[Discuss] Fidelity voice-recognition security?

[Richard Pieri]

On 11/22/2017 10:42 PM, Robert Krawitz wrote:
> With that much leeway, there's more chance for collision, right?

It depends on a lot of factors. Leeway -- the degrees of deviations
allowed for a match -- is just one of the more easily quantifiable factors.

https://www.scientificamerican.com/article/human-voices-are-unique-but-were-not-that-good-at-recognizing-them/

Any biometric system can be spoofed. This is as much a tautology as "any
password can be cracked". The difference, ideally, is that a specific
password can be cracked by anyone with sufficient power but spoofing a
specific voice requires a willing twin sibling with similar enough
habits (eating, drinking, smoking, exposure to atmospheric pollution,
injuries or lack thereof, etc) to force a match. I don't see (hear?)
voice spoofing to be a credible threat except in rare circumstances. Or
financial executives cutting corners on security in order to maximize
their personal wealth.

-- 
Rich P.