Passive Voice Print Generation from Historical Recordings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice authentication methods require active enrollment, which can be cumbersome and lead to low take-up rates, as individuals need to actively utter specific phrases multiple times, whereas passive enrollment methods that utilize historical recordings can reduce effort but may lack regulatory compliance and accuracy.
Innovation Solution
A system that generates a voice print by searching for a predetermined or repeated phrase in historical recordings, allowing for passive authentication without requiring positive action from the individual, using a combination of text-dependent and text-independent biometric techniques to create a voice print based on multiple utterances of a phrase.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If active enrollment methods are used where individuals must actively utter specific phrases multiple times, then authentication accuracy is improved, but ease of operation deteriorates leading to low take-up rates
Solution Approach 1:
The system performs preliminary action by automatically searching for and extracting phrases from historical voice recordings before authentication is needed. Voice prints are generated in advance from past communications, eliminating the need for active enrollment at the time of authentication. This resolves the contradiction by maintaining high accuracy through multiple phrase samples while dramatically improving ease of operation through automated background processing.
Solution Approach 2:
The system enables self-service by allowing individuals to be authenticated using their own historical voice data without requiring their active participation in an enrollment process. The automated phrase extraction and voice print generation occurs without human intervention, transforming a cumbersome active process into a seamless passive experience that maintains security and accuracy.
2Ease of operation
If passive enrollment methods utilizing historical recordings are used, then ease of operation is improved, but measurement precision and regulatory compliance deteriorate
Solution Approach 1:
The system applies feedback by automatically analyzing historical recordings to identify and extract phrases that meet specific criteria (repetition frequency, clarity, duration). The system provides feedback loops where voice prints are continuously refined using additional historical data, ensuring both regulatory compliance and high authentication accuracy while maintaining ease of operation through automated processing.
Solution Approach 2:
The system changes parameters by transforming raw historical voice recordings into structured voice print data through automated phrase extraction. By adjusting parameters such as phrase repetition thresholds, audio quality filters, and temporal patterns, the system optimizes both accuracy and compliance while keeping the enrollment process passive and user-friendly.
3Reliability
If multiple utterances of phrases are searched for in historical recordings, then voice print quality is improved, but loss of time increases due to extensive searching
Solution Approach 1:
The system performs preliminary action by pre-processing and indexing historical voice recordings to enable rapid phrase searching. Voice prints are generated in advance from available historical data, and the system maintains updated voice prints as new recordings become available. This resolves the time-quality contradiction by completing extensive searching beforehand while maintaining high voice print quality through multiple phrase samples.
Solution Approach 2:
The system applies dynamics by making the voice print generation process adaptive and ongoing rather than static. As new historical recordings become available, the system dynamically updates voice prints by searching for additional phrase utterances. This allows the system to balance processing time and voice print quality by continuously refining authentication data without requiring batch processing of all historical data at once.
Data Source
AI summary
A computer-implemented method for enrolling in a database voice prints generated from audio streams may include receiving an audio stream of a communication session and creating a preliminary association between the audio stream and an identity of a customer that has engaged in the communication session based on identification information. The method may further include determining a confidence level of the preliminary association based on authentication information related to the customer and if the confidence level is higher than a threshold, sending a request to compare the audio stream to a database of voice prints of known fraudsters. If the audio stream does not match any known fraudsters, sending a request to generate from the audio stream a current voice print associated with the customer and enrolling the voice print in a customer voice print database.