covolution_reverb_and_irs

This is an old revision of the document!


Convolution Reverb and IRs (Impulse Responses)

The idea behind convolution reverb based on impulse responses is to emulate the placement and recording of a sound source in a real physical space. Decades ago, one would have had to move the sound source into that space (a small room, a live stage, a concert hall, a church or simply set up a loudspeaker cabinet) and set up all the required recording equipment. Since the invention of IR-based convolution, that process has become a matter of providing the sound to process, the impulse response and software that is able to do the mathematical convolution of both input signals.

How are IRs recorded? You can read more about the process here (link as of January 2020):

https://en.m.wikipedia.org/wiki/Convolution_reverb

A more technical description with details can be found here (link as of January 2020):

https://sourceforge.net/p/hesuvi/wiki/How-To%20Record%20Impulse%20Responses%20Digitally/

Convolution reverbs can use impulse responses (often just referred to as IR or IRs) in different ways. Guitar amp and cabinet emulations often use monaural impulse response files because both the guitar and the microphone used to record the sound of amp speaker are monaural too. If you hear today's guitar tracks, the stereo effect usually originates from doubling (recording multiple takes of the same guitar riff and panning them to different positions in the stereo image), giving it a much more lively character due to the more or less tiny (human) differences between each recorded take.

Mono signals can be effected in stereo or even surround however, given that IR files are used that contain the respective amount of channels that have been recorded in the environment we want to emulate. Ideally, the loudspeakers for reproducing the effected signal have a position that reminds the positioning of the microphones used for recording the IR.

Stereo signals can also be processed in “true stereo” which is desirable when 1. the sound source has enough difference between the left and right channels (e.g. a choir, a big band or an orchestra), and 2. the natural reverberation of the left and right sound sources are different enough to make separate processing worthwhile.

Here's a short explanation of the different processing modes, taken from the Liquidsonics web site:

Parallel Stereo The left input channel is convolved with the left impulse response file channel and the right input channel is convolved with the right impulse response file channel. This is the typical configuration for stereo convolution reverbs when used with stereo impulse responses, although when input audio is panned left or right, using Mono to Stereo may provide more intuitive results.

True Stereo The left input channel is convolved with the left and right impulse response file channels from IR1-A and the right input channel is convolved with the left and right impulse response file channel from IR1-B. The two output convolutions’ respective left and right components are then summed into a single stereo output. This configuration is necessary to take full advantage of true stereo impulse responses. True stereo impulse responses are required to be provided as two separate stereo files and loaded into IR1-A and IR1-B (or IR2-A and IR2-B). This configuration is typically found in high-end algorithmic reverbs.

Mono to Stereo The left and right input channels are mixed to mono and then independently convolved with the left and right impulse response file channels. When using a single stereo impulse response file, this is useful when input audio is panned hard left or right; this configuration is often encountered in low/medium-end stereo algorithmic reverbs.

(Source URL from January 2020: https://www.liquidsonics.com/knowledge-base/reverb-processing-topologies/)

Convolution reverbs can be used to interesting effect when you use non-IR (impulse response) files instead of impulse responses. The convolution acts as a sort of spectral filter. The result is that the source sound is “filtered” with the file you as the IR. Since convolution is basically the multiplication of the signal spectrum with the IR's spectrum, it cannot add harmonics that aren't present in the source signal.

A quick demo of a decidely non-reverb application of convolution

How to Import Import Ownhammer Cabinet IRs into THAFKNAR or Impulsation

Some IRs can be found for download on the following pages:

this page is under construction and needs your help to add information

  • covolution_reverb_and_irs.1606632775.txt.gz
  • Last modified: 2020/11/29 17:52
  • by espiegel123