The age-old debate about whether a robot will be able to replace a person in real creative work has moved to a new field of confrontation. We are talking about the field of mastering - the final and most complex processing of music tracks before launching them into circulation. This service can cost from several hundred to several tens of thousands of dollars and is sometimes not available to ordinary indie musicians. However, for the third year in a row, neural networks have been successfully performing human work in this area.
Mastering is the most accurate, on the verge of perception, adjustment of the shades of a musical composition. As the saying goes, in this case you need to make "the sky is bluer and the grass is greener" than it already is. This is a whole science, in which the most important element is the personal experience and philosophical approach of the mastering engineer, which is almost impossible to reflect in the form of algorithms and instructions. Therefore, AI developers in this area took a different path.
The most famous service for automatic mastering is called Landr, which analyzes a track and then offers one of the typical processing options. There are only three schemes available, but they are universal and correspond to 90% of audience requests. In blind tests with major labels and professional sound engineers, Landr was named the best mastering service last year.
A different approach is taken by iZotope, whose mastering AI divides the track into fragments, and then prompts the user to choose a processing scheme for each. AI takes over the rough, routine work, for which non-professionals do not have the knowledge and skills, but the responsibility for the result lies entirely with the customer of the service. Today, about 10 million tracks have already been released in the world using the automatic mastering function.
Mastering engineers themselves do not lose heart in the light of new trends. On the contrary, now the important line is finally drawn, musicians can choose between cheap and versatile mastering from AI and expensive but professional mastering from humans. And this is much better than when there was no alternative at all and many promising compositions disappeared due to the fact that the author or publisher did not have the money to finalize their product.