Federated Local Remote Speech Recognition System
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Mobile devices with limited storage face challenges in efficiently processing complex speech recognition tasks due to the need for large data access, which can be exacerbated by the limitations of local storage and network latency in automatic speech recognition systems.
Innovation Solution
A system that federates local and remote speech recognition results by simultaneously sending audio data to both local and remote recognizers, allowing for the selection of the most likely recognition result based on grammar complexity and data availability, thereby improving response times and accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If speech recognition is performed locally on mobile devices, then response time is reduced, but accuracy decreases for complex grammars and large data requirements
Solution Approach 1:
The speech recognition system is segmented into local and remote components. The local recognizer handles simple, fast recognition tasks on the mobile device, while the remote recognizer handles complex tasks requiring large data sets. This segmentation allows the system to optimize for both speed and accuracy depending on the task complexity.
Solution Approach 2:
The system merges local and remote recognition results through a federation component that combines outcomes from both recognizers. By integrating results from both local fast processing and remote accurate processing, the system achieves both rapid response times and high recognition accuracy for complex grammars.
2Measurement precision
If large amounts of data are stored locally for speech recognition, then recognition accuracy improves, but device storage requirements increase
Solution Approach 1:
The system extracts the data storage burden from the mobile device and relocates it to a remote server. The local device retains only essential recognition capabilities, while large grammars and data sets are stored remotely and accessed as needed, reducing local storage requirements while maintaining recognition accuracy.
Solution Approach 2:
The remote recognizer serves multiple functions: it stores large data sets, processes complex grammars, and provides accurate recognition results for various applications. This universal remote resource eliminates the need for each device to store complete data sets locally, reducing storage requirements while maintaining accuracy.
3Quantity of substance
If speech recognition data is accessed over a network, then storage requirements are reduced, but network latency increases processing time
Solution Approach 1:
The system dynamically selects between local and remote recognition based on task requirements. For simple tasks, local recognition provides immediate responses without network latency. For complex tasks requiring large data sets, the system transitions to remote recognition, accepting the network latency as necessary for achieving accurate results.
Solution Approach 2:
The local recognizer is pre-configured with essential recognition capabilities and simple grammars, allowing it to handle common tasks immediately without network access. This preliminary local preparation eliminates network latency for routine operations while maintaining the option to access remote resources when needed.
4Adaptability or versatility
If only local recognition is used, then device independence is maintained, but recognition capability is limited by local resources
Solution Approach 1:
The federation component acts as an intermediary that coordinates between local and remote recognizers. It manages the complexity of integrating multiple recognition sources, handling result fusion, and selecting appropriate recognition paths, thereby enabling enhanced recognition capability without requiring the end device to directly manage complex distributed system architecture.
Data Source
AI summary
Techniques to provide automatic speech recognition at a local device are described. An apparatus may include an audio input to receive audio data indicating a task. The apparatus may further include a local recognizer component to receive the audio data, to pass the audio data to a remote recognizer while receiving the audio data, and to recognize speech from the audio data. The apparatus may further include a federation component operative to receive one or more recognition results from the local recognizer and/or the remote recognizer, and to federate a plurality of recognition results to produce a most likely result. The apparatus may further include an application to perform the task indicated by the most likely result. Other embodiments are described and claimed.