Phoneme-Based Reading Location Detection for Spoken Text

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Traditional virtual assistants struggle to accurately detect when a user has finished reading a text source aloud, leading to unnecessary resource consumption and potential recording of private conversations due to non-linear reading behaviors such as skipping, repeating, or adding content.

Innovation Solution

A computing device enhances detection by comparing audio data with text source data using phoneme analysis and fuzzy matching logic to determine correspondence measures, deactivating audio processing when a threshold is not met, thereby optimizing resource use and preventing private audio capture.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If traditional virtual assistants continuously analyze audio data to detect reading status, then they can respond to user commands, but they consume unnecessary resources and may record private conversations when users stop reading

Engineering Contradiction:
Improvereading status detection accuracyVSAvoidpower consumption
Core Design Contradiction:
ReliabilityVSUse of energy by moving object

Solution Approach 1:

The system performs preliminary phoneme extraction from audio data and compares it with phoneme data from the text source before making a final determination about reading status. This preliminary analysis allows the system to detect reading discontinuation more accurately without requiring continuous full-scale speech recognition processing, thereby reducing power consumption while maintaining detection reliability

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system continuously compares extracted phoneme data with expected phoneme data from the text source and uses this feedback to dynamically adjust its operation. When the comparison shows significant deviation beyond a threshold, the system determines reading has stopped and reduces processing, creating a feedback loop that optimizes resource usage based on actual reading status

Inventive Principle:
Principle #23Feedback

2Adaptability or versatility

If traditional virtual assistants use speech recognition to identify reading content, then they can understand user input, but they introduce time delays and consume additional computing resources

Engineering Contradiction:
Improvereading behavior adaptationVSAvoidprocessing delay
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The system extracts only phoneme data from the audio signal, which is the essential acoustic feature needed for reading detection, rather than performing complete speech recognition that would convert audio to text. This extraction approach maintains adaptability to different reading behaviors while significantly reducing processing time and computational resource consumption

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The system replaces the traditional speech recognition mechanism (which involves complex language modeling and text generation) with a phoneme-based acoustic analysis mechanism. This substitution maintains the ability to adapt to various reading behaviors while eliminating the time-consuming steps of full speech-to-text conversion

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

3Productivity

If traditional virtual assistants perform continuous speech analysis, then they can detect user commands, but they cannot accurately determine when users have finished reading due to non-linear reading behaviors

Engineering Contradiction:
Improvereading detection efficiencyVSAvoidreading discontinuation detection accuracy
Core Design Contradiction:
ProductivityVSMeasurement precision

Solution Approach 1:

The system segments the audio data into discrete phoneme units and compares each segment with the corresponding phoneme data from the text source. This segmentation allows the system to detect deviations caused by non-linear reading behaviors (skipping, repeating, adding content) at a granular level, improving detection precision without requiring continuous analysis of the entire audio stream

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4191563B1Determination of a story readers current reading location
Publication Date: 2025.10.15 GOOGLE LLC
  • EP4191563B1 patent drawingFigure 1
  • EP4191563B1 patent drawingFigure 2
  • EP4191563B1 patent drawingFigure 3

AI summary

A location in a sequence of words being read aloud is identified by determining phoneme data of a text source, receiving audio data comprising a spoken word associated with the text source, comparing the phoneme data of the text source and phoneme data of the audio data to identify the location being currently read.