Speech Sample Alignment for Physiological State Diagnosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current medical diagnostic systems for physiological conditions affecting speech lack effective methods to accurately assess changes in speech patterns over time, particularly in identifying onset or deterioration of conditions like congestive heart failure, COPD, or depression, using speech analysis.
Innovation Solution
A method and system that construct speech models from reference speech samples with known physiological states, using local distance functions and allowed transitions to map test speech samples to minimum-distance sequences, generating outputs indicating the current physiological state based on total distance calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If speech samples are analyzed using traditional methods, then the analysis process is simple, but the accuracy in identifying physiological state changes is insufficient
Solution Approach 1:
The speech signal is segmented into multiple feature vectors representing different temporal aspects of speech production. Each feature vector captures specific acoustic characteristics that can be independently analyzed to detect physiological state changes, thereby improving measurement precision through detailed segmentation of the speech signal.
Solution Approach 2:
The system transforms speech samples into multiple derived parameters including feature vectors, distance metrics, and temporal characteristics. By changing from raw speech analysis to analyzing multiple transformed parameters, the system achieves higher accuracy in detecting physiological state changes while managing complexity through systematic parameter transformation.
2Measurement precision
If multiple feature vectors are computed from speech samples to improve diagnostic accuracy, then the measurement precision improves, but the computational complexity increases
Solution Approach 1:
The system extracts only the most relevant features from speech samples that are directly indicative of physiological state changes. By selecting and computing only essential feature vectors rather than analyzing all possible speech characteristics, the system maintains high diagnostic accuracy while reducing unnecessary computational energy consumption.
Solution Approach 2:
The system computes a specific set of feature vectors that are sufficient for detecting physiological state changes without computing all possible speech parameters. This partial action approach provides enough information for accurate diagnosis while avoiding the excessive computational burden of complete speech analysis.
3Loss of time
If speech samples are collected and analyzed in real-time, then the response time is reduced, but the measurement precision may be compromised due to limited data
Solution Approach 1:
The system performs preliminary processing of speech samples by extracting key feature vectors and computing baseline characteristics in real-time. This preliminary action enables immediate detection of obvious physiological state changes while maintaining the option to perform more comprehensive analysis when needed, thus reducing response time without permanently compromising measurement precision.
Solution Approach 2:
The system continuously analyzes speech samples as they are collected, providing ongoing assessment of physiological state. This continuous useful action ensures that diagnostic information is available in real-time while accumulating data over time, allowing the system to maintain both rapid response and high measurement precision through continuous rather than intermittent analysis.
Data Source
AI summary
A method includes obtaining a first sequence of reference-sample feature vectors that quantify acoustic features of different respective portions of at least one reference speech sample, which was produced by a subject at a first time while a physiological state of the subject was known, and a second sequence of test-sample feature vectors that quantify the acoustic features of different respective portions of at least one test speech sample, which was produced by the subject at a second time while the physiological state of the subject was unknown. The test-sample feature vectors are mapped to respective ones of the reference-sample feature vectors, under predefined constraints, such that a total distance between the test-sample feature vectors and the respective ones of the reference-sample feature vectors is minimized. In response to the mapping, an output indicating the physiological state of the subject at the second time is generated.