Speech Anaphora Resolution via Speaker Context Tracking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current speech processing systems face challenges in resolving anaphora, exophora, and other ambiguous references across multiple devices and conversations, leading to difficulties in understanding user intents and executing commands accurately.
Innovation Solution
A speech processing system that utilizes speaker identification, device identification, and contextual information to determine the intended entities in utterances, employing ASR and NLU modules to process audio inputs and resolve ambiguous references by linking them to previous exchanges and environmental context.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If speaker identification and contextual information are used to resolve anaphora, then accuracy of command execution is improved, but device complexity increases
Solution Approach 1:
The patent introduces speaker identification as an intermediary mechanism that bridges the gap between audio input and contextual understanding. By identifying the speaker and using this information to select appropriate context, the system resolves anaphoric references more accurately without requiring complex real-time analysis of all possible contextual factors simultaneously.
Solution Approach 2:
The system performs preliminary speaker identification and context selection before processing the actual command. This preliminary action allows the system to pre-determine which contextual information is relevant, reducing the complexity of subsequent anaphora resolution while maintaining high accuracy in command execution.
2Reliability
If contextual information is linked to previous exchanges, then understanding of user intent is improved, but processing time increases
Solution Approach 1:
The patent applies local quality by selecting and processing only the specific contextual information relevant to the current speaker and situation, rather than processing all available historical context. This targeted approach maintains accurate intent understanding while significantly reducing processing time by focusing computation on locally relevant data.
Solution Approach 2:
The system changes the parameter of context selection based on speaker identity. By using speaker information as a filtering parameter, the system dynamically adjusts which contextual information is retrieved and processed, optimizing the balance between understanding accuracy and processing speed for different speaking scenarios.
Data Source
AI summary
A speech-processing system configured to determine entities corresponding to ambiguous words such as anaphora (“he,”“she,”“they,” etc.) included in an utterance. The system may associate incoming utterances with a speaker identification (ID), device ID, and other data. The system then tracks entities referred to in utterances so that if a later utterance includes an ambiguous entity reference, the system may take the speaker ID, device ID, etc. from the ambiguous reference, along with the text of the utterance and other data, and compare that information to previously mentioned entities (or other entities that may be relevant) to identify the entity mentioned in the ambiguous statement. Once the entity is determined, the system may then complete command processing of the utterance using the identified entity.