Universal Speech Processing with User-Specific Intent Routing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current speech processing systems face challenges in efficiently configuring and processing user utterances across multiple domains, leading to resource wastage and complexity in identifying user intents and executing corresponding actions, as they require domain-specific training and parallel processing of multiple domains simultaneously.
Innovation Solution
A natural language processing system with a universal catalog of actions and an action manager that selects the appropriate speechlets based on user-specific data, history, and preferences, using a trained machine-learning model to navigate the execution of user intents, even activating multiple intermediate speechlets to obtain necessary data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If domain-specific training and parallel processing of multiple domains are used, then user intent identification accuracy is improved, but system complexity and resource consumption increase
Solution Approach 1:
The patent implements a universal speech processing system that can handle multiple domains through a single unified architecture. The system uses a universal intent classifier and common processing pipeline that serves all domains (music, news, weather, etc.), eliminating the need for separate domain-specific processing paths. This allows the system to maintain high accuracy across domains while reducing overall system complexity.
Solution Approach 2:
The patent segments the speech processing system into modular components: audio input, feature extraction, universal intent classifier, and domain-specific action executors. This segmentation allows the complex intent identification task to be distributed across independent modules, reducing the complexity of any single component while maintaining overall system accuracy.
2Measurement precision
If domain-specific training and parallel processing are implemented, then user intent identification accuracy is improved, but resource wastage occurs
Solution Approach 1:
The universal speech processing system reuses the same computational resources (feature extraction models, intent classification algorithms, processing pipelines) across all domains. Instead of maintaining separate resource-intensive domain-specific systems, a single universal system handles music, news, weather, and other domains, significantly reducing resource wastage while maintaining identification accuracy.
Solution Approach 2:
The patent merges previously separate domain-specific processing systems into a unified architecture. By combining the intent classification and processing pipelines into a single shared system that serves multiple domains, the patent eliminates redundant resource consumption while preserving the accuracy needed for diverse user intents.
3Reliability
If domain-specific configurations are used, then domain performance is optimized, but ease of configuring new domains decreases
Solution Approach 1:
The universal speech processing system provides a standardized configuration interface and common processing pipeline that can accommodate new domains without requiring complex reconfiguration. The system maintains domain-specific performance optimization through modular action executors while using a unified intent classification framework, making it easy to add new domains by simply registering new action handlers rather than building entire domain-specific systems.
Data Source
AI summary
A system configured to process an incoming spoken utterance and to coordinate among multiple speechlet components to execute an action of the utterance, where a trained model considers user history and preference information to select the primary speechlet to execute the action as well as any intermediate speechlets that may be provide input data to the speechlet that will ultimately perform the action. The trained model may also consider current dialog information, feedback data, or other data when determining how to process a dialog.