The paper embarks on measuring masking in an audio mix. Masking is the process where the masker simultaneously raises the audibility threshold of the maskee. In audio production, spectral masking makes it difficult to distinguish the sounds and the sound sources. The amount of masking varies from one listener to another and also depend on the characteristics of the maker and the maskee audio sounds. In general, masking makes a sound mix unclear, confusing and underwhelming. The experiment used both real-time and off-line autonomous multitracking systems to reduce the masking in the multitrack audio. Multitrack equalization was both independent and low latency (Hafezi & Reiss, 2015). The researchers use subjective and objective techniques to compare the results of the different audio mix results implemented by professional and amateur mix engineers. Four different methods were used; Offline semi, offline fully, real-time constrained and real-time unconstrained.
The results of the experiment indicated that autonomous systems have a positive effect of improving the quality of a mix by reducing the perceived and objective spectral masking. However, measuring masking in a multitrack musical audio using objective techniques would be difficult and ill-suited for the task. Both professional and amateur engineers find it challenging to equalize a multitrack mix. The challenges stem from identifying tracks with problems in their frequencies as well as determining their spectral locations (Hafezi & Reiss, 2015). Through the experiment, the researchers were able to automate the whole multitrack equalization process. The offline semi-autonomous automated system was able to improve the quality of a mix by just controlling one parameter of the user. The quality of the system was close to that of the professional mix engineer and much better that an amateur’s raw mix.
Reference
Hafezi, S. & Reiss, J. (2015). Autonomous Multitrack Equalization Based on Masking Reduction. Journal of the Audio Engineering Society, 63 (5), 312-323.
