Voice Announcement System Speech-to-Text Error Correction

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Public announcement systems face challenges in maintaining audio clarity, especially in emergency situations where users may speak quickly and excitedly, leading to potential errors and increased power consumption in wireless implementations due to high data rates, causing interference issues.

Innovation Solution

A high intelligibility voice announcement system that converts spoken announcements to text using a speech-to-text engine, identifies errors with an error recognition engine, allows user correction, and transmits corrected text to speakers for audible broadcast, reducing bandwidth requirements and power consumption.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If a high data rate is used to transmit audio to maintain high audio legibility, then audio clarity is improved, but power consumption increases and interference issues arise

Engineering Contradiction:
Improveaudio legibilityVSAvoidpower consumption
Core Design Contradiction:
Measurement precisionVSUse of energy by moving object

Solution Approach 1:

The patent extracts only the essential information from spoken announcements by converting speech to text and identifying key error patterns, rather than transmitting the complete high-fidelity audio signal. This selective extraction maintains intelligibility while dramatically reducing data rate requirements and power consumption.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The system creates a text-based copy of the spoken announcement instead of transmitting the original audio signal. This text representation preserves the essential information content while occupying minimal bandwidth and requiring minimal power for transmission and processing.

Inventive Principle:
Principle #26Copying

2Measurement precision

If a high data rate is used to transmit audio to maintain high audio legibility, then audio clarity is improved, but transmission interference increases

Engineering Contradiction:
Improveaudio legibilityVSAvoidtransmission interference
Core Design Contradiction:
Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The patent extracts only the essential information from spoken announcements by converting speech to text and identifying key error patterns, rather than transmitting the complete high-fidelity audio signal. This selective extraction maintains intelligibility while dramatically reducing data rate requirements and power consumption.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The system creates a text-based copy of the spoken announcement instead of transmitting the original audio signal. This text representation preserves the essential information content while occupying minimal bandwidth and requiring minimal power for transmission and processing.

Inventive Principle:
Principle #26Copying

3Speed

If the user speaks quickly and excitedly in emergency situations, then response time is improved, but announcement accuracy deteriorates

Engineering Contradiction:
Improveresponse timeVSAvoidannouncement accuracy
Core Design Contradiction:
SpeedVSManufacturing precision

Solution Approach 1:

The system provides immediate visual feedback by displaying the transcribed text to the user after speech-to-text conversion. This allows the user to review the accuracy of their quickly spoken announcement and make corrections if needed, ensuring accuracy is maintained even when speaking rapidly during emergencies.

Inventive Principle:
Principle #23Feedback

Solution Approach 2:

The system performs speech-to-text conversion and error identification in advance of the final broadcast, allowing time for review and correction of quickly spoken announcements before they are transmitted over the public address system.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP3032531B1High intelligibility voice announcement system
Publication Date: 2019.10.09 HONEYWELL INTERNATIONAL INC
  • EP3032531B1 patent drawingFigure 1
  • EP3032531B1 patent drawingFigure 2
  • EP3032531B1 patent drawingFigure 3~4

AI summary

A high intelligibility voice announcement system is described herein. One system includes an announcement station containing a speech to text engine in which a spoken announcement from a user is converted to text data, wherein any errors are identified and marked in the converted text data and displayed to the user. After the user has corrected any errors, the corrected text data is transmitted to one or more speakers, where a text to speech engine converts the text data to an audible message that is then broadcast via the speakers.