Music created using Google DeepMind’s Lyria artificial intelligence model will contain invisible watermarks. The technology in question is SynthID, which will allow people to identify that audio was created with the help of AI, writes The Verge.
DeepMind explained that the watermark will not affect the sound quality and is not perceived by the human ear. At the same time, the mark will always be visible, even if the audio track is slowed down, sped up, or otherwise altered.
The developers explain that SynthID technology works by “converting a sound wave into a two-dimensional visualization that shows how the frequency spectrum of sound changes over time.” They claim that this approach is unlike anything else that exists today.
In the summer, Google started testing SynthID technology to recognize images created by AI. The watermark will be embedded in individual pixels of the image, so the human eye will not be able to see it, but the computer will recognize it.
Meanwhile, the developers have warned that the technology is not a “reliable protection against extreme image manipulation.”