Privacy-Enabled Voice Assistant Audio Path Switching
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Users of voice-assisted intelligent automated assistants, such as smart speakers, are concerned about privacy breaches due to the recording and storage of their speech data without direct control, as these devices often rely on cloud-based services for speech recognition and processing, leading to potential unauthorized access and data misuse.
Innovation Solution
A user interface device, like a smart speaker, is configured to implement a privacy mode where microphone and speaker functions are available locally, but audio data is not transmitted to remote intelligent automated assistant services, using a switch element to bypass the remote system and ensure local handling of audio data, thereby preventing undesired recording and storage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If audio data is transmitted to remote intelligent automated assistant services for speech recognition and processing, then speech recognition accuracy and natural language processing capabilities are improved, but user privacy security deteriorates due to potential unauthorized access and data misuse
Solution Approach 1:
The patent segments the audio data processing into two distinct paths: a local processing path that handles speech recognition and natural language processing privately on the user's device, and a remote service path that can be optionally accessed for additional capabilities. This segmentation allows users to maintain privacy while still accessing advanced features when desired, resolving the contradiction between speech recognition accuracy and privacy security.
Solution Approach 2:
The patent introduces a local processing intermediary (the user's device) that acts as a mediator between the microphone input and the remote intelligent automated assistant service. This intermediary processes audio data locally first, ensuring privacy protection, and only transmits necessary data to remote services when needed, thereby maintaining both speech recognition accuracy and user privacy security.
2Adaptability or versatility
If cloud-based services are used for speech recognition and processing, then processing capabilities and intelligence are improved, but user control over audio data deteriorates
Solution Approach 1:
The patent implements dynamic user control where users can switch between local-only processing mode and hybrid mode with optional cloud services. This dynamic approach allows users to adjust their level of control and privacy based on their needs, maintaining both processing capability and user control by enabling them to choose when to engage remote services.
Solution Approach 2:
The patent enables self-service by allowing users to independently control their own audio data processing preferences and settings. Users can configure their device to handle speech recognition locally without external intervention, maintaining full control over their data while still accessing advanced processing capabilities when they choose to enable remote services.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
A user interface device for an intelligent automated assistant service uses a plurality of audio data paths that can be selectively switched to operate in a privacy mode. There is provided a first audio data input path between the audio input subsystem and an internal buffer; a first audio data output path between the internal buffer and the first communications subsystem; a second audio data output path between the internal buffer and the second communications subsystem; and a first audio data bypass path between the audio input subsystem and the second communications subsystem. In a first state, the first audio data input path directs audio data to the internal buffer, and to the first and second audio data output paths. In a second state, the first audio data input path is terminated and audio data is directed to the second communications subsystem on the first audio data bypass path.