OpenAI claims its software can clone your voice from 15 seconds of you talking OpenAI's latest trick needs just 15 seconds of audio of someone speaking to clone that person's voice – but don't worry, no need to look behind the curtain, the biz wants everyone to know it's not going to release this Voice Engine until it can be sure the potential for mischief has been managed. Described as being a "small …
"Described as being a "small model" that uses a 15-second clip and a text prompt to generate natural-sounding speech resembling the original vocalist"
I just wonder how close they can generate a specific accent, like Geordie, Scouse etc. from a mere 15 seconds of speech. Although I guess that using Siri or Alexa they'll have hours of eavesdropped speech to use. We should really start worrying when it can deduce how you pronounce "scone" without having heard you ordering tea in Fortnums.
"3 shampla gutha sintéiseithe ginte ó ionchur 15 secs...dáiríre, scanrúil" ... I'm confident that Brendan Behan (icon) would have said "I am an AI atheist." (in a Youth Custody Centre these days) ... So these days Brendan would have had to be writing a book titled "The Youth Custody Centre Boy" but an AI voice would never have made it sound as unique a "Borstal Boy"
There are other AI models out there with similar claims which do work to varying degrees, so I would rate it on the same level of credibility as Google announcing that they've launched a spreadsheet program.
Where it likely will go wrong is if for example an English person provides a voice recording and asks it to convert it to Donald Trump's voice, it could sound like Donald Trump speaking in an English accent.
Also if there is a word that a particular person always says in a weird way, like for example Nigella Lawson's meekrowavey, it is probably not going to replicate that, even if it is in the sample.
The article may have been on April 1st, but it is dated after midday GMT+1.
This bit however is pure comedy:-
the biz wants everyone to know it's not going to release this Voice Engine until it can be sure the potential for mischief has been managed
Like if they care, or there is any way to put this back in the box.
This post has been deleted by its author
Just make such tools illegal too.
"Since then the FCC has formally declared AI-generated robocalls to be illegal, and the FTC has issued a $25,000 bounty to solicit ideas on how to combat the growing threat of AI voice cloning."
How about making and commercialising in any way such tools a criminal offence with jail time for all willingly involved up to the top? I am well aware it won't stop all abuse, but it will stop all claims of "but we just use this tool that is legally available so we thought what we did was legal" and "but we just provide this tool to companies and organization that promise they will follow the law and have strong procedures to avoid abuse" and makes any and all detected usage very easy legally qualifiable as illegal and criminal without the need of 10 years of legal battle of the giants to get a slap on the wrist only.
Can I claim my $25,000 now?
The option of "just don't have the technology" is always considered, suggested, superficially functional, and impossible. The same way that "don't have an internet" didn't work when the first abuses were known. The same way that, when there were ten computers in the world and people's ideas of what they could do came from science fiction stories, fear of them did not mean that we just decided to ban them and keep going with manual methods. You can try to ban developing the technology, but it won't stop people, especially as multiple open source versions already exist. They have valid uses, and anyone making one will say they're intending those, whether they actually are or not. If you ban it in one country, it will just be developed in another one.
You can only try to ban a technology when it is prohibitively difficult to develop it, and even that doesn't always work as demonstrated by the number of countries that have or could develop nuclear weapons. Those take a lot of money and things that are hard to just buy, and yet programs to do so have succeeded. Someone can build a voice cloning tool on a home computer, even though it won't be as good as an organized corporate effort. You won't be able to do very much to prevent that.
This post has been deleted by its author
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
