Hearing aid having an adaptive classifier

A method of selecting one of at least two modes of operation for a signal processing sub-system for a hearing aid including a signal processor processing an electric input signal according to audio processing parameters of the hearing aid, said method including:

• connecting a personal communication device to the hearing aid by means of a short range data communication link;
• analyzing in a classifier multiple specific characteristics of said electric input signal statistically in order to determine the acoustic environment of the hearing aid, said statistical analysis including comparison of said specific characteristics to one or more thresholds;
• automatically selecting one of said at least two modes of operation for a signal processing sub-system according to the classifiers classification;
• offering the user by means of said personal communication device a user interface for controlling and interacting with a program selector component of the hearing aid;
• generating a notification according to the operation of the user interface and transmitting the notification to said hearing aid via said short range data communication link;
• and
• adjusting at least one of said one or more thresholds used by the classifier component upon reception of the notification from said personal communication device;
• wherein said classifier is based upon said multiple specific characteristics handled in a multi-dimensional feature space for classification of the auditory environment, and mutates adaptively over time from a standard setting to a personalized setting based on the user experience and feed-back.

A hearing aid communicating with the personal communication device via a short range data communication link, and including:

• a signal processor processing an electric input signal according to audio processing parameters of the hearing aid;
• a classifier component analyzing multiple specific characteristics of said electric input signal statistically, said statistical analysis including comparison of said multiple specific characteristics to one or more thresholds;
• a program selector component automatically selecting one of at least two modes of operation for a signal processing sub-system according to said statistical analysis performed by the classifier component;
• said signal processor being adapted to receive a notification from said personal communication device and initiated by a user operating a user interface allowing the user to control said program selector component;
• and
• said signal processor adjusting said one or more thresholds of the classifier component upon reception of the notification from said personal communication device;
• wherein said classifier is based upon said multiple specific characteristics handled in a multi-dimensional feature space for classification of the auditory environment, and mutates adaptively over time from a standard setting to a personalized setting based on the user experience and feed-back.

A hearing aid including:

• a signal processor processing an electric input signal according to audio processing parameters of the hearing aid;
• a feature extractor for extracting value representations relating to at least one specific characteristic for said samples of said electric input signal;
• a classifier component analyzing said value representations statistically by comparing said value representations to one or more thresholds for said at least one specific characteristic;
• wherein the signal processor is furthermore analyzing the statistical distribution of said value representations for evaluating the homogeneity of the auditory environment,
• wherein the signal processor, upon recognizing that the auditory environment as being homogeneous, is adapted to compare the mean value of the distribution to said one or more thresholds for said at least one specific characteristic;
• and
• wherein the signal processor, upon recognizing that the difference between the mean value and one of said one or more thresholds is below a first predetermined value, adjusts said one of said one or more thresholds so that said difference at least corresponds to said first predetermined value.

A method of classifying an acoustic environment for a hearing aid including a signal processor processing an electric input signal according to audio processing parameters of the hearing aid, said method including:

• extracting in a feature extractor value representations relating to at least one specific characteristic for samples of said electric input signal;
• analyzing in a classifier component said value representations statistically by comparing said value representations to one or more thresholds for said at least one specific characteristic;
• analyzing in the signal processor the statistical distribution of said value representations for evaluating the homogeneity of the auditory environment,
• comparing in the signal processor the mean value of the distribution to said one or more thresholds for said at least one specific characteristic when the signal processor has recognized the auditory environment as being homogeneous;
• and
• adjusting said one of said one or more thresholds so said difference at least corresponds to said first predetermined value when the signal processor has recognized that the difference between the mean value and one of said one or more thresholds is below a first predetermined value.