A.I. Mastering

 
 

There is a new online mastering service, which goes by the name of Landr and provides remote, (instant) mastering services free of charge. That’s right “free” but only for 192 kbps MP3s. If you want uncompressed wav files you will have to dish out $9-$19 per month, still not a bad deal if these services are what they say. It is true that mastering is critical to achieving that “polished” sound, but the reality is that it can be an expensive undertaking for independent artists. Just think about how much that an artist has already spent on production, recording and mixing. By the time a project reaches the mastering phase the budget has been depleted.

MixGenius claims that more than 8 years of research went into the algorithms utilized to master the audio. The website goes on to say, “Our team is composed of music industry veterans - award winning mixing engineers, top-level DSP programmers, musicians, producers and label owners - who know exactly what the mastering process needs to deliver.” With that amount of brainpower and resources, my expectations that this might actually work increased significantly.

I read mostly cynical and negative comments in audio forums with a few positive remarks sprinkled in. Most of the doubt was based on the A.I. and the fact that it was attempting to emulate human decision making on something as subjective as music.  I decided to try the service for myself. First I tried the free service and uploaded a song that I had mixed which was a blend of hip hop and reggae. The initial results of the 192 kbps MP3 was less than great. The loudness it provided does indeed compete with industry releases but the quality of the compression at 192 was not the very good. Even with the less than ideal quality of the mp3, it showed enough potential for me to dig further into their premium services. I tried the $9 per month “amateur” service, which allows 4 uncompressed wav masters. The result of receiving wav files verses the 192 kbps mp3 was night and day. The low end was tamed a bit and the premium services gave you the option to pick the level of compression. Low, Medium, and High. The low setting was the most musical and dynamic, but lacked the loudness to compete with mainstream releases. The High setting was very competitive with industry loudness standards but lacked musicality and dynamic range. My favorite setting was the medium, which gave you a little bit of both. 

In my opinion, Landr might not be the ideal mastering situation for the independent artists, but it can work with specific genres especially if you are trying to compete with industry “loudness.” I can see these types of A.I. mastering services getting relatively better in the coming years. Another application that producers and mix engineers may take advantage of is using Landr to expose potential problems in the mix.