Audio Engineering Society Convention Paper Presented at the 127th Convention 2009 October 9–12 New York, NY, USA The papers at this Convention have been selected on the basis of a submitted abstract and extended precis that have been peer reviewed by at least two qualified anonymous reviewers. This convention paper has been reproduced from the author's advance manuscript, without editing, corrections, or consideration by the Review Board. The AES takes no responsibility for the contents. Additional papers may be obtained by sending request and remittance to Audio Engineering Society, 60 East 42 nd Street, New York, New York 10165-2520, USA; also see www.aes.org. All rights reserved. Reproduction of this paper, or any portion thereof, is not permitted without direct permission from the Journal of the Audio Engineering Society. Automatic equalization of multi-channel audio using cross-adaptive methods Enrique Perez_Gonzalez 1 , and Joshua Reiss 1 1 Centre for Digital Music, Queen Mary, University of London, London, E1 4NS, England enrique.perez@elec.qmul.ac.uk, josh.reiss@elec.qmul.ac.uk ABSTRACT A method for automatically equalizing a multi-track mixture has been implemented. The method aims to achieve equal average perceptual loudness on all frequencies amongst all multi-track channels. The method uses accumulative spectral decomposition techniques together with cross-adaptive audio effects to achieve equalization. The method has applications in live and recorded audio mixing where the audio engineer would like to reduce set-up time, or as a tool for inexperienced users wishing to perform audio mixing. Results are reported which show how the frequency content of each channel is modified, and which demonstrate the ability of the automatic equalization method to achieve a well-balanced and equalized final mix. 1. INTRODUCTION Equalizing a sound mixture is one of the most expert human tasks related to live music mixing. The main problem of determining the amount of equalization to be used is that the perceived amount of equalization is different from the physical amount of equalization applied. In order to achieve a perceptually pleasant equalization several things should be considered; whether or not the channel needs equalization at all, how many filters should be used, the type of filters and ultimately the amount of boost or cut they should have. Some studies on how the sound engineer performs these decisions have been performed by [1, 2]. Automatic mixing of speech and music levels has been attempted by [3-6]. However, very little has been done to attempt self-equalization of musical signals. The only notable example of such an attempt is [7], where the use of an off-line machine learning approach to the problem where humans need to manually train the machine. Once the machine is trained, it equalizes using nearest neighbor techniques. In this paper the authors will propose a method for use in live mixing situations driven by perceptual indicators. The proposed system does not require off-line machine learning methods. Instead it uses a real time cross-adaptive accumulative spectral decomposition approach to the problem based on a multiband implementation of [6]. A cross-adaptive algorithm is based on inter-channel dependency