Cross-device Voiceprint Recognition via Mapping Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Differences in hardware between microphone devices lead to uneven sound collection qualities, resulting in lower accuracy of voiceprint recognition, especially when a voice control is performed using one device but the voiceprint is registered on another, causing misidentification of users.
Innovation Solution
Establishing a voiceprint mapping model between different devices through collecting and processing voice data to create a mapping function that maps a voiceprint from one device to another, using deep learning to improve recognition accuracy while protecting user privacy by not transmitting raw voice data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If voiceprint recognition is performed using different microphone devices, then user identification can be achieved across devices, but hardware differences cause uneven sound collection quality leading to lower recognition accuracy
Solution Approach 1:
The patent introduces a voiceprint mapping model as an intermediary that bridges voiceprints from different devices. The mapping model transforms voiceprints collected by devices with different hardware characteristics into a unified representation space, enabling accurate cross-device recognition without requiring raw voice data transmission. This mediator resolves the conflict between cross-device adaptability and recognition precision by adapting voiceprint representations rather than raw signals.
Solution Approach 2:
The patent changes the parameter space of voiceprint representation through the mapping model. Instead of working with raw voice signals that are highly sensitive to hardware differences, the system transforms voiceprints into a standardized parameter space where device-specific variations are normalized. This parameter transformation enables accurate comparison and identification across different microphone devices.
2Measurement precision
If raw voice data is transmitted for voiceprint recognition, then recognition accuracy can be improved, but user privacy is compromised
Solution Approach 1:
The patent extracts only the essential voiceprint features from raw voice data and transmits these extracted features for recognition. By separating the voiceprint extraction from the raw voice data transmission, the system achieves accurate recognition while minimizing privacy exposure. Only the compressed voiceprint representation is transmitted, not the original voice recording that could reveal sensitive information.
Solution Approach 2:
The patent creates and transmits a copy of the voiceprint representation rather than the original voice data. This copy contains the necessary identification information but lacks the contextual content of the original speech. The mapping model generates this copied representation that can be used for recognition without exposing the user's actual voice recordings or sensitive spoken information.
3Measurement precision
If voiceprint mapping model is established through collecting and processing voice data, then recognition accuracy across devices is enhanced, but device load increases
Solution Approach 1:
The patent segments the voiceprint recognition process into distinct stages: voiceprint extraction at the local device, mapping model application for cross-device adaptation, and recognition decision-making. This segmentation allows computationally intensive operations to be distributed appropriately, reducing the processing load on individual devices while maintaining high recognition accuracy through the mapping model.
Data Source
AI summary
According to an embodiment, an electronic device is provided. The electronic device includes: at least one processor; and a memory comprising instructions, which when executed, control the at least one processor to: receive a voice instruction of a user at the electronic device; transmit information regarding the voice instruction to a control device for identifying the user by mapping to a first voiceprint which is registered by another electronic device, a second voiceprint of the voice instruction of the user based on a voiceprint mapping model; and perform an operation corresponding to the voice instruction upon the identification of the user.