Dear MEL Topic Readers,
OpenAI says it’s working on AI that mimics human voices
Last month, OpenAI introduced Sora, a video tool that creates realistic-looking
60-second videos from text instructions that include scenes with characters,
motions, and elaborated background details. Then on April 1, it revealed a new
AI tool called Voice Engine that can mimic human voices from a 15-second sample.
Once a text is provided, the text-to-voice AI tool reads it in a way that the
voice sample speaks not only in the original language but also in other
languages like Chinese, German, or Japanese. Since text can be created by a speech-to-text
app easily and quickly, the voice AI tool can be used for simultaneous translators
in multiple languages. But if this kind of voice-mimicking app becomes widely
available, voice can no longer be used for authentication. Also, one can call
you to do something as if he or she is your boss, or disinformation can be
easily created and spread. For example, if such mimic voices are incorporated into fake
videos or Sora, one can become the President of the USA and speak in the Oval
Office.
How easily will AI help or deceive humans?
Read the article and learn about OpenAI’s new voice-mimicking AI tool.
No comments:
Post a Comment