Singing its way forward

Can music be machine made Google’s Project Magenta is here to answer in the affirmative and how In the latest edition of Moogfest in North Carolina, Google previewed its Project Magenta, an Artificial Intelligence (AI) product with tech robots to show the way to sing.

As what Deep Dream does on images by imparting surreal twists, Project Magenta has music to experiment on. At the fest, a Google Brain (the machine learning wing) researcher demonstrated music generated on a digital synth. A five-note sequence was transformed into a piece of melody before the audience. Google expects to improvise the quality of output by feeding in more data to the software, i.e. more examples of the music.

Using TensorFlow, Google’s open source distributed computing machine learning and neural network framework, Magenta is to generate algorithms to make music, which would go on to create videos and images subsequently.

Its future implications on apps are predictable. Once released, Magenta, as it is codenamed, would be Google’s advanced competitor for other tech giants’ ‘intelligent assistants’ on voice commands. From what is hinted, as a precursor, an app release is on the cards. It is going to be a platform where users can experience the Project Magenta creations.