DE102009043775A1 - Hearing device i.e. combined hearing and tinnitus masker device, adjusting method, involves analyzing speech signal for recognizing emotional state of user and adjusting parameter of hearing device as function of recognized emotional state - Google Patents

Abstract

The method involves detecting and recognizing a speech signal of a user. The signal is analyzed to recognize an emotional state of the user. Tension, stress or relaxation is recognized as the emotional state. A parameter of a hearing device i.e. combined hearing and tinnitus masker device (1), is adjusted as a function of the recognized emotional state. Selection of a hearing program, volume control, frequency response and/or noise suppression of the hearing device is adjusted as a function of the recognized emotional state. The parameter is transferred from an initial value to a final value. An independent claim is also included for a hearing aid comprising a signal source.

Description

The invention relates to a method for adjusting a hearing device, in particular a hearing aid and / or tinnitus masker. Furthermore, the invention relates to a corresponding hearing device with at least one signal source for providing a source signal, a signal processing unit for processing the source signal in response to at least one automatically adjustable parameter and for generating an electrical output signal, an output transducer for converting the electrical output signal into an acoustic output signal, an input transducer for receiving an acoustic input signal and conversion into an electrical input signal and a signal analysis unit for analyzing the electrical input signal and for automatically adjusting the parameter as a function of an analysis result.

Many users of hearing aids do not want to or can not adapt the signal processing by the hearing aid to the current hearing situation, but expect that a hearing aid automatically "optimal" 'sets. However, an "optimal" setting depends on a variety of factors.

To adapt a hearing aid to the current hearing situation, the user has various options available today. Usually, there are switches on the hearing aids themselves with which the volume can be changed or the program selection can be made. Furthermore, many manufacturers offer remote controls with advanced settings (volume, sound, etc). In order to meet the desire of the users for an automatic tracking of the settings to the current listening situation, classifiers are used, which analyze the microphone signal in order to estimate from which listening situation the user is currently located. The signal processing is then automatically adapted to the detected hearing situation.

Furthermore, there are recently hearing aids that try to recognize user preferences (learning ability) when the user has manually readjusted his hearing aid in certain listening situations. However, the recognition of the hearing situation is up to now in the program selection, which in turn requires a user action, or the input signal analyzed by the classifier, from which the actual user request can only be guessed.

From the publication EP 0 064 042 A1 a hearing aid is known which analyzes an acoustic input signal and, on the basis of the result of the signal analysis, recognizes the instantaneous hearing situation in which the hearing aid is located and adjusts the parameters relating to signal processing automatically as a function of the detected auditory situation.

Furthermore, from the published patent application DE 10 2005 046 168 A1 a hearing device is known in which by means of a sensor biometric data, such as pulse, temperature or skin resistance of a user are detected and a signal processing in the hearing aid related parameter is set automatically depending on the detected biometric data and derived therefrom mental state of the user. The disadvantage here is that a sensor for capturing the biometric data must be provided.

From the pamphlets DE 10 2005 032 274 A1 and EP 1 017 253 A2 Hearing aids are known in which an existing in an acoustic input signal speech signal is detected. In this case, a speech analysis in particular detects whether it is the user's own voice. If a plurality of speech signals are present in the input signal at the same time, they are separated by means of a blind-source separation (BSS) algorithm and optionally further processed independently of one another.

Tinnitus maskers are provided by the industry for tinnitus retraining therapy. These usually generate a white noise, which can be configured by a hearing care professional by means of suitable fitting software still in terms of its volume and spectral coloring and thus adapted to the individual needs of a person affected. The aim of the adaptation is to best mask the tinnitus while at the same time the lowest possible level of the masking signal.

The tinnitus noise, however, may vary in volume and sound, e.g. B. in noise or stress in volume increase. In this case, the tinnitus is no longer obscured by the preconfigured masker signal, but is audible to the tinnitus patient, possibly disturbing, and represents an additional burden for a hearing aid wearer in an already demanding situation. The therapy device, the tinnitus masker , does not seem to work in the perception of the tinnitus patient.

