Hacker News new | ask | show | jobs
by cwmma 9 days ago
I (41 man with very deep voice) told it in a higher then normal voice that I was a kid, and it said I was under 18. I suspect if I was a kid that used a deep voice to say that I was an adult it would also fool it.

No offense but this is probably worthless is an adversarial environment.

3 comments

All good! Posting here to get the opinions and feedback from engineering peers.

> I... told it in a higher then normal voice that I was a kid, and it said I was under 18...

Good. If you're pretending to be a kid, you should be blocked. The launch is for sites that allow adults only (should have made that more clear in the post).

Deep voiced teens are THE HARDEST. What I commented earlier about the amount of data available in a clip of audio, that's real - teens have unique signatures that only show in their cohort.

Thanks for testing!

post puberty boys have absolutely have overlap with over 18 adults, if for no other reason then kids can go through puberty at different ages, so some 18 years olds have voices that are just dropping while others have had their voice drop for 6+ years.

You aren't going to be able to find a set of signatures that keeps out a useful number of <18s that doesn't also flag >18s

You are correct with inferring it's a scale, completely.

Politely disagree with "You aren't going to be able to find a set of signatures that keeps out a useful number..."

That's AGEWARDEN's job and the last year of my life has been diving deep into audio data to find those exact signals.

Good call out and thank you for the pushback :)

What do you consider the appropriate false negative/fall positive rate for voices of people aged 16-20? Since you don't give a confidence rating to the determination it's on the service to figure out what it's comfortable with.
Reliable Age Verification is nearly impossible and your antecdote gives weight to that assertion. Your comment brought my mind to voice actors. They easily change their voices. Deep fakes also come to mind.

So what is a solution? Parents. What about those that don't have parents? Surely they have guardians. Adults who guide the next generation instead of delegating them to technology.

> Your comment brought my mind to voice actors. They easily change their voices. Deep fakes also come to mind.

Voice actors, deep fakes, replay attacks (recordings from a TV/computer/phone) - mitigated. As in my earlier comment where teen voices have a signature, these all have unique signatures as well. The ongoing part is staying up-to-date with the latest synthetic (deep fakes/AI generated) as those are getting better each day.

How do you mitigate kids recording their parent's voices and playing them back to the AI?
That would be considered a "replay attack" and recorded audio carries a unique signature: BLOCKED
> I suspect if I was a kid that used a deep voice to say that I was an adult it would also fool it.

There should exist frequencies impossible to reach for anyone but post-pubescent men.

Of course that would make it teen-age verification at best.

> There should exist frequencies impossible to reach for anyone but post-pubescent men.... Of course that would make it teen-age verification at best.

Yes and no. Like geology, the more you dig into audio, the more is uncovered and available.

A child/teen with a deep voice has very unique perturbations that don't exist in other cohorts across feature sets. Adults may share some characteristics, but not at the levels found in specific age groups.

What about women, though? Do they not get to age verify themselves? :P
Women also undergo voice change, just less pronounced.

Now the problem turns into being able to tell men from women.

EDIT: this of course isn't a serious proposition.

Then again I wonder what effect this would have implemented on e.g. a porn site. Women also watch porn, but with enough tuning I think it shouldn't produce too many false positives - if only due to the demographic of the users.

Surely the vocal cords and larynx of an adult woman are on average larger than those of a 10yo boy?

The AGEWARDEN system is ^^ agnostic and answers a simple question only: Is this human over or under 18?

From my research, each person is incredibly unique :)