Accent recognition by AI

[Madman1981]

Hi,
I've sent this audio clip to the AI and asked whether the speaker is American or not. The AI's response is below. Did it answer correctly?"
(I've did not put the direct link to the audio and instead put the link to the page because it is possible that your device download the audio from the direct link automatically, which you may not want.)

After listening to the clip, the speaker does not sound American. The accent leans more toward a non-native English speaker with possibly a European or Middle Eastern background — though the pronunciation is clear and quite understandable.

 

[probus]

While perhaps it sounds more American than British, the problem is that it doesn't sound like natural human speech.at all. It sounds like it was generated by AI.

 

[PeterCW]

While perhaps it sounds more American than British, the problem is that it doesn't sound like natural human speech.at all. It sounds like it was generated by AI.

I would say American text to speech software.

 

[güey]

the problem is that it doesn't sound like natural human speech.at all.

It sounds quite human-like AE to me. Now, in 2025, we know that AI can generate natural-sounding speech with perfect intonations, realistically imitating human features like breathing patterns, lip smacking -- all of which this one does -- even occasional stumbles and hesitations.

Ten years ago I would've been sure this one was impeccable human speech.

 

[mrmvp]

While perhaps it sounds more American than British, the problem is that it doesn't sound like natural human speech.at all. It sounds like it was generated by AI.

I agree the speech is connected smoothly with no short pauses. I think the text was reproduced by AI.

@Madman1981 regarding the AI response about Middle Eastern background of the speaker, I can confirm that the speaker isn't from Arabic background at all. I know exactly what sounds are difficult to pronounce and I have the experience to specify the country the speaker from.

 

[Madman1981]

It sounds quite human-like AE to me. Now, in 2025, we know that AI can generate natural-sounding speech with perfect intonations, realistically imitating human features like breathing patterns, lip smacking -- all of which this one does -- even occasional stumbles and hesitations.

Ten years ago I would've been sure this one was impeccable human speech.

Good point. We can find the history of the page where this audio is located by searching its link on the Archive website. I’m currently unable to do this (in Iran, a VPN is needed for access). You said ten years ago, I believe that even six years ago it would not have been possible to generate such a speech with AI.

 

[jutfrank]

Did it answer correctly?"

No. The voice there is an AI-generated, female, American English voice.

 

[emsr2d2]

No. The voice there is an AI-generated, female, American English voice.

I agree entirely. It sounds like a slightly quieter version of the AI-generated voice on my car's satnav.

 

[Skrej]

After listening to the clip, the speaker does not sound American. The accent leans more toward a non-native English speaker with possibly a European or Middle Eastern background — though the pronunciation is clear and quite understandable.

AI is flat out wrong. It's very clearly text-to-speech engine meant to sound like a generic American accent. There's nothing remotely Middle Eastern or "European" in it (whatever that's supposed to mean). The only accent is a digital one.

"European" as a reference to accent is so vague as to be meaningless anyway. There are so many different languages spoken across Europe it is ludicrous to assume you can pick one as representative of the entire continent.

'Middle Eastern' is vague as well, but at least Arabic is spoken so widely throughout much of the Middle East one could at least assume it's referring to an Arabic accent.