Known on-market systems have volume controls and the user can manually track the volume of the masking signal in critical listening situations.

From the publication EP 0 820 211 A1 Furthermore, a tinnitus masker is known in which the masking signal is automatically transmitted to the listening environment. in which the tinnitus masker is currently operated, can be adjusted.

In the field of pattern recognition the topic "Recognition of emotional user states with the help of acoustic and linguistic features (multimodal recognition of attention focus)" is currently a topic focus. First applications are already available on the market, for example the Skype extension "KishKish" or the "Voice Stress Analysis System" of the City of London detect lies in telephone conversations.

In science, the view has prevailed that there are six primary emotions: joy, sadness, fear, anger, surprise and disgust. In addition, other derived emotions can be distinguished. Current procedures separate speaker-independent (that is, without speaker-specific training) up to 20 mental states.

For example, from the document EP 2 063 416 A1 For example, a method for analyzing a speech signal is known by which the emotional state (state of mind) of the speaker is recognized.

The object of the invention is to improve the automatic adjustment options for a hearing device.

This object is achieved by a method with the method steps according to claim 1. Furthermore, the object is achieved by a hearing device with the features according to claim 9.

In the context of the invention, a speech signal is generally understood to be a signal generated by the human voice. These include not only the pure language but also other human sounds, vocals etc. generated by the voice.

The hearing device according to the invention is in particular a hearing aid. This has at least one microphone for receiving an acoustic input signal and conversion into an electrical input signal. The microphone is the signal source in the context of the invention. Preferably, the microphone is designed as a directional microphone for direction-dependent reception of acoustic input signals. It can be made up of a plurality of individual microphones and also the electrical input signal can comprise a plurality of parallel input signals, which can optionally be further processed separately. The electrical input signal (source signal) is processed and amplified in a signal processing unit to compensate for an individual hearing loss of the user. The signal processing is advantageously carried out as a function of the current listening environment in which the hearing aid is operated.

Furthermore, the hearing device according to the invention may also be a tinnitus masker. This comprises as a signal source a signal generator for generating a source signal or a memory in which the source signal is stored and can be read out for further processing. Again, the signal processing of the source signal depends advantageously on adjustable parameters.

The adjustable parameters in a hearing device according to the invention relate in particular to adjustments in terms of volume, sound or frequency response. In addition, however, a variety of other settings is possible, for example, whether filters are effective or not, whether a directional microphone is switched on, etc.

The basic idea of the invention is now generally to operate the setting of the parameter as a function of the emotional state of the user of the hearing device. In this case, the instantaneous emotional state is determined on the basis of a voice signal emitted by the user. This has the advantage that at least one microphone is usual for receiving an acoustic input signal in hearing devices and therefore usually no further sensor for receiving the speech signal is required. Sensors for recording biometric data (blood pressure, pulse, etc.) of the user, however, are unusual in hearing devices and therefore require additional hardware effort. Further, with a hearing aid formed in one piece and worn on or in the ear, it is difficult to acquire biometric data with a sensor integrated into the respective hearing device, since measurements on or in the ear are difficult. For example, blood pressure or pulse can only be measured if the sensor is pressed with a certain pressure on the tissue of the user. This can quickly become uncomfortable for the user in a hearing aid devices, longer wearing time.

In the implementation of the invention can be made of known algorithms for speech analysis. In this case, it is first necessary to detect the user's own voice in the case of a voice signal entering the hearing aid device. Here, too, can be used on known algorithms. If a plurality of speech signal sources are present, they can advantageously be separated by a Blind Source Separation (BSS) algorithm so that an analysis based on the speech signal emitted by the user can be used to detect the emotional state.

The manner in which certain parameters are changed as a result of a particular detected emotional state is advantageously determined during the programming of the hearing device, for example with a hearing aid acoustician. For example, if a very relaxed own voice is detected, the device is turned down or a special program for "rest" selected to support the relaxed mood. If, on the other hand, tension or stress is detected by means of an excited or aggressive voice, the amplification is boosted or ZEN sounds are played in for reassurance, if the user so wishes.

Advantageously, the hearing device according to the invention by the evaluation of the determined from one's own voice emotions in addition to the acoustic situation, which is already evaluated by the classifier, an additional indication from which the real acoustic focus of the user can be estimated. The likelihood increases that the hearing device is then automatically adjusted as the user desires.

The signal analysis for detecting the current emotional state of the user is preferably adaptive and trainable. By way of example, current emotional states are given to the algorithm during a training phase by manual user input, so that the algorithm learns the assignment over time and can distinguish between different states even better. In voice analysis, the pitch is of particular importance. The detection of the individual mental state takes place - like the detection of the acoustic environment - preferably continuously.

Advantageously, the hearing device has a neural network for signal analysis.

Especially with a tinnitus masker, an advantage of the invention is to interpret tension and stress as tinnitus triggers and adaptively adjust the volume of the masker signal in this regard. If the mental state "stress" is detected, for example, the volume of the masker signal can be adaptively increased by × dB in order to effectively mask the tinnitus. In addition, a situation recognizer may be used to adaptively adjust the volume of the masker signal when the user has been in a noisy environment for a long time. This application, in its preferred embodiment, additionally takes into account values from a tone audiogram, in particular the auditory and discomfort threshold, and concludes thereon based on levels perceived as "loud". Advantageously, the level of the tinnitus masker is tracked only when the current acoustic situation is perceived by its wearer as "loud".

For the user, this application has the advantage that the tinnitus remains concealed even in critical situations, without having to manually make changes to the system on a control element and does not disturb it in uncritical situations. Another advantage is that once the device has learned the user preferences, no further user interaction is needed. The devices can therefore be built very small (= inconspicuous), since neither controls nor a wireless coil are needed.

The chronological course of the parameter change in a hearing device according to the invention depends in particular on empirical values and statements of the user, for example which noise at which level and which exposure time how long a specific tinnitus noise entails. The tracked level of the masker signal preferably remains even after a change in the state of mind or the acoustic situation for a certain time and then falls slowly back to the predefined level in order to conceal the increased tinnitus in his loudness even then effectively ,

The hearing device according to the invention preferably comprises at least one device unit which can be worn on or in the ear. However, it can also be a system with two unit units, which can each be worn on or in one ear of the user. In addition, it may also include a non-portable or in-ear external device unit, such as a remote control. The signal analysis for detecting the emotional state according to the invention can also be carried out completely or partially in the external device unit, in which case a control signal for adapting a respective parameter is transmitted to the device units which can be worn on or in the ear.

The invention will be explained in more detail with reference to embodiments. Showing:

1 a hearing device according to the invention in a highly simplified block diagram and

2 the adjustment of the volume over time as a function of different emotional states.

1 shows in the simplified block diagram a combined hearing and tinnitus masker device (HTM) 1 according to the invention. This includes a microphone 2 for receiving an acoustic input signal and conversion into an electrical input signal. This is after a frequency-dependent processing and amplification as an electrical output signal a listener 4 supplied, the latter back into an acoustic output signal converts. In addition, the HTM includes a signal generator 5 for generating a masker signal for masking a tinnitus noise.

The HTM also includes an auto-tuning detector 7 to recognize the user's own voice. This is in the eigenvoice detector 7 separated from possibly present in the electrical input signal further speech signals other speakers, z. By a BSS algorithm. The user-initiated voice signal is then sent to a signal analysis unit 8th concerning the current emotional state of the user. Furthermore, a sensor detects 6 the heart rate (pulse) of the user, which is also used to determine the user's emotional state. The signal analysis unit 8th then generates at an output a control signal, which the signal processing unit 3 is supplied. The control signal makes at least one adjustable parameter of the signal processing unit 3 For example, the gain or volume setting, adapted to the current state of mind of the user.

