Consumer Voice Authentication: the Australian Tax Office's example
How do you feel about the government maintaining a database of voice samples of everyone? If you're not against the practice, then Australia might be the place for you.
A new voice recognition system in use by the Australian Tax Office (ATO) to identify customers. The system asks taxpayers to repeat a phrase in order to compare it to a voiceprint by which they can prove their identity. The system is expected to save up to 45 seconds per phone call. The ATO receives about 8 million calls each year, 6 million of which require identification checks, for a total of about 3125 days, or 75,000 hours, each year spent performing identity checking.
On the one side, that's interesting tech, was probably a fun database project for some team, and saves money. What's not to like? On the other hand, you can bet that the NSA and similar agencies are also employing some version of this technology. So, who would like to step up to the microphone?
[Author note: On the flip side, it could be useful for people to record telemarketers so we can build a database of the people who spam us with annoying phone calls.]
A new voice recognition system in use by the Australian Tax Office (ATO) to identify customers. The system asks taxpayers to repeat a phrase in order to compare it to a voiceprint by which they can prove their identity. The system is expected to save up to 45 seconds per phone call. The ATO receives about 8 million calls each year, 6 million of which require identification checks, for a total of about 3125 days, or 75,000 hours, each year spent performing identity checking.
On the one side, that's interesting tech, was probably a fun database project for some team, and saves money. What's not to like? On the other hand, you can bet that the NSA and similar agencies are also employing some version of this technology. So, who would like to step up to the microphone?
[Author note: On the flip side, it could be useful for people to record telemarketers so we can build a database of the people who spam us with annoying phone calls.]
What happens in the future when they have enough people on record to identify them in other areas? Compare voice prints against recorded phone calls to match exactly who is speaking perhaps?