JP4543014B2 - Hearing device - Google Patents

Description

  The present invention relates to a hearing device such as a hearing aid that is worn on both ears and listens to voice and ambient noise.

  Human hearing processes a wide variety of sound information coming from various directions with a pair of left and right ears, which is generally called the binaural effect. In the actual environment, there are various environmental noises in addition to main sound information mainly used for communication such as voice. Human hearing is based on differences in the level of sound information input to both ears (interaural level difference: ILD) and phase differences (interaural phase difference: IPD). By analyzing and recognizing (sound source direction) in the brain and focusing on the main information in it, we communicate under noise.

  However, as humans become older, their hearing gradually decreases. There are various patterns of hearing loss, and there are many cases where imbalance occurs in the hearing ability of the left and right ears. In such a case, the ILD and IPD that should be extracted by the hearing of both ears are not accurately transmitted to the brain, and there is an error in the analysis of the sound source direction, or the high frequency sound for at least one of the ears A masking phenomenon that occurs when both a component and a low-frequency sound component are input at the same time may occur, resulting in a hindrance to voice communication under noise.

  With regard to such problems, technologies that provide the user with the direction of the sound source appropriately while exchanging data between hearing aids such as hearing aids, etc. And a technique for presenting appropriate sound information to each ear, a technique for performing data exchange between both ears using in-vivo communication (biological communication) technology, and the like.

  In Patent Document 1, (1) means for inputting acoustic signals generated from a plurality of sound sources from both the left and right sound receiving units, (2) means for dividing the left and right input signals for each frequency band, and (3) Means for obtaining IPD for each frequency band from the cross spectrum of both the left and right input signals, and ILD from the level difference of the power spectrum, (4) In each frequency band, each frequency by comparing IPD and / or ILD with that of the database Means for estimating a sound source direction candidate for each band; (5) means for estimating a direction of high appearance frequency among sound source directions obtained for each frequency band as a sound source direction; and (6) estimated from the above. There is a description of a sound source separation system including means for separating sound sources by mainly extracting a frequency band in a specific sound source direction based on the sound source direction information.

  Patent Document 2 includes a sound component detection unit, a sound component separation unit that separates a detected sound into a low-frequency sound component and a high-frequency sound component, and an intensity comparison unit that compares the strengths of the separated sound components. The sound source direction detection unit that detects the sound source direction of the sound detected by the sound component detection unit, the comparison result in the intensity comparison unit, and the sound source direction detected by the sound source direction detection unit, A selector that distributes the output destination of the low-frequency sound component, and the selector selects the sound component determined by the intensity comparison unit as having a higher intensity in the right ear output unit and the left ear output unit. There is a description of a sound processing device, a sound processing method, and a sound processing program that allow a user to accurately perceive the direction of a sound source while suppressing the occurrence of a masking phenomenon by outputting the sound corresponding to the direction of the sound source.

  In Patent Document 3, a band in which masking between frequency components occurs is detected from the frequency characteristics of an input signal, and a frequency band causing masking and a frequency band to be masked are a first sound output means and a second sound output. There is a description of an acoustic processing device that can set the passbands of a low-pass filter and a high-pass filter so as to be divided into means and output, thereby improving speech intelligibility.

U.S. Patent No. 6,099,075 comprises one or more acoustic I / O components operable to receive acoustic input and processing components, at least these components being formed by the user's skin. Electrically coupled together by a closed electrical circuit with one conductive path, electrical signals are transmitted between components using the closed electrical circuit, and communication between components can be one-way communication or two-way communication There is a description of a hearing aid system that is possible.
JP 2004-325284 A JP 2005-268964 A JP 2006-87018 A Special table 2003-533153

  Patent Document 1 describes in detail the direction estimation and separation technique of a sound source, but means and noise for enhancing a user's listening by