Speech production and the management/prediction of hearing loss

Abstract

A method, including obtaining hearing data and speech data for a statistically significant number of individuals, and analyzing the obtained hearing data and speech data using a neural network to develop a predictive algorithm for hearing loss based on the results of the analysis, wherein the predictive algorithm predicts hearing loss based on input indicative of speech of a hearing impaired person who is not one of the individuals.

What is claimed is: 
     
         1 . A method, comprising:
 obtaining hearing data and speech data for a statistically significant number of individuals; and   analyzing the obtained hearing data and speech data using machine learning to develop a predictive algorithm for hearing loss based on the results of the analysis, wherein the predictive algorithm predicts hearing loss based on input indicative of speech of a hearing impaired person who is not one of the individuals.   
     
     
         2 . The method of  claim 1 , wherein:
 the predictive algorithm is not focused on a specific speech feature.   
     
     
         3 . The method of  claim 1 , wherein:
 the action of analyzing the obtained hearing data machine-trains a system that results in the developed predictive algorithm.   
     
     
         4 . The method of  claim 1 , wherein:
 of the obtained hearing data and speech data, a portion thereof is used, in a neural network, for training and a portion thereof is used, in the neural network, for verification.   
     
     
         5 . The method of  claim 1 , wherein:
 the machine learning develops the predictive algorithm by internally identifying important features from the obtained hearing data and speech data.   
     
     
         6 . The method of  claim 1 , wherein:
 the predictive algorithm utilizes an complex arrangement number of features present in the speech data to predict hearing loss.   
     
     
         7 . The method of  claim 1 , further comprising:
 obtaining biographical data for the individuals of the statistically significant number of individuals;   analyzing the obtained hearing data and speech data and the obtained biographical data using a neural network to develop a predictive algorithm for hearing loss based on the results of the analysis, wherein the predictive algorithm predicts hearing loss based on input indicative of speech and biographical data of a hearing impaired person who is not one of the individuals.   
     
     
         8 . A method, comprising:
 obtaining data based on speech of a person; and   analyzing the obtained data based on speech using a code of and/or from a machine learning algorithm to develop data regarding hearing loss of the person, wherein   the machine learning algorithm is a trained system trained based on a statistically significant population of hearing impaired persons.   
     
     
         9 . The method of  claim 8 , wherein:
 the code utilizes non-heuristic processing to develop the data regarding hearing loss.   
     
     
         10 . The method of  claim 8 , wherein:
 the data is developed without identified speech feature correlation to the hearing loss.   
     
     
         11 . The method of  claim 8 , wherein:
 the data regarding hearing loss of the person is an audiogram.   
     
     
         12 . The method of  claim 8 , wherein:
 the code of the machine learning algorithm is a trained neural network.   
     
     
         13 . The method of  claim 8 , wherein:
 the code is agnostic to the speech features.   
     
     
         14 . The method of  claim 8 , further comprising:
 obtaining biographical data of the person; and   analyzing the obtained data based on speech and the obtained biographical data using the code from the machine learning algorithm to develop the data regarding hearing loss of the person.   
     
     
         15 . A method, comprising:
 obtaining data based on speech of a person; and   developing a prescription and/or a fitting regime for a hearing prosthesis based on the obtained data.   
     
     
         16 . The method of  claim 15 , wherein:
 the prescription is developed based on relationships as opposed to correlations between speech and hearing loss.   
     
     
         17 . The method of  claim 15 , wherein:
 the prescription and/or fitting regime is a gain model.   
     
     
         18 . The method of  claim 15 , further comprising:
 utilizing a code written in the language of a neural network to develop the prescription and/or fitting regime.   
     
     
         19 . The method of  claim 15 , wherein the action of developing the prescription and/or the fitting regime is executed directly from the obtained data. 
     
     
         20 . The method of  claim 15 , further comprising:
 obtaining non-speech and non-hearing related data and developing the prescription and/or fitting regime based on the non-speech and non-hearing related data.