Speech Input Component Selection with Variable Touch Areas

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current speech recognition systems face challenges in allowing users to quickly and reliably correct incorrectly recognized voice inputs, particularly in environments where user attention cannot be diverted, such as in vehicles, as existing methods require precise selection or awareness of recognition probabilities.

Innovation Solution

A method and device that process voice inputs to determine recognition probabilities for each component, adjusting the selection parameter based on these probabilities, allowing easier selection of low-probability components through a touchscreen interface with variable touch areas, enabling users to correct errors without precise targeting.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If the user is presented with all recognized components equally for selection, then the user can select any component for correction, but the user cannot quickly identify and select components with low recognition probability, leading to increased selection time and user distraction

Engineering Contradiction:
Improveaccuracy of speech recognitionVSAvoidtime for user to locate and correct errors
Core Design Contradiction:
ReliabilityVSLoss of time

Solution Approach 1:

The patent applies visual highlighting to components with low recognition probability, making them stand out from correctly recognized components. This allows users to quickly identify potential errors without having to carefully examine each component, thereby reducing the time needed to locate and correct speech recognition errors while maintaining high accuracy.

Inventive Principle:
Principle #32Color changes

2Ease of operation

If the touch area for selecting components is made uniform, then the interface is simple to implement, but components with low recognition probability are difficult to select, requiring precise user interaction that increases distraction

Engineering Contradiction:
Improvesimplicity of selection interfaceVSAvoiduser distraction from road
Core Design Contradiction:
Ease of operationVSObject-generated harmful factors

Solution Approach 1:

The patent implements variable touch areas where components with low recognition probability are assigned larger selection areas than components with high recognition probability. This local differentiation allows users to easily select potentially erroneous components with larger target areas, reducing the precision required for interaction and minimizing user distraction, especially in safety-critical environments like vehicle operation.

Inventive Principle:
Principle #3Local quality

3Device complexity

If all components are displayed with the same visual prominence, then the display is simple and clean, but the user cannot quickly identify components that need correction

Engineering Contradiction:
Improvecomplexity of display systemVSAvoidspeed of error correction
Core Design Contradiction:
Device complexityVSProductivity

Solution Approach 1:

The patent uses visual highlighting to differentiate components based on their recognition probability. Components with low recognition probability are highlighted to draw user attention, enabling rapid identification of potential errors. This approach maintains display simplicity while significantly improving the speed of error correction by guiding user attention to the most likely candidates for correction.

Inventive Principle:
Principle #32Color changes

Data Source

PatentEP3113178B1Method and device for selecting a component of a speech input
Publication Date: 2018.04.11 VOLKSWAGEN AG
  • EP3113178B1 patent drawingFigure 1
  • EP3113178B1 patent drawingFigure 2~3

AI summary

The invention relates to a method for selecting a component (13) of a speech input. In the method according to the invention, the speech input is captured, wherein the speech input comprises several components (13). The captured speech input is processed, and a text is recognized. For each component (13) of the speech input, a recognition probability is determined, which indicates the probability with which the component (13) has been correctly recognized. The recognized text of the speech input is displayed on a screen (3), and for each component (13), the value of a parameter for selecting the component (13) is set by means of an input device (4, 9) depending on the determined recognition probability. Furthermore, the invention includes a device (1, 10) for selecting a component (13) of a speech input.