Multi Hotword Voice Command Detection in Mobile Devices
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current speech recognition systems in computing devices require manual activation or detection of a hotword, which can lead to inefficiencies and false triggering, especially when users need to speak additional input to launch applications or perform actions.
Innovation Solution
Implementing a 'multi hotword' mode where the device listens for specific activation phrases to invoke applications, using a restricted language model for initial detection and a wider model for subsequent speech recognition, allowing for seamless application launch and reduced false triggers.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If a single hotword detection mode is used to activate speech input, then the system can be simple to implement, but it causes false triggering and inefficiency when users need to speak additional input to launch applications
Solution Approach 1:
The patent segments the speech detection process into two distinct modes: a first speech detection mode that listens for activation phrases, and a second speech detection mode that processes application-specific commands. This segmentation allows the system to differentiate between activation intent and command intent, reducing false triggering while maintaining manageable complexity through structured organization of detection functions.
2Reliability
If a restricted language model is used for activation phrase detection, then false triggering is reduced, but the system cannot recognize diverse speech inputs for application commands
Solution Approach 1:
The patent implements dynamic switching between language models based on operational context. The restricted language model is used during the first speech detection mode for accurate activation phrase recognition, while the unrestricted language model is activated during the second speech detection mode to handle diverse application-specific commands. This dynamic adaptation resolves the contradiction by adjusting model complexity according to the current task requirements.
3Productivity
If manual activation is required to speak additional input for application launch, then processing power consumption is reduced, but user convenience and productivity decrease
Solution Approach 1:
The patent employs periodic action by switching between two distinct operational modes: the first speech detection mode that consumes minimal processing power while listening for activation phrases, and the second speech detection mode that engages the more resource-intensive unrestricted language model only when an activation phrase has been detected. This periodic switching ensures that high-power processing is activated only when necessary, thereby improving productivity without excessive energy consumption.
Data Source
AI summary
In a first speech detection mode, a computing device listens for speech that corresponds to one of a plurality of activation phrases or “hotwords” that cause the computing device to recognize further speech input in a second speech detection mode. Each activation phrase is associated with a respective application. During the first speech detection mode, the computing device compares detected speech to the activation phrases to identify any potential matches. In response to identifying a matching activation phrase with a sufficiently high confidence, the computing device invokes the application associated with the matching activation phrase and enters the second speech detection mode. In the second speech detection mode, the computing device listens for speech input related to the invoked application.