Google takes massive first steps into the world of synthesizers
A group of scientists on the Google Brain team or pushing the boundaries of sound creation through machine learning.
Machine learning is “the field of study that gives computers the ability to learn without being explicitly programmed” and when the geniuses at Google applied it to music and sound design, sparks started flying. The idea is to remove a large portion of the complex sound design tasks and give the musician new and intuitive ways of creating music.
Nsynth is Google’s web-based app that learns the fundamental aspects of what makes up each sound, and uses an algorithm to combine the two sounds – resulting in a totally new tone. The concept is simple but the results are complex, rich timbres. Within minutes after using the Nsynth web app for the first time, my mind was wondering if and when we would see a plugin version.
It looks like another groups of scientists at Google’s Creative Lab have taken it a step further by taking the complex algorithms of Nsynth and incorporating them into a tactile hardware device. Dubbed the Nsynth Super, it builds on the existing technology by offering four sound sources to mix between, and 100 000 pre-loaded sounds.
The best news of all – It’s completely open-source
That means everyone has access to the source code, schematics and designs, you can build an Nsynth Super at home if you have the technical know-how. It also means that it is only a matter of time until programmers from other audio companies start incorporating this revolutionary algorithm.
On closer inspection it looks like the system is built around a Raspberry Pi 3, a relatively inexpensive system that a lot of electronics hobbyists already have, alongside some other easy-to-find components. That being said, it’s probably much easier to order the board pre-assembled, although Google have not announced an official date of availability.
Check out the Nsynth website for more details.
If you are keen to get hands-on and make your own, check out their GitHub page
