I am not envious of the policy folks at Youtube who will have to parse out all the edge cases over the next few years. They are up against a nearly impossible task.
Most interesting example to me: "Digitally altering audio to make it sound as if a popular singer missed a note in their live performance".
This seems oddly specific to the inverse of what happened recently with Alicia Keys from the recent Superbowl. As Robert Komaniecki pointed out on X [1], Alicia Keys hit a "sour note" which was silently edited by the NFL to fix it.
Correct, it's the inverse that requires disclosure by Youtube.
Still, I find it interesting. If you can't synthetically alter someone's performance to be "worse", is it OK that the NFL synthetically altered Alicia Key's performance to be "better"?
For a more consequential example, imagine Biden's marketing team "cleaning up" his speech after he has mumbled or trailed off a word, misleading the US public during an election year. Should that be disclosed?
I don't understand the distinction. if the intent is to protect the user, then what if I make the sound better for rival contestants on American idol and don't do it for singers of a certain race.
This is a great example as a discussion point, thank you for sharing.
I will be coming back to this video in several months time to check whether the "Altered or synthetic content" tag has actually been applied to it or not. If not, I will report it to YouTube.
However autotune has existed for decades. Would it have been better if artists were required to label when they used autotune to correct their singing? I say yes but reasonable people can disagree!
I wonder if we are going to settle on an AI regime where it’s OK to use AI to deceptively make someone seem “better” but not to deceptively make someone seem “worse.” We are entering a wild decade.
A lot of people do! Tone correction [1] is a normal fact of life in the music industry, especially in recordings. Using it well takes both some degree of vocal skill and production skill.
You'll often find that it's incredibly obvious when done poorly, but nearly unnoticeable when done well.
In my opinion, these kinds of apps are the future of data science / data analyst work. Forget no-code, just enable these professionals to work in a single programming language that they're familar with and give them visualization superpowers. The Python ecosystem has https://www.streamlit.io/ and https://gradio.app/ now. R has https://shiny.rstudio.com/. I think we'll see more.
Pretty cool. The embedding similarity approach makes a lot of sense. I actually this project by experimenting with computing cosine similarities of sentence embeddings [1]. But I wasn't very impressed with out-of-the-box results, and I found it difficult to set a similarity threshold for a match. QA was the second try, and the pretrained models worked better out of the box. I'm wondering if I should revisit the embedding approach now...
The extension feeds it context data in chunks (e.g. only one <p> element at a time). I could see TensorFlowJS BERT struggling with a single sequence of 87 KB -- have you tried breaking them up?
Nope, I've developed and hosted my own summarization API here. It does heavily rely on open-source machine learning libraries and research / datasets released by Facebook AI [1]
https://novehiclesinthepark.com/