Local Keyword Engine Memory Allocation for Voice Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing voice-assisted media playback systems rely on cloud-based keyword recognition, which may compromise user privacy and experience delays due to the need for remote processing of voice commands, and lack efficient local processing capabilities for voice control of smart devices.
Innovation Solution
Implementing a local keyword engine within network microphone devices that uses machine learning techniques, such as neural networks, to detect and process voice commands locally, reducing the need for remote data transmission and enhancing privacy by processing audio inputs for recognized keywords within a limited library, thereby enabling quicker responses to voice inputs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If cloud-based keyword recognition is used, then the system can process voice commands remotely, but user privacy is compromised and processing delays occur
Solution Approach 1:
The patent extracts the keyword recognition functionality from the cloud and implements it locally within the network microphone device. The local keyword engine includes a neural network model that processes audio data on-device, eliminating the need to transmit voice commands to remote servers while maintaining accurate keyword detection capabilities
Solution Approach 2:
The network microphone device performs self-processing of audio data through its integrated local keyword engine. The device autonomously detects and processes keywords using machine learning models without requiring external cloud services, enabling independent operation while preserving privacy and reducing latency
2Productivity
If cloud-based keyword recognition is used, then the system can process voice commands remotely, but processing delays occur due to remote data transmission
Solution Approach 1:
The patent removes the data transmission step by implementing keyword recognition locally within the network microphone device. The local keyword engine processes audio data on-device, eliminating the round-trip time required for cloud communication and enabling immediate response to voice commands
Solution Approach 2:
The system performs preliminary audio processing and keyword detection locally before any potential cloud communication would occur. The local keyword engine analyzes audio data in real-time on-device, preparing and responding to commands without waiting for remote processing, thus maximizing response speed
Data Source
AI summary
Network microphone devices configured to detect keywords can include microphones for capturing sound samples. Features can be extracted from the sound samples by storing the sound samples in a first portion of a dynamic-access memory block, performing first computations based on spectral coefficients of the sound samples using a second portion of the memory block, and storing results of the first computations as extracted features in a third portion of the memory block. The second and third portions of the memory block can be designated as temporary memory. The extracted features are then processed using a neural network by storing the extracted features in a fourth portion of the memory block, performing second computations on the extracted features using the temporary memory, the second computations comprising computing at least one layer of the neural network, and storing an output of the neural network as a classification in the temporary memory.