Joint-Encoder Audio Detection with Contextual Wakeword Decoders
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The increasing availability of multiple wakewords and assistant voice services in devices strains computing resources due to the need for multiple wakeword detectors, making it impractical to operate them simultaneously, especially in devices with limited resources.
Innovation Solution
A two-layered approach is implemented, where a joint encoder processes input audio data into feature vectors, and individual decoders determine specific wakewords, allowing devices to recognize multiple sounds simultaneously without overwhelming resources by activating and deactivating decoders based on context.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If multiple wakeword detectors are operated simultaneously to recognize multiple wakewords, then the device can detect multiple sounds simultaneously, but the computing resources are overwhelmed
Solution Approach 1:
The wakeword detection system is segmented into two distinct layers: a shared encoder that processes audio input into feature vectors, and multiple specialized decoders that each detect specific wakewords. This segmentation allows the computationally intensive encoding to be performed once, while the lighter decoding operations can be distributed across multiple wakeword-specific detectors, reducing overall resource consumption while maintaining multi-wakeword detection capability
Solution Approach 2:
The encoder component serves as a universal, multi-functional element that processes audio input for all wakeword detectors simultaneously. By making the encoder universal and shared across multiple decoders, the system avoids redundant processing and enables multiple wakeword detection capabilities without proportionally increasing computing resource usage
2Adaptability or versatility
If multiple wakeword detectors are deployed to increase detection capability, then the device can handle more assistant voice services, but the device complexity increases
Solution Approach 1:
The detection system is divided into modular components: a single shared encoder and multiple independent decoders. This segmentation creates a clear architectural structure where the complex encoding logic is centralized and reused, while the decoding logic is simplified and replicated. The modular design makes the system easier to manage and extend compared to having fully independent detectors for each wakeword
Solution Approach 2:
The encoder functionalities for multiple wakeword detectors are merged into a single shared component. By combining the encoding operations that would otherwise be duplicated across multiple detectors into one universal encoder, the system reduces architectural complexity while maintaining the ability to detect multiple wakewords through the shared feature extraction
Data Source
AI summary
A device is configured to detect multiple different wakewords. A device may operate a joint encoder that operates on audio data to determine encoded audio data. The device may operate multiple different decoders which process the encoded audio data to determine if a wakeword is detected. Each decoder may correspond to a different wakeword. The decoders may use fewer computing resources than the joint encoder, allowing for the device to more easily perform multiple wakeword processing. Enabling/disabling wakeword(s) may involve the reconfiguring of a wakeword detector to add/remove data for respective decoder(s). Specific decoders may be activated/deactivated depending on device context, thereby efficiently managing device resources.