There is a body of work in the field of intelligent music production covering a range of specific audio effects. However, there is a distinct lack of any purely machine learning approaches to automatic mixing. This could be due to a lack of suitable data. This paper presents an approach to used human produced audio mixes, along with their source multitrack, to produce the set of mix parameters. The focus will be entirely on the gain mixing of audio drum tracks. Using existing reverse engineering of music production gain parameters, a target mix gain parameter is identified, and these results are fed into a number of machine learning algorithms, along with audio feature vectors of each audio track. This allow for a machine learning prediction approach to audio gain mixing. A random forest approach is taken to perform a multiple output prediction. The prediction results of the random forest approach are then compared to a number of other published automatic gain mixing approaches. The results demonstrate that the random forest gain mixing approach performs similarly to that of a human engineer and outperforms the existing gain mixing approaches.
Click to purchase paper as a non-member or login as an AES member. If your company or school subscribes to the E-Library then switch to the institutional version. If you are not an AES member and would like to subscribe to the E-Library then Join the AES!
This paper costs $33 for non-members and is free for AES members and E-Library subscribers.
The Engineering Briefs at this Convention were selected on the basis of a submitted synopsis, ensuring that they are of interest to AES members, and are not overly commercial. These briefs have been reproduced from the authors' advance manuscripts, without editing, corrections, or consideration by the Review Board. The AES takes no responsibility for their contents. Paper copies are not available, but any member can freely access these briefs. Members are encouraged to provide comments that enhance their usefulness.