Facebook AI Research (FAIR) team published a paper detailing a new artificial intelligence system it developed for the purposes of remixing and emulating learned music in different styles. The short video that can be seen below demonstrates how the technology is capable of accepting an original piece of music from someone like Mozart or Bach, then emulating the underlying style of such compositions using different instruments. Facebook’s solution could allow for extremely swift music making if properly developed, with the experimental build of the service already being able of recognizing melodies and styles of various origins, including basic whistling.
The neural network-based solution can work entirely unsupervised and successfully produces easily recognizable results, FAIR claims. While not the first technology of its kind, the “universal music translation network,” which is how Facebook is referring to the creation, is an industry first in the sense that its output is of unprecedentedly high quality. Such level of fidelity was achieved by teaching the neural network to encode audio on its own and while the company’s AI experts claim the system is still only capable of melody transfers and not style translations, the limited demonstration of its capabilities does indicate that it’s already much more than just an automated audio remixer.
AI remains one of Facebook‘s core focus points, with the company now integrating such technologies into a broad range of its products and services, including its main social media platform. During a Tuesday Q&A session with members of the European Parliament, Facebook co-founder and Chief Executive Officer Mark Zuckerberg said AI is an important part of the firm’s strategy for combating fake news and misinformation, as well as illegal content. The tech giant’s automated solutions can already detect 99-percent of all extremist content the moment it’s posted on Facebook, Mr. Zuckerberg claimed. Google, Amazon, and the rest of the technology juggernauts are also presently embracing AI on a significant scale, both in terms of their consumer-facing products and internal operations.