Hierarchical environmental classification in a hearing prosthesis
Presented herein are techniques for generating a hierarchical classification of a set of sound signals received at hearing prosthesis. The hierarchical classification includes a plurality of nested classifications of a sound environment associated with the set of sound signals received at hearing prosthesis, including a primary classification and one or more secondary classifications that each represent different characteristics of the sound environment. The primary classification represents a basic categorization of the sound environment, while the secondary classifications define sub-categories/refinements of the associated primary classification and/or other secondary classifications.
1. A hearing device, comprising:
one or more input elements configured to receive a plurality of sets of sound signals over a period of time;
a memory; and
one or more processors coupled to the memory and to the one or more input elements, wherein the one or more processors are configured to:
process the plurality of sets of sound signals with a sound processing path to generate output signals for use in delivering stimulation signals to a recipient of the hearing device;
perform an environmental classification of a sound environment associated with each of a number of the plurality of sets of sound signals, wherein each environmental classification results in a classification output that represents the environmental classification for a corresponding set of sound signals; and
store each of the classification outputs in the memory to form a classification data set.
2. The hearing device of claim 1 , wherein the one or more processors are configured to:
analyze the classification data set; and
automatically adjust one or more settings of the sound processing path based on the analyzing of the classification data set.
3. The hearing device of claim 2 , wherein the one or more settings comprise at least one of: noise reduction settings, tinnitus masking settings, microphone settings, gain settings, channel dynamic range, maxima selection, or comfort settings.
4. The hearing device of claim 2 , wherein analyzing the classification data set comprises:
analyzing the classification data set with a machine learning algorithm.
5. The hearing device of claim 1 , wherein at least one environmental classification of at least one of the plurality of sets of sound signals includes a hierarchical classification that includes a primary classification and a secondary classification, the secondary classification sub-categorizing the primary classification.
6. The hearing device of claim 5 , wherein the primary classification indicates that the sound environment is an environment that includes speech, and wherein the secondary classification sub-categorizes the sound environment based on attributes of the speech.
7. The hearing device of claim 5 , wherein the primary classification indicates that the sound environment is an environment that includes noise, and wherein the secondary classification sub-categorizes a type of noise in the sound environment.
8. The hearing device of claim 5 , wherein the primary classification is a classification of at least one of speech or speech-in-noise, and wherein the secondary classification is at least one of own voice or external voice.
9. A non-transitory computer media comprising instructions that, when executed by one or more processors, are configured to cause the one or more processors to perform operations comprising:
processing a plurality of sets of sound signals received at a hearing device to generate output signals for use in delivering stimulation signals to a recipient of the hearing device;
performing an environmental classification of a sound environment associated with each of a number of the plurality of sets of sound signals, wherein each environmental classification results in a classification output that represents the environmental classification for a corresponding set of sound signals; and
storing each of the classification outputs to form a classification data set.
10. The non-transitory computer media of claim 9 , wherein performing the environmental classification of the sound environment generates a primary classification and a secondary classification, the secondary classification sub-categorizing the primary classification.
11. The non-transitory computer media of claim 10 , wherein the operations comprise:
conditionally generating the secondary classification based on an attribute of the primary classification.
12. The non-transitory computer media of claim 9 , wherein the operations comprise:
analyzing the classification data set; and
automatically adjusting one or more settings of a sound processing path used to generate the output signals based on the analyzing of the classification data set.
13. A method, comprising:
receiving a plurality of sets of sound signals at a hearing device over a period of time;
processing the plurality of sets of sound signals with a sound processing path to generate output signals for use in delivering stimulation signals to a recipient of the hearing device;
performing an environmental classification of a sound environment associated with each of a number of the plurality of sets of sound signals, wherein each environmental classification results in a classification output that represents the environmental classification for a corresponding set of sound signals; and
storing each of the classification outputs in memory to form a classification data set.
14. The method of claim 13 , further comprising:
analyzing the classification data set; and
automatically adjusting one or more settings of the sound processing path based on the analyzing of the classification data set.
15. The method of claim 14 , wherein the one or more settings comprise at least one of: noise reduction settings, tinnitus masking settings, microphone settings, gain settings, channel dynamic range, maxima selection, or comfort settings.
16. The method of claim 14 , wherein analyzing the classification data set comprises:
analyzing the classification data set with a machine learning algorithm.
17. The method of claim 13 , wherein at least one environmental classification of at least one of the plurality of sets of sound signals includes a hierarchical classification that includes a primary classification and a secondary classification, wherein the secondary classification sub-categorizes the primary classification.
18. The method of claim 17 , wherein the primary classification indicates that the sound environment is an environment that includes speech, and wherein the secondary classification sub-categorizes the sound environment based on attributes of the speech.
19. The method of claim 17 , wherein the primary classification indicates that the sound environment is an environment that includes noise, and wherein the secondary classification sub-categorizes a type of noise in the sound environment.
20. The method of claim 17 , wherein the primary classification is a classification of at least one of speech or speech-in-noise, and wherein the secondary classification is at least one of own voice or external voice.