What can one accomplish in 60 seconds? The fastest clapper in the world can do 1,020 claps in a minute. A car can be stolen in just 60 seconds. And now, an AI can copy a human voice in one minute. A company has successfully created artificially intelligent technology that is capable of recreating the voice of an actual person and it only needs 1 minute of audio samples to accomplish the feat. Impressive? Yes. Scary? Definitely.
According to Tech Crunch, the Montreal-based Lyrebird successfully developed a voice generator that can copy a human's voice and how he or she actually talks. That means the recreated voice comes with emotions, stresses, intonations, dictions, and the like. In short, the AI does not sound monotonous unlike existing digital assistants such as Alexa, Siri, Google Assistant, Cortana and even the yet-to-be-heard Bixby.
The AI startup recorded samples of what the voice imitation algorithm can do. The company chose audio samples from former U.S. President Barack Obama, Hillary Clinton, and Donald Trump. The samples can be heard here. In the audio samples, each personality can be heard talking. Another recorded audio involves the three having a conversation with each other and talking about Lyrebird.
This development is definitely impressive if looked upon scientifically or technologically. However, the possible repercussions can already be imagined and they are nothing to scoff about. A tech or device that can copy someone's voice can easily be taken advantage. With the ongoing problem with fake news, this technology can only worsen the situation. Another possible scenario involves smart devices such as Google Home and the Amazon Echo. Lots of homes today are now equipped with smart devices that rely on the owner's voice to do specific tasks like opening the door to the house or turning the engine of the car on. Now imagine someone using the tech to copy the owner's voice. How about a voice-activated robot army or a nuclear missile, if ever that exists?