Voice Biometrics Analysis Microservice Architecture
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current voice biometrics systems in cloud-based environments are inefficient due to their session-based approach, which requires all data to be processed by a single engine, leading to reduced flexibility and resource utilization, and results in wasted computational power if not enough speech is present.
Innovation Solution
A two-phase analysis system where audio packets are processed by multiple micro-services, with a real-time authentication manager sending audio packets to a voice processor for feature extraction, and then to voice biometrics engines for quality grading and voiceprint generation, allowing for parallel processing and efficient resource allocation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If a session-based approach is used where all data is processed by a single designated engine, then data consistency is maintained, but system flexibility and resource utilization are reduced
Solution Approach 1:
The system divides the monolithic session-based processing into independent microservices (audio packet receiver, audio feature calculator, voiceprint generator, fraud detector, authenticator). Each microservice handles specific tasks independently, eliminating the need for complex session management while maintaining data consistency through shared access to the repository of audio features and voiceprints.
Solution Approach 2:
The repository of audio features and voiceprints serves as a universal data source that can be accessed by multiple microservices simultaneously. This allows different engines to process different aspects of the same audio data without requiring session-based exclusivity, thereby improving system flexibility and resource utilization.
2Reliability
If voiceprint generation is performed on all audio packets, then comprehensive authentication is achieved, but computational resources are wasted when insufficient speech is present
Solution Approach 1:
The system performs preliminary analysis by storing audio features in the repository before voiceprint generation. The audio feature calculator continuously processes audio packets and stores features in the repository, so when sufficient speech is detected, the voiceprint generator can immediately use the pre-computed features without re-processing the audio data, saving computational resources.
Solution Approach 2:
The system monitors the accumulated speech duration and automatically determines when sufficient data is available for voiceprint generation. This self-regulating mechanism ensures voiceprints are generated only when necessary, avoiding waste of computational resources on insufficient speech while maintaining authentication reliability.
3Productivity
If multiple voice biometrics engines are deployed, then processing capacity is increased, but data must be routed to a single designated engine reducing efficiency
Solution Approach 1:
The system segments the processing tasks among multiple specialized microservices instead of using multiple general-purpose voice biometrics engines. Each microservice handles specific functions (audio reception, feature extraction, voiceprint generation, authentication, fraud detection), allowing parallel processing without complex data routing since each service independently accesses the shared repository.
Solution Approach 2:
The repository of audio features and voiceprints acts as an intermediary data structure that decouples the microservices from direct data exchange. This eliminates the need for complex routing logic to direct data between engines, as each microservice independently retrieves or stores data in the repository based on its specific needs.
Data Source
AI summary
In a system and method for audio analysis in a cloud-based computerized an authentication (RTA) manager micro-service may send an audio packet to a voice processor micro-service. The voice processor may extract features of the audio. The RTA manager may obtain the extracted features from the voice processor; calculate, based on the extracted features, a quality grade of the audio packet, and send the extracted features to an at least one voice biometrics engine if the quality grade is above a threshold. Each of the at least one voice biometrics engines may be configured to generate a voiceprint of the audio packet, based on the extracted features of the audio packet and to perform at least one of: authenticate a speaker, detect fraudsters, and enrich a previously stored voiceprint of the speaker with the voiceprint of the audio packet.