Newsletter
Right now, YouTube has a massive task on its hands. It has to simultaneously provide generative AI tools for creators to create media and police AI-generated media so that it doesn’t infringe on copyright. A huge example of YouTube’s struggle is the song “Heart On My Sleeve”, a song using AI-generated vocals from The Weeknd and Drake. Well, YouTube is working on AI music tools that will let people clone notable artists; the catch is that it will be legal this time.
What most people don’t realize is that whenever someone emulates a person’s voice or music style and posts it to YouTube, that could possibly make YouTube itself a target of legal repercussions. It is the platform that’s hosting these videos. Also, many people hold YouTube responsible for detecting AI-generated content.
There are a few issues with this. For starters, AI-generated content is getting harder and harder for people to detect. Also, hundreds of thousands of videos are posted to YouTube each day, and the company does not have an algorithm in place to detect every single instance of AI-generated content.
In any case, YouTube is developing AI tools to let you emulate music artists
Right now, YouTube is working on a strict set of rules for people posting AI-generated content to YouTube. While this is going on, the company is coming out with some tools to help people create AI content legally, according to the report. Notably, it’s coming out with a few tools. The first tool is called Dream Track, and this is probably the headlining tool.
Think of it like ChatGPT but for music. If you want to create a piece of music in the style of one of your favorite music artists, you can simply type what type of song you want and what style you want it. The tool will then generate a clip emulating that artist’s Style with AI-generated voices modeled after the artist.
This seems to be a bit ironic seeing as the company is dealing with an issue of AI emulation on the platform. However, it will only generate 30-second clips. So, no one will be able to make full 3 minute long songs. This eliminates the possibility of people using the tool to deceive viewers.
There’s another tool
There is another tool in the works that will give people the ability to create music. This tool is powered by Lyria, which was created by Google Deepmind. It will allow you to sing or hum a short piece of a Melody. Then, the tool will create a short piece of music in whatever style you want.
For example, if you want it to make a short saxophone solo based on a melody that you hum, you can tell the tool to do that, and it will generate that track for you. This is something that the company showed off not too long ago.
This feature is called Dream Screen, and YouTube has a way of detecting when the Tool was used to create music. Each clip will have a very high-pitched tone added to it (one that is undetectable to human ears). So, YouTube will be able to detect that tone and tell whether the music was generated using AI.
Right now, the company is testing out these new tools, so we should hear more about them as time goes on.