Take the test to see if you can distinguish real and AI VOICES

Take the test to see if you can distinguish real and AI VOICES,

Our voices are just about as unique as our fingerprints – so can you tell the difference between a real one from one that has been AI–generated? 

AI voice clones are a new generation of artificial voices that can recreate someone’s voice from just a few seconds of recorded speech.

A new study has revealed that these synthetic clones are actually clearer and easier to understand than the real person they’re based on.

Scientists from University College London initially expected that voice clones would be poor representations of human voices. 

However, what they found was the total opposite.

‘I thought initially that voice clones would be less intelligible because they were unfamiliar,’ said lead author Professor Patti Adank.

‘I found they were up to 20 per cent more intelligible, which was quite shocking.’

So, do you think you could tell the difference? Watch the video to hear some side–by–side comparisons of real and fake voices, then scroll down to the bottom for the answers. 

Scientists have found that AI–generated voice clones are easier to understand than their human counterparts, especially when there is lots of disruptive background noise 

In the past, voice assistants like Siri or the one in your satnav used so–called ‘synthetic voices’.

These require voice actors to spend hours in the recording studio, meticulously sampling all the different words and phrases that the assistant might need.

Voice clones, on the other hand, have revolutionised how synthetic voices are created, by using AI to digitally recreate someone’s speech patterns.

These clones can be created with as little as a few seconds of recorded audio, even using clips from social media or snippets of conversation as the raw material.

This has sparked concerns that criminals using AI could easily impersonate friends, family, or co–workers to manipulate their targets.

According to the National Trading Standards, criminals are already using AI to clone people’s voices and set up unauthorised direct debits over the phone.

In the study, the researchers created voice clones of human participants using just 120 pre–recorded sentences.

Participants listened to 80 unique sentences – 40 spoken by a real person and 40 spoken by an AI voice clone.

The researchers compared human (top) AI–generated (bottom) voice recordings to see why this might be the case, but couldn’t find any clear explanation 

They were asked to copy down exactly what they heard, so the researchers could see how easy the voices were to understand.

Each participant was then asked to rate how clear they thought the voice was, how strong the regional accent was, and whether or not they thought it was an AI.

To the scientists’ surprise, the AI–generated voices were consistently rated as easier to understand.

This was not something that previous research into AI–generated voices had suggested, and the researchers were baffled as to why it might be the case.

Professor Adank says: ‘A small part of our paper is talking about that experiment, and then a large part is me and my collaborator frantically trying to find out what it is that makes those voice clones more intelligible.’

The researchers repeated the experiment with elderly participants and a special filter designed to mimic the effects of a cochlear implant to see if being hard of hearing might have an effect.

They then tried the test again with Americans to see if the British accents of the voice clones would create any extra confusion.

However, no matter what the researchers did, the AI–generated clones were consistently rated 13 per cent more intelligible than their human counterparts.

What makes this especially unusual is the fact that the participants were rarely tricked by the AI clones.

When presented with a human and an AI voice, participants correctly chose the human 70.4 per cent of the time.

This means they were rating the AIs as being easier to understand, even though they knew that they were artificial.

After examining over 100 different acoustic measurements to try to determine what could be causing the strange effects, the researchers remain stumped.

Professor Adank now believes the only way to solve the mystery will be to work with the engineers who build voice clones to understand how the AIs really work.

She says: ‘I am now going to try and recreate [the effect] by studying how synthesisers work and how they use digital signal processing to generate those voices, just to get a bit of a handle on this.’