can trigger and record in that space. In these circumstances, short percussive sounds like a snare drum hit, or starter-pistollike sound, trigger a wide enough range of frequencies for a satisfying result when imported into your convolution reverb as a new impulse response. Once recorded, trim that audio file so its start and end points are clearly defined - then it’s ready to load.