Speaker Verification Using Co-location Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Speaker verification systems face challenges in accurately identifying enrolled users in environments with multiple potential impostors, leading to high false acceptance rates, especially in mobile devices where the number of possible impostors is constantly changing and unbound.
Innovation Solution
The implementation of co-location information and imposter speaker models, which utilize APIs available on mobile devices to adjust similarity thresholds and share enrolled speaker models among devices, allowing for improved verification decisions by normalizing scores and reducing false positives.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional speaker verification systems are used without co-location information, then the system is simple to operate, but the false acceptance rate is high in environments with multiple potential impostors
Solution Approach 1:
The patent combines co-location information from multiple devices with traditional speaker verification. Multiple devices detect the presence of enrolled users and share this location data, merging it with audio verification results to improve overall system reliability and reduce false acceptance rates.
Solution Approach 2:
The system uses multi-functionality by utilizing both audio processing capabilities and location detection capabilities of mobile devices. The same device serves multiple purposes: recording audio for verification and detecting presence through sensors, thereby improving verification accuracy without requiring separate dedicated hardware.
2Reliability
If co-location information from multiple devices is used, then false positives are reduced by 60 to 80%, but the device complexity increases
Solution Approach 1:
The patent introduces a server as an intermediary that receives audio data and co-location information from multiple devices, processes this information centrally, and returns verification results. This mediator approach allows complex multi-device coordination without requiring direct peer-to-peer communication between all devices, thereby managing system complexity.
Solution Approach 2:
The system implements feedback loops where devices continuously share co-location data and verification results with both the server and each other. This real-time feedback mechanism allows the system to dynamically adjust verification decisions based on current presence information from multiple sources, significantly reducing false positives.
3Reliability
If imposter speaker models are used, then the ability to reject impostor utterances is improved, but the loss of information increases due to score normalization
Solution Approach 1:
The patent changes the parameters used for verification by incorporating co-location data alongside audio features. Instead of relying solely on audio similarity scores, the system adjusts verification parameters to include presence detection data, enabling better impostor rejection while maintaining information integrity through multi-parameter evaluation.
Data Source
Figure 1A~1C
Figure 2
Figure 3
AI summary
Methods, systems, and apparatus, including computer programs encoded on computer storage media, for identifying a user in a multi-user environment. One of the methods includes receiving, by a first user device, an audio signal encoding an utterance, obtaining, by the first user device, a first speaker model for a first user of the first user device, obtaining, by the first user device for a second user of a second user device that is co-located with the first user device, a second speaker model for the second user or a second score that indicates a respective likelihood that the utterance was spoken by the second user, and determining, by the first user device, that the utterance was spoken by the first user using (i) the first speaker model and the second speaker model or (ii) the first speaker model and the second score.