University of Washington (UW) researchers have developed a system that generates audio from silent piano performances.
UW's Eli Shlizerman explained the goal of the research “was to see if artificial intelligence could generate music that was played by a pianist in a video recording."
Audeo decodes what transpires in the video and translates it into music, by first detecting which keys are pressed in each frame to produce a diagram; it then converts the diagram into something a music synthesizer would recognize as a sound a piano would make.
This step de-noises the data and adds more information, like how strongly each key is pressed and for how long.
When the UW team tested Audeo's compositions with music-recognition applications like SoundHound, the apps correctly identified the piece with about 86% accuracy.
Said Shlizerman, “We hope that our study enables novel ways to interact with music.”
From UW News
View Full Article
Abstracts Copyright © 2021 SmithBucklin, Washington, DC, USA
No entries found