Utilizing AI to revive the voice of a deceased pilot

2 Min Read
2 Min Read

Within the newest signal of the age of AI, the Nationwide Transportation Security Board briefly eliminated entry to its docket system after discovering that the audio of a pilot who died in final 12 months’s UPS aircraft crash was recreated utilizing AI and circulating on the web.

The NTSB is prohibited by federal regulation from together with cockpit audio recordings in its transcript system. In any other case, it accommodates a considerable amount of information about investigations and has traditionally been made obtainable to the general public. Nonetheless, the aircraft’s accident document included a voice recorder spectrogram file. A spectrogram makes use of a mathematical course of to transform an audio sign containing high and low frequencies into a picture.

Scott Manley, a preferred YouTuber whose channel combines physics, astronomy, and video video games, identified on X that it might be attainable to reconstruct audio from megabytes of information encoded in photos.

And that is what occurred. In keeping with the NTSB, folks obtained spectrograms together with publicly obtainable transcripts to create an approximation of the cockpit voice recorder sounds of UPS Flight 2976 in Louisville, Kentucky. In keeping with social media posts, they used AI instruments resembling Codex.

See also  TruthScan vs. Grammarly: Which AI detector is most effective?

The company restored public entry to its docket system on Friday, however 42 investigations, together with these associated to Flight 2976, remained closed pending overview.

TAGGED:
Share This Article
Leave a comment