Real-Time Speech Recognition and Correction for Deaf Communication
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Severely hearing impaired and profoundly deaf individuals struggle to communicate effectively due to their inability to hear themselves speak, leading to unintelligible speech and difficulties in understanding conversational partners, as traditional solutions like sign language and lip reading are limited or inaccurate.
Innovation Solution
A computer-based system with a processor, display, and storage that executes speech recognition to convert speech into text, allows for corrections, and converts corrected text back into speech signals, potentially integrated into devices like eyeglasses, mobile phones, or tablets, enabling real-time communication assistance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of information
If speech recognition is executed on distorted speech from hearing-impaired individuals, then communication intelligibility is improved, but accuracy of speech recognition deteriorates due to speech distortions
Solution Approach 1:
The system displays recognized text to the hearing-impaired speaker and allows them to review and correct the transcription. This feedback loop enables the speaker to verify accuracy and make corrections, compensating for the reduced accuracy of speech recognition on distorted speech while maintaining improved communication intelligibility.
Solution Approach 2:
The system introduces text as an intermediary representation between the distorted speech and the final communicated message. The speech is converted to text, which then serves as the reliable medium for communication, bypassing the accuracy limitations of direct speech recognition.
2Loss of information
If sign language is used for communication, then understanding between hearing-impaired and normal-hearing individuals is improved, but vocabulary limitation and expression capability deteriorate
Solution Approach 1:
The system replaces the mechanical hand movements of sign language with electronic speech synthesis and text processing. This substitution enables access to the full vocabulary and expression capabilities of spoken language while maintaining improved understanding between hearing-impaired and normal-hearing individuals.
3Loss of information
If lip reading is used for communication, then understanding of speech content is improved, but accuracy and usability deteriorate when speaker is not facing the deaf person
Solution Approach 1:
The system uses audio signals as an intermediary to capture speech content, eliminating the need for visual lip reading. This allows accurate speech understanding regardless of the speaker's orientation, as the microphone can capture audio from any direction.
4Measurement precision
If speech correction features are added to the system, then speech recognition accuracy is improved, but device complexity increases
Solution Approach 1:
The system enables the hearing-impaired speaker to self-correct the recognized text by displaying it and allowing direct input of corrections. This self-service approach improves speech recognition accuracy without requiring complex automated correction algorithms, thus limiting the increase in device complexity.
Data Source
AI summary
An application on a mobile phone or tablet computer or eyeglasses listens to the speech of a profoundly deaf person and interprets it in real time. It can then repeat it in a “normal voice”. The voice recognition is trained to the voice of the profoundly deaf person.