Google's machine learning can now turn any sound into music (sort of)

Google loves dabbling in weird and wonderful things with artificial intelligence and machine learning technology used in various experiments. This latest experiment involves applying machine learning to sound to create an entirely different sound, in this case, music.

Tone Transfer, as the experiment is known, uses machine learning algorithms to transform your voice, the sounds of birds chirping, pots and pans clanging or any other sound into a short ditty.

Record 15 seconds of sound and then let the tool does its work and you'll hear a tune played back with a flute, saxophone, violin or trumpet.

To give it a go, head over to the Tone Transfer site. You can do this from an Android phone or from a desktop machine. Then click to try it out. The easiest way to try it out is to click to "add your own" audio and then record yourself singing, humming or making some inane noise.

Once that's done, you can then submit the recording and let the machine learning algorithm work its magic. After a short while, you'll then hear the sound returned via a musical instrument. Simply click between them to hear the different examples.

You might need to play about a bit to get a great result, but it's fun to input different sounds and hear what comes back.

If you're interested in finding out more about what's happening and how it all works, take a look at this blog post.

Original Article