When a customer calls up ICICI Bank in coming weeks, in the 10 seconds that he introduces himself the bank's servers will identify his account details and authenticate his identity by picking a voice sample and matching it with the database. Not just that, in future voice data would be analyzed to identify whether the caller is agitated, irate or in a hurry to avoid cross-sell pitches.
ICICI Bank has become the first in the country and among the few in the world to deploy voice-recognition technology for biometric authentication. The new facility will enable customers to call up and ask for funds to be transferred to registered beneficiaries or bills to be paid without having to enter 16-digit numbers or keying in their PIN.
The cutting-edge technology comes from Nuance Communications, a Massachusetts-based firm that created the technology behind Siri - Apple's voice-recognition software. According to Rajiv Sabharwal, executive director, ICICI Bank, the technology will provide a big leg-up for financial inclusion as the process of authentication can be done seamlessly without requiring the caller to speak specific words on in any particular language.
"The software uses 100 parameters for matching sound samples. To collect the sample, the servers need to record only 35 seconds of voice during any conversation. Besides authentication, the software will help the bank in analytics as it will immediately inform the operator about the state of mind of the caller," said Sabharwal.
"If we were to use other biometrics like fingerprint or iris, we would need to deploy scanners. In the case of voice recognition, any call from a basic phone to the call centre is good enough. Since the voice samples and servers are in the call centre, there is no requirement for data connectivity," said Sabharwal.
Unlike speech recognition used in phones, voice recognition is a different technology altogether and has nothing to do with the language or contents of the speech. The voice print which helps identification is a hashed string of numbers and characters that represent how specific an individual's voice rates on a host of characteristics being measured.
According to Sabharwal, voice recognition would be the second factor authentication, the first being the mobile number itself. "To ensure safety, we are enabling this feature only on calls made from the customers' registered phone," he said.
He said that besides phone banking, there is potential to use voice recognition as a second factor in online. Instead of an OTP, the servers could call up the customer on his registered number and authenticate him with a voice sample.
Source : Economic Times
ICICI Bank has become the first in the country and among the few in the world to deploy voice-recognition technology for biometric authentication. The new facility will enable customers to call up and ask for funds to be transferred to registered beneficiaries or bills to be paid without having to enter 16-digit numbers or keying in their PIN.
The cutting-edge technology comes from Nuance Communications, a Massachusetts-based firm that created the technology behind Siri - Apple's voice-recognition software. According to Rajiv Sabharwal, executive director, ICICI Bank, the technology will provide a big leg-up for financial inclusion as the process of authentication can be done seamlessly without requiring the caller to speak specific words on in any particular language.
"The software uses 100 parameters for matching sound samples. To collect the sample, the servers need to record only 35 seconds of voice during any conversation. Besides authentication, the software will help the bank in analytics as it will immediately inform the operator about the state of mind of the caller," said Sabharwal.
"If we were to use other biometrics like fingerprint or iris, we would need to deploy scanners. In the case of voice recognition, any call from a basic phone to the call centre is good enough. Since the voice samples and servers are in the call centre, there is no requirement for data connectivity," said Sabharwal.
Unlike speech recognition used in phones, voice recognition is a different technology altogether and has nothing to do with the language or contents of the speech. The voice print which helps identification is a hashed string of numbers and characters that represent how specific an individual's voice rates on a host of characteristics being measured.
According to Sabharwal, voice recognition would be the second factor authentication, the first being the mobile number itself. "To ensure safety, we are enabling this feature only on calls made from the customers' registered phone," he said.
He said that besides phone banking, there is potential to use voice recognition as a second factor in online. Instead of an OTP, the servers could call up the customer on his registered number and authenticate him with a voice sample.
Source : Economic Times
0 comments:
Post a Comment