Speech Input Correction Using Timing and Device Acceleration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Determining whether a user intends to correct an error in a spoken input generated by automatic speech recognition without explicit indications, such as phrases like 'No' or 'I meant', is challenging, especially in the absence of clear cues in subsequent speech.
Innovation Solution
A media guidance application uses time differences and acceleration changes of a user input device to infer the user's intent to correct errors in speech recognition, by analyzing the time elapsed between presenting search results and subsequent speech, and the change in device acceleration during this period.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If automatic speech recognition is used to convert spoken input to text, then the system can process voice commands and search queries, but recognition errors occur without explicit correction cues
Solution Approach 1:
The system monitors user behavior feedback (time to speak correction, acceleration patterns) after search results are displayed to determine whether the user intends to correct the ASR input. This feedback loop allows the system to detect correction intent without requiring explicit correction phrases, thereby improving recognition accuracy while maintaining automation.
Solution Approach 2:
The system introduces intermediate signals (acceleration data from the user input device, timing information) as mediators between the ASR output and the correction mechanism. These intermediaries provide indirect evidence of user intent to correct, allowing the system to resolve recognition errors without direct explicit commands.
2Reliability
If the system waits for explicit correction phrases like 'No' or 'I meant', then correction intent can be clearly identified, but the user experience becomes less natural and more cumbersome
Solution Approach 1:
The system enables self-service correction detection by automatically analyzing user behavior patterns (timing, acceleration) to infer correction intent without requiring the user to provide explicit verbal cues. This makes the correction process more natural and effortless while maintaining reliable detection of user intent.
Solution Approach 2:
The system changes the parameters used to detect correction intent from explicit verbal phrases to implicit behavioral parameters (time elapsed since results display, acceleration magnitude). This parameter shift allows the system to detect correction intent more naturally through user actions rather than requiring specific verbal formulations.
3Ease of operation
If the system analyzes user behavior cues like time and acceleration to detect correction intent, then explicit correction phrases are not needed, but the complexity of the detection system increases
Solution Approach 1:
The system makes the user input device multi-functional by using it both for receiving speech input and for detecting acceleration patterns. This universal use of existing hardware components allows the system to gather correction intent data without adding separate dedicated sensors or devices, thereby managing complexity while improving ease of operation.
Solution Approach 2:
The system replaces the mechanical/verbal system of explicit correction phrases with an automated analysis system that processes timing and acceleration data. This substitution eliminates the need for users to learn specific correction phrases and allows the system to automatically infer intent from natural user behaviors.
4Productivity
If the system quickly processes and displays search results, then user productivity is improved, but the user may not have enough time to review results before intending to correct errors
Solution Approach 1:
The system performs preliminary action by quickly displaying search results based on the initial ASR input, allowing the user to immediately see whether correction is needed. The system then monitors subsequent user behavior (time to speak correction, acceleration patterns) to detect correction intent, enabling fast result delivery while still capturing correction opportunities.
Solution Approach 2:
The system maintains continuous monitoring of user behavior (timing, acceleration) after displaying results to detect correction intent. This continuous observation allows the system to bridge the gap between fast result delivery and user review time, capturing correction opportunities that arise naturally during the user's brief review period.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Methods and systems for correcting, based on subsequent second speech, an error in an input generated from first speech using automatic speech recognition, without an explicit indication in the second speech that a user intended to correct the input with the second speech, include determining that a time difference between when search results in response to the input were displayed and when the second speech was received is less than a threshold time, and based on the determination, correcting the input based on the second speech. The methods and systems also include determining that a difference in acceleration of a user input device, used to input the first speech and second speech, between when the search results in response to the input were displayed and when the second speech was received is less than a threshold acceleration, and based on the determination, correcting the input based on the second speech.