There are occasions when you need to reduce the amount of reverb in a recording and fortunately there are now plenty of tools available to assist with this. Paul Maunder takes five of the most popular plugins that claim to do the job and pits them against each other in a comparison test - which do you think performs the best?
Update - the results are in - check out the link at the bottom of this article.
They are:
Accentize DeRoom Pro 2
Based on machine-learning, DeRoom Pro 2 from Accentize uses a trained artificial neural network to separate direct sound from the reflection components. The Spectral Focus mode allows the reduction amount to be tailored across 3 user definable frequency bands.
iZotope RX De-Reverb
Due to the fact that this is included with iZotope RX, De-Reverb is one of the most widely used reverb removal tools. It offers a learn function which can intelligently detect the reverb profile in the audio signal. Frequency controls are provided so you can selectively adjust the reverb reduction amount across 4 bands. The main reduction control allows reverb to be reduced or increased.
Zynaptiq Unveil
Released in 2012, Unveil was one of the very first general purpose reverb removers. It’s designed to work well with complex material like full mixes, or any source in the presence of significant amounts of background sound. It allows the reverb level to be increased as well as decreased.
Acon Digital DeVerberate 3
DeVerberate 3 uses an algorithm based on deep learning for fully automatic (or manual) reverb reduction of dialogue. An early reflections filter works in conjunction with the reverb suppression algorithms and estimates the impulse response of the early reflections which is then used to cancel out the effect of the early reflections.
Plug-in Alliance SPL De-Verb
This transient designing plug-in is the simplest of the bunch to operate. Reverb reduction is controlled with just a single dial. SPL De-Verb also has the option to switch in a limiter if needed plus a dry/wet knob which allows you to set a balance between the unprocessed and processed signals. There’s also an output gain control.
The Recordings
For each plug-in, 4 dialogue recordings were used, each of which includes reverb with its own characteristics. The recordings are intended to test the de-reverb plug-ins in different ways.
In each case there is a short introduction followed by a count from 1 to 5. The purpose of the count is to provide short bursts of speech so the reverb tail can be easily assessed. A description of each recording is provided below, along with the audio. The original, unprocessed recoding is followed by the processed versions from each of the 5 reverb reduction plug-ins. The examples have been randomised and labelled A to E.
Take a listen to each example and vote on which one you think is best. The results will be revealed soon along with details of which specific processors from the iZotope, Acon Digital, Accentize, Zynaptiq and Plugin Alliance processors were used for each.
Long Stereo Reverb
This is a clean studio recording with a long artificial stereo reverb added. The reverb time is about 1.8 seconds.
Hallway
This recording is intended to represent the kind of audio which you’re likely to actually use de-reverb plug-ins on in audio post. It was made in a hallway with smooth plaster walls and ceiling. There is some background noise in addition to the obvious reflections.
Bathroom
Made in a small tiled bathroom, this recording exhibits a fairly short but obvious reverb due to the highly reflective walls, floor and ceiling.
Basement
A midsized domestic basement with very few soft surfaces and a concrete floor is not an ideal place to make a recording, but it’s a good way to test de-reverb plug-ins with this stereo recording.
UPDATE: The results have been analysed, check out our article De-Reverb Listening Test - The Results.