Anyone know how many hours of training data it takes to build up a convincing model of someone’s voice? It was 10’s of hours when I did a bit of research a year ago… the article says social media is the likely source of training data for these scams, but that seems unlikely at this point.
The technology has clearly come a long way in a short time, really fascinating.
I remember the first examples I read about being trained with celebrity read audiobooks because they needed so much audio data. I want to say Tom Hanks or Anthony Hopkins but I could have that confused with something else.
A current state of the art ai model from Microsoft can achieve acceptable quality with about 3 seconds of audio. Commercially available stuff like eleven labs about 30 minutes. But quality will obviously vary heavily but then again they’re using a low quality phone call so maybe not that important
That’s downright scary :-) I think it took longer in the last Mission Impossible.
30 minutes is still pretty minimal for the kind of targeted attack it sounds like this is used for. I suppose we all need to work with our families on code words or something.
I went in thinking the article was a bit alarmist, but that’s clearly not the case. Thank for the insight.
With that little, they may be able to recreate the timbre of someone’s voice, but speech carries a multitude of other identifiers and idiosyncrasies that they’re unlikely to get with that little audio, like personal vocabulary (we don’t choose the same words and phrasings for things), specific pronunciations (e.g. “library” vs “libary”), voice inflections, etc. Obviously, the more training data you have, the better the output.
Anyone know how many hours of training data it takes to build up a convincing model of someone’s voice? It was 10’s of hours when I did a bit of research a year ago… the article says social media is the likely source of training data for these scams, but that seems unlikely at this point.
I don’t remember the exact number but I did see an article recently that said it was videos on social media like you surmised.
And it was a pretty minimal amount of data needed. Definitely not tens of hours. Less than one hour iirc.
Is it safe to assume that if you don’t have any family that posts videos to Facebook/socials you are in a safer place?
You are safe only if you don’t have any people at all whom you trust.
But then you are having some other problems…
Make sure to use a fake accent when talking to strangers on the phone
I certainly am hoping so myself.
The technology has clearly come a long way in a short time, really fascinating.
I remember the first examples I read about being trained with celebrity read audiobooks because they needed so much audio data. I want to say Tom Hanks or Anthony Hopkins but I could have that confused with something else.
A current state of the art ai model from Microsoft can achieve acceptable quality with about 3 seconds of audio. Commercially available stuff like eleven labs about 30 minutes. But quality will obviously vary heavily but then again they’re using a low quality phone call so maybe not that important
That’s downright scary :-) I think it took longer in the last Mission Impossible.
30 minutes is still pretty minimal for the kind of targeted attack it sounds like this is used for. I suppose we all need to work with our families on code words or something.
I went in thinking the article was a bit alarmist, but that’s clearly not the case. Thank for the insight.
With that little, they may be able to recreate the timbre of someone’s voice, but speech carries a multitude of other identifiers and idiosyncrasies that they’re unlikely to get with that little audio, like personal vocabulary (we don’t choose the same words and phrasings for things), specific pronunciations (e.g. “library” vs “libary”), voice inflections, etc. Obviously, the more training data you have, the better the output.
ElevenLabs only needs 1 minute, but it also works with even shorter clips.
I literally just cloned someone’s voice for a presentation on AI and did it using maybe 30 total minutes of audio…
Took me about an hour and it was free. Hardest part was clipping the audio to get the ‘good bits.’
The voice was absolutely convincing.
Wow! That’s really impressive.
TorToiSe can work off of just three ten second clips when you’re using a pre-trained model. No telling if that’ll sound any good.
I’ll have to check that out, thanks for the link.