Slashdot Mirror
An AI System For Editing Music in Videos (mit.edu)
Amateur and professional musicians alike may spend hours poring over YouTube clips to figure out exactly how to play certain parts of their favorite songs. But what if there were a way to play a video and isolate the only instrument you wanted to hear? MIT News: That's the outcome of a new AI project out of MIT's Computer Science and Artificial Intelligence Laboratory (CSAIL): a deep-learning system that can look at a video of a musical performance, and isolate the sounds of specific instruments and make them louder or softer. The system, which is "self-supervised," doesn't require any human annotations on what the instruments are or what they sound like. Trained on over 60 hours of videos, the "PixelPlayer" system can view a never-before-seen musical performance, identify specific instruments at pixel level, and extract the sounds that are associated with those instruments.
For example, it can take a video of a tuba and a trumpet playing the "Super Mario Brothers" theme song, and separate out the soundwaves associated with each instrument. The researchers say that the ability to change the volume of individual instruments means that in the future, systems like this could potentially help engineers improve the audio quality of old concert footage. You could even imagine producers taking specific instrument parts and previewing what they would sound like with other instruments (i.e. an electric guitar swapped in for an acoustic one).
Really? What substance do they pour over the clips? And to what end? Do they pour a liquid, like coffee? Or a fluid-like solid, like sand?
I'm sorry. Pouring vs poring is one that really bugs me, for some reason.
... it can isolate first chair clarinet farting.
It little behooves the best of us to comment on the rest of us.
...than being able to re-edit video game sound tracks.
--
"A trumpet says what?" - H. Stern
It can do image recognition frequency filtering and change the volume too? AI is amazing. I am so glad I live in 2018! One of these days these AI experts are gonna make AI do something really useful, and then watch out! The sky is the limit!
This sounds like the system is combining object recognition of musical instrument and processing the audio in order to separate the musical instruments. But that's not the hard part. A middle school band student can provide a list of the instruments in a video. The hard part is going to be separating the instruments in a way that sounds good.
As for previewing the same line on a different instrument, that's what MIDI is for. The issue would be the quality of the samples used.
This is my signature. There are many like it, but this one is mine.
That is a hell of an AI that can do anything intelligent with only 60 hours of training videos
I wonder if they can remove Yoko Ono's banshee cries from this live performance of Johnny B Goode
What does that mean?
I'm sure they can be fooled by playback, when the musicians only fake the playing.
It's called training your ear to hear, and it's a necessary musical skill for a musician. This may still be useful, but man, most engineers are so left-brained it's a miracle they don't tip over.
would be to use it to remove annoying background irrelevant heavy metal music that people add to their favorite movie clips so you can actually watch and hear the real in-movie clip without the annoying distraction.
Also, on Twitch (and YT) many clips either get muted by the service because of copyright claims. Having this feature could strip out the offending sound but leave the streamer's voice (and room sounds) unscathed. Everyone wins.
If your fart noise can no longer stay hidden under layers of surrounding noises, imagine what this system could do for spies? Consider the surveillance feed of any noisy/busy place where people meet and converse. Now you can isolate and ease drop in on each conversation with clarity far exceeding what a normal surveillance system could ordinarily do.