Tech

OpenAI’s Johansson gaffe pushes voice cloning into spotlight

Scarlett Johansson voiced an AI character within the movie ‘Her’, a film favored by Sam Altman.

OpenAI was pressured to apologize to actor Scarlett Johansson final week for utilizing her voice –- or one thing very comparable –- on its newest chatbot, throwing the highlight on to voice-cloning tech.

Though OpenAI denied the voice they used was Johansson’s, their case was not helped by CEO Sam Altman flagging the brand new mannequin with a one-word message on social media—”Her”.

Johansson voiced an AI character within the movie “Her”, which Altman has beforehand mentioned is his favourite movie in regards to the know-how.

Proper from the beginning, AI voice cloning has proved problematic.

Final 12 months, British agency Elevenlabs went viral for all of the unsuitable causes when it launched its voice-cloning software program.

Web pranksters instantly started pushing out deepfaked celebrities—Harry Potter star Emma Watson was proven studying Hitler’s Mein Kampf.

Regulation enforcement warned that AI clones could possibly be used to extort cash from family members over the cellphone.

The know-how has developed quickly previously 12 months changing into way more sensible and nuanced.

Danish entrepreneur Victor Riparbelli, CEO of British AI agency Synthesia, advised AFP it was largely all the way down to a program referred to as Tortoise that was launched two years in the past.

This system’s builders threw 1000’s of hours of voice knowledge into their mannequin in an unstructured means and found it not solely learnt what to say however find out how to say it.

“That was a pretty big paradigm shift,” Riparbelli mentioned on the sidelines of final week’s VivaTech convention in Paris.

Tortoise was an open supply program and Elevenlabs was the primary to go to market utilizing it.

OpenAI makes use of comparable techniques although they don’t launch any particulars.

‘Not superb’

A lot of the controversy round voice cloning has centered on issues over folks misusing the software program.

However the declare in opposition to OpenAI is uncommon as a result of it’s the firm itself accused of enjoying quick and unfastened.

“It was very unfortunate that OpenAI did that—really not very good,” Katya Laine, CEO of TALKR.ai, advised AFP at VivaTech.

“If they actually cloned her voice without her knowing then I think that’s very very bad,” mentioned Riparbelli.

The 2 entrepreneurs are amongst a whole lot harnessing AI voice packages for makes use of that they argue will make firms extra environment friendly.

Laine’s agency supplies digital voice assistants—basically AI customer support brokers.

She mentioned her agency’s system might now resolve 25 to 30 p.c of calls with none human involvement.

Synthesia makes a speciality of video avatars, which Riparbelli mentioned allowed any workplace employee to show textual content or slides right into a video carried out by a sensible AI.

Each Riparbelli and Laine enable their shoppers to make use of their very own avatars, off-the-shelf merchandise or these provided by the likes of OpenAI and Elevenlabs.

Riparbelli mentioned Synthesia used actors whose likenesses and voices had been licensed for 2 years with an choice to renew after the preliminary interval.

The issues come up if actors’ voices are used with out their consent.

‘Odd precedent’

The fiasco overshadowed a developer convention in Paris final week when OpenAI was exhibiting off a collection of recent instruments.

In entrance of a giant display screen in an auditorium, Romain Huet, OpenAI’s Head of Developer Expertise, breezily chatted into his cellphone.

Seconds later, his quick voice pattern had been processed and could possibly be heard commentating over a generated video—in 5 languages.

The demonstration confirmed how shortly the sector is transferring, however the headlines had already been written.

The Washington Submit requested in a publication “How dumb is OpenAI?”, different commentators had been suggesting wunderkind Altman was nothing greater than a huckster.

Nonetheless, Riparbelli was open to OpenAI’s argument that that they had used one other actor who simply sounded favored Johansson.

“If it’s not her but someone who sounds a lot like her… where do you draw that line,” Riparbelli requested.

“If they’re not allowed to use someone who sounds a lot like her, then it sets a very odd precedent.”

© 2024 AFP

Quotation:
OpenAI’s Johansson gaffe pushes voice cloning into highlight (2024, May 27)
retrieved 27 May 2024
from https://techxplore.com/information/2024-05-openai-johansson-gaffe-voice-cloning.html

This doc is topic to copyright. Other than any truthful dealing for the aim of personal research or analysis, no
half could also be reproduced with out the written permission. The content material is offered for data functions solely.



Click Here To Join Our Telegram Channel


Source link

In case you have any issues or complaints concerning this text, please tell us and the article will likely be eliminated quickly. 

Raise A Concern

Show More

Related Articles

Back to top button