In recent years , the hearing time demand by a piece of AI to clone someone ’s voice has beengetting shorter and shorter .
It used to be minutes , now it ’s just seconds .
OpenAI , the Microsoft - backed fellowship behind the viral procreative AI chatbot ChatGPT , recently bring out that its own voice - cloning technology requires just 15 second of audio material to reproduce someone ’s voice .
In a mail service on its internet site , OpenAI shared a small - ordered series preview of a model call Voice Engine , which it ’s been developing since belated 2022 .
Voice Engine works by feed in it a lower limit of 15 moment of spoken fabric . The user can then input text edition to produce what OpenAI describes as “ affective and naturalistic ” language that “ closely resemble the original speaker system . ”
OpenAI insists it is take a “ cautious and informed approach to a broader release due to the electric potential for synthetical representative misuse , ” adding that it want to “ jump a duologue on the responsible deployment of synthetic interpreter , and how fellowship can accommodate to these new potentiality . ”
It sum up : “ Based on these conversations and the results of these low scale test , we will make a more informed decision about whether and how to deploy this technology at scale . ”
One of the misuses that OpenAI refer to is a scam that some malefactor are already carrying out using alike technology that ’s been publicly available for some time . It involvescloning a phonation and then calling a friendor relative of that person to trick them into handing over immediate payment via a bank conveyance . There are also fears about how such applied science might be used in the coming presidential election , an issue highlighted by a recent high - profile incident in which a robocall using a ringer of President Joe Biden ’s voicetold people not to votein January ’s New Hampshire primary election .
Another business is how the rapidly improving technology willimpact the sustenance of vocalisation actorswho fear that they ’ll be increasingly asked to sign over the rights to their part so that AI can be used to produce a synthetic variant , with compensation for such a contract likely to be much dispirited than if the actor was ask to execute the job in person .
Looking at more positive deployments of the engineering , OpenAI suggests that it could be used to provide reading assistance to non - reader and children using natural - sounding , affective voices “ represent a wider range of speakers than what ’s potential with preset spokesperson , ” as well as instant transformation of videos and podcasts , something thatSpotify is already trialing .
It could also be used to aid patients who are step by step drop off their voice through illness to keep communicating using what sounds like their own voice .
OpenAI has some example of the AI - yield sound recording and the mention audio on its website , and we ’re sure you ’ll concord that they ’re pretty extraordinary .