2 shows by way of example the time course of the volume setting of a tinnitus masker as a function of different mental states of the user. In a relaxed phase of the user whose tinnitus noise is perceived as relatively quiet. It can therefore already be masked by a relatively quiet masker signal. This is shown by curve section A. At some point in time, the user is detected with an emotional tension / stress. As a result, the Tinnutus sound is noticeably louder. The tinnitus masker reacts by adjusting the volume gradually up to a maximum value. This is shown in section B of the curve. If the user relaxes again from a second point in time and the tinnitus noise returns to the background, a smooth transition of the volume setting takes place from this point until the original setting is restored is. This is illustrated in the curve section C.

The course of the volume setting shown avoids sudden changes that can lead to irritation.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• EP 0064042 A1 [0005]
• DE 102005046168 A1 [0006]
• DE 102005032274 A1 [0007]
• EP 1017253 A2 [0007]
• EP 0820211 A1 [0011]
• EP 2063416 A1 [0014]

Claims (12)

Method for adjusting a hearing device ( 1 ), in particular a hearing aid and / or tinnitus masker, characterized by - detecting and recognizing a speech signal of the user, - analyzing the speech signal for recognizing an emotional state of the user and - adjusting at least one parameter of the hearing device ( 1 ) depending on the detected emotional state. The method of claim 1, wherein additionally biometric data of the user are detected and the parameter also takes place in dependence on the biometric data of the user. The method of claim 1 or 2, wherein additionally the acoustic hearing situation is detected, in which the hearing device ( 1 ), and the parameter is also dependent on the auditory situation. Method according to one of claims 1 to 3, wherein as emotional states at least tension or stress or relaxation are detected. Method according to one of claims 1 to 4, wherein depending on the detected emotional state, a selection of a hearing program and / or a volume adjustment and / or a frequency response and / or a noise reduction of the hearing device ( 1 ) be adjusted. Method according to one of claims 1 to 5, wherein predetermined acoustic signals are emitted depending on the detected acoustic condition. Method according to one of claims 1 to 6, wherein the recognition of a speech signal of the user and / or the recognition of the emotional state of the user are trained and / or learned. The method of any one of claims 1 to 7, wherein after the recognition of an emotional condition, the parameter is slidably transitioned from an initial value to a final value. Hearing device ( 1 ), in particular hearing device and / or tinnitus mask, comprising at least one signal source ( 2 . 5 ) for providing a source signal, a signal processing unit ( 3 ) for processing the source signal in dependence on at least one automatically adjustable parameter and for generating an electrical output signal, an output transducer ( 4 ) for converting the electrical output signal into an acoustic output signal, an input transducer ( 2 ) for receiving an acoustic input signal and conversion into an electrical input signal, a signal analysis unit ( 8th ) for analyzing the electrical input signal and automatically adjusting the parameter as a function of an analysis result, characterized in that a speech signal emanating from a user is detectable and by an analysis of the speech signal in the signal analysis unit ( 8th ) an emotional state of the user is recognizable and the automatic adjustment of the parameter takes place as a function of the emotional state. Hearing device ( 1 ) according to claim 9, wherein at least one sensor ( 6 ) is present for detecting a biometric signal of the user and the automatic adjustment of the parameter also takes place as a function of the biometric signal. Hearing device ( 1 ) according to claim 9 or 10, wherein in the signal analysis unit ( 8th ) An analysis of the electrical input signal for detecting the current hearing situation is carried out and the automatic adjustment of the parameter also takes place in dependence of the current hearing situation. Hearing device ( 1 ) according to one of claims 9 to 11, wherein the signal analysis unit ( 8th ) comprises a neural network.

Images