Cough Detection via Self-Supervised Audio Embeddings
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing cough tracking systems are cumbersome, expensive, and unreliable, limiting their effectiveness in remote monitoring and clinical diagnosis.
Innovation Solution
A method using an audio feature set derived from speech samples to detect cough episodes and generate metrics automatically, implemented in computer devices like smartphones, for de-identified cough data collection and characterization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional cough tracking equipment (vests, neck mics, chest straps) is used, then cough detection capability is achieved, but device complexity and cost increase significantly
Solution Approach 1:
The patent applies universality by enabling smartphones and computers to perform cough detection in addition to their primary functions. The existing audio recording capability of these devices is leveraged for health monitoring purposes, eliminating the need for specialized equipment while maintaining reliable cough detection capability.
Solution Approach 2:
The patent uses the audio recording function of existing devices as a copy of specialized cough detection equipment. Instead of requiring dedicated medical devices, the system replicates cough detection functionality using the microphone and audio processing capabilities of consumer electronics, thereby reducing device complexity and cost.
2Reliability
If traditional cough tracking systems are used, then cough detection is possible, but cost increases significantly
Solution Approach 1:
The patent employs inexpensive consumer electronics (smartphones, computers) that are already widely owned by patients, replacing expensive medical-grade cough detection equipment. These devices are mass-produced at low cost and can be disposed of or replaced easily, significantly reducing the overall system cost while maintaining adequate detection capability.
Solution Approach 2:
By utilizing the existing audio recording functionality of universally owned devices, the patent eliminates the need for expensive specialized equipment. The same device serves multiple purposes including communication, entertainment, and health monitoring, thereby reducing the marginal cost of adding cough detection capability.
3Ease of operation
If self-reporting method is used for cough monitoring, then data collection is simple, but reliability decreases significantly
Solution Approach 1:
The system applies self-service by automatically detecting and recording cough events without requiring patient intervention or subjective reporting. The audio processing algorithm autonomously identifies cough patterns in recorded audio, eliminating human error and bias while maintaining operational simplicity for the user.
Solution Approach 2:
The patent replaces the mechanical/manual process of self-reporting with an automated electronic detection system. Audio signals are processed through algorithms that objectively identify cough events, substituting human judgment with machine-based acoustic analysis to improve measurement reliability while keeping the interface simple.
Data Source
AI summary
A method of detecting a cough in an audio stream includes a step of performing one or more pre-processing steps on the audio stream to generate an input audio sequence comprising a plurality of time-separated audio segments. An embedding is generated by a self-supervised triplet loss embedding model for each of the segments of the input audio sequence using an audio feature set, the embedding model having been trained to learn the audio feature set in a self-supervised triplet loss manner from a plurality of speech audio clips from a speech dataset. The embedding for each of the segments is provided to a model performing cough detection inference. This model generates a probability that each of the segments of the input audio sequence includes a cough episode. The method includes generating cough metrics for each of the cough episodes detected in the input audio sequence.