Microsoft has developed an advanced artificial intelligence (AI) text-to-speech program that achieves human-like believability. VALL-E 2 is the first program of its kind to achieve “human parity,” meaning its speech cannot be distinguished from that of a human. However, the technology remains strictly a research project and is unavailable to the public.
“It may carry potential risks in the misuse of the model, such as spoofing voice identification or impersonating a specific speaker,” researchers say. There are therefore “no plans to incorporate VALL-E 2 into a product or expand access to the public.”
The program can replicate voices with remarkable fidelity after processing as little as three seconds of audio, surpassing previous systems in speech robustness, naturalness, and similarity to the original speaker.
There are concerns about voice spoofing, with the technology used for impersonation or identity fraud, particularly in phone scams.
Misuse of AI remains a concern in the upcoming presidential election, with AI being used for fake robocalls using Joe Biden’s voice earlier this year in New Hampshire.
Some Biden supporters are suggesting AI should be used by the Biden campaign to mask the 81-year-old’s obvious cognitive decline, frailty, and confusion from the public.
