Inventing music has always been one of our proudest achievements as a species. We have been composing and consuming music for millennia. Over the years, thousands of different styles emerged to satisfy the needs of every individual. Human creativity is the key ingredient to composing new melodies and cannot be replaced by any other means. At least that’s what we thought up until recently.

Composing Music with Artificial Intelligence

What if I told that we’re now at the point where we have managed to create AI that is able to generate music. Several research teams and even startups have developed AI that can compose somewhat decent melodies. In some cases, the pieces created by the AI are pretty good actually.

Although this type of research has been going on for many years, we have now reached the point where the outcome is good enough to create human-like composed music. Google’s DeepMind has developed AI-composers using deep learning that are able to create melodies based on pieces of famous composers like Bach,Tchaikovsky and many others. Sony CSL research team has created flow-machines, which is a collection of AI algorithms that aids musicians in composing new music. The results of both teams are quite fascinating and really show a glimpse of how the music industry is going to look like in the very near future.

Some AI Music

These two examples were generated using AI in collaboration with real artists. The outcome is objectively good compared to what most of us expect from AI-composers.

This piece was created by flow-machines from scratch. It was composed in the style of the Beatles.

This piece was created in collaboration with real artists and it’s part of the first ever album released, called Hello World, where AI-composers “met” with real artists to create original work. You can find out more about this album at their official website.

The post Is it time to leave music composing to AI ? appeared first on BadFive.

Making music using new sounds generated with machine learning

Magenta, a team within Google that combines machine learning with art and music, has come up with a new way to make music. NSynth Super is a simple interface that makes use of the NSynth algorithm to create countless new sounds. It uses deep learning to capture and understand a variety of instruments and later combine them to create new, previously unheard sounds.

NSynth doesn’t simply combine or blends sounds. It is able to create a new sound from scratch. During experimentation, NSynth super was able to create 100.000 new sounds from just 4 sources. Sounds exciting doesn’t it? Although NSynth Super is only a prototype, Google provides instructions on how you can make your own using Raspberry Pi and their open source code on GitHub.

Is Artificial Intelligence the new way to make music?

Well, no! At least not yet! After listening to the NSynth Super demo provided by Google, it is clear that we have a long way to go before we create actual music with the help of artificial intelligence. Their demo is mediocre and that’s putting it politely.

NSynth Super Demo

The post Google NSYNTH Super: Is Artificial Intelligence the new way to make music? appeared first on BadFive.