Advantages

• 
  High Voiceprint Recognition Accuracy
  Original voiceprint recognition algorithm, high recognition accuracy, and multiple technological patents
• 
  Anti-synthetic Speech Attack
  Quickly distinguishes voice of real persons from recordings to effectively prevent against synthetic speech and recording attacks
• 
  Cross-channel Validation
  Allows multi-terminal access and provides cross-channel validation through networks and telephones
• 
  Best Practices in the Financial Industry
  Has passed the practical verification by Ping An Group in a large number of financial scenarios

Typical Business Scenarios

• Identity Verification of Call Center Agents
• Anti-financial Fraud
• Identify Verification for Remote Signature in Person
• X-list Database

Overview

Ping An Bank Credit Card Center uses the 1:1 identity verification technology to initiate 1:1 registration/verification to determine whether a call is dialed by a customer under his or her own identity. The voiceprint-based identity verification has reduced the average call time of the credit card customer service by 5 seconds, and improved the work efficiency by 2.5%, with the passing rate of voiceprint verification at 86% and the verification success rate at 96.67%.

Value Proposition

The lengthy Q&A identity verification during telephone transactions is removed, which effectively simplifies the verification process and improves the efficiency of agents’ work and customer experience.

Overview

In order to increase the risk control dimensions in the loan approval process and reduce the risk of non-performing loans, Ping An Bank has extracted the characteristics of recordings of users with confirmed on-performing loan records to establish a voiceprint blacklist database. Upon a call by a customer, 1:N blacklist database search and comparison will be initiated to confirm whether the customer is on the blacklist. When there is a high risk prompt, the agent can use other credit qualifications of the customer to conduct a review and consider whether to offer the loan. Since the use of the voiceprint technology, 43 cases per quarter have been confirmed on average, blocking unqualified loans of RMB 576,000. The approval line and across-business line have confirmed 1 case of agent counterfeiting.

Value Proposition

It enables customers to effectively prevent financial fraud, put an end to impostors, loans for financial blacklist users and falsely claiming insurance benefits, reduce financial risks for financial institutions, and strengthen financial risk control.

Overview

In order to improving customer identity verification in the remote signature in person of auto loan agreements, Ping An Bank’s auto financing business has, based on facial recognition and micro-expression recognition, introduced imperceptible voiceprint verification in phone calls with customers for comprehensive judgment of customer identity matching, thus assisting agents in remote signature in person and mitigating fraud.

Value Proposition

In business scenarios that need on-site verification, such as financial account opening and insurance withdrawal, you only need to conduct remote online verification through the unique voiceprint of each customer, thus avoiding face-to-face identity verification and improving service efficiency.

Overview

Ping An voiceprint solution has helped China Guangfa Bank establish its own databases of credit card voiceprint blacklists and whitelists by automating batch processing of the previous audio files stored by the bank and creating a new standard process for voiceprint collection. In the bank’s daily credit card business, it can use voiceprint data in the blacklist database for imperceptible caller identity verification. At present, the voiceprint database has more than 10 million voiceprints, and more than half of the callers have gone through voiceprint-based identity verification. It is estimated that the monthly unqualified loans blocked is more than RMB 500,000.

Value Proposition

You can effectively use the audio files in storage, establish and regulate the voiceprint collection process, and build your own voiceprint X-list databases.