AI Voices Indistinguishable from Human Ones, Study Finds

A study from Queen Mary University of London found that artificial intelligence-generated voices are now indistinguishable from human voices.

The study, published in the journal PLOS One, compared real human voices with AI-generated voices cloned to mimic a specific person and those created from large generic voice models. It found that the synthetic voices could sound as real as human voices, with many participants unable to tell them apart.

In the research, the cloned voices were mistaken for real 58 percent of the time and human voices were correctly identified only 62 percent of the time.

To test people’s ability to distinguish human voices from AI-generated ones, the researchers created 40 synthetic AI voices and 40 voice clones using ElevenLabs' AI voice generator tool.

The study also found that the synthetric voices were often perceived as more dominant and, in some cases, more trustworthy than their human counterparts.

"AI-generated voices are all around us now. We've all spoken to Alexa or Siri, or had our calls taken by automated customer service systems," Nadine Lavan, a senior lecturer in psychology at Queen Mary University of London and co-chair of the study, said in a statement. "Those things don't quite sound like real human voices, but it was only a matter of time until AI technology began to produce naturalistic, human-sounding speech. Our study shows that this time has come, and we urgently need to understand how people perceive these realistic voices."

Lavan highlighted the ease and rapidity with which the team could create voice clones using commercially available software. They each took roughly four minutes of voice recordings to create

"The process required minimal expertise, only a few minutes of voice recordings, and almost no money," Lavan added. "It just shows how accessible and sophisticated AI voice technology has become."

And despite the profound ethical concerns about misinformation, fraud, and impersonation, Lavan emphasized the potential of AI voice technology to offer exciting opportunities.

"There might be applications for improved accessibility, education and communication, where bespoke high-quality synthetic voices can enhance user experience," she said.