Voice Biometrics potentials in FSI
Voice biometrics often refers to speaker authentication- this is the technology that confirms if caller identity is the one, who claims. However, voice biometrics has more to offer which have been less investigated.
Voice biometrics has gained a lot of traction in the last few years in FSI. The technology has moved from the realm of science fiction into the risk management toolkit of an increasing number of enterprises. Most recently, Bank of Australia implemented the technology to prevent malicious account takeovers [1]. Tech giants have integrated voice biometrics into home assistant platforms so that online payments could be made just by voice with reliable authentication.
Voice biometrics often refers to speaker authentication- this is the technology that confirms if caller identity is the one, who claims. However, voice biometrics has more to offer which have been less investigated. These are mostly characterised by speaker identification technology.
Speaker authentication is a one-to-one matching. It meets the relatively simple problem statement “Is the person speaking, the person who claims to be?” People largely communicate with businesses via contact centres so use of voice in the verification process comes in handy. Authentication using voice is convenient, and done well can improve customer experience, and significantly reduce average handle time, drives efficiency, and reduces operational costs. As opposed to speaker authentication, speaker identification is a one to many (or none) matching problem. This is a more complex technology comparing to earlier one which is capable of monitoring audio files to identify some targeted voices in tonnes of conversations. Let’s review some offerings of technology in FSI.
Call centre conversations are a crucial data source in conduct investigations. Voice biometric can be employed to identify targeted agents, staff with a previous misconduct or a complaining customer to gather more evidence. Further, voice models could be trained to detect voice of targeted fraudsters in call centres to prevent further losses as imposter could be identified in seconds and agent notified immediately. Technology can also search old conversations for fraudsters voices to get more evidence, or to counteract further attacks.
Voice fingerprint can be searched in recorded calls to link conversations to individuals being customers or agents for further assessments for surveillance purposes. This is especially helpful for historic recordings where there is no metadata to determine speakers- which is quite the case for older call centre platforms.
The other area voice biometrics has potential is law enforcement and more specifically Anti-Money Laundering (AML)/Counter Terrorism Financing (CTF). The cost of AML/CTF incompliance is increasing as regulatory bodies tighten up mandated obligations [3]. Rapidly growing and complex regulations under their dynamic nature makes compliance quite challenging for financial sector. This put companies at risk of hefty penalties, loss of reputation, and can jeopardize long-term growth. Voice biometrics can proactively be employed to detect and monitor voice interactions with sensitive customers by using their voice prints. Such customer could be an individual with a suspicious activity or a Politically Exposed Person (PEP). Technology recognises target voice print, over phone calls or VOIP (Voice Over IP) to capture and link recordings to provisioned automated or manual risk assessments or to collect evidence to support investigations. Even synthesised voices could be tracked by this technology so there is less chance to escape from being identified.
Lastly in area of culture, ethics, and behavioural monitoring- audio communications of staff in internal and external conversations could be basis of individually curated training and remunerations or for audit and inspection purposes.
We are here to help
Deloitte Risk Advisory has developed a solid voice capability and have already delivered a couple of successful deliveries in this space. Please reach out and our team will be delighted to work with you to explore potentials in risk management.
References
[1] Bank of Australia https://www.bankaust.com.au/support/voiceid
[2] Consumer Remediation Draft https://asic.gov.au/about-asic/news-centre/find-a-media-release/2021-releases/21-307mr-asic-consults-on-consumer-remediation-draft-guidance/
[3] Anti-Money Laundering and Counter-Terrorism Financing ACT 2006, https://www.austrac.gov.au/business/legislation/amlctf-act
Author
Bahram Nezhad - Specialist Senior Manager, Financial Regulatory and Risk Insights
