Google embeds "watermarks" into music generated by artificial intelligence, which is imperceptible to human ears.

CTOnews.com, November 16 (Xinhua)-- DeepMind, an artificial intelligence company owned by Google, announced that its AI audio generation model Lyria will use SynthID technology to add "watermarks" to the generated audio so that people can later recognize that they were made by AI.

CTOnews.com noted that DeepMind said in a blog post that the watermark will not be detected or affect the listening experience, and that the watermark can still be detected even if the audio is compressed, accelerated or slowed down, or added extra noise.

DeepMind says SynthID audio watermarking works by "converting audio waves into two-dimensional visualization, showing how the spectrum in the sound changes over time", which the company claims is "unlike anything that exists today".

Watermarking tools such as SynthID are regarded as an important protection against the harm caused by generated AI, which is a promising field, but the current technology is far from becoming a master key to defense against forgery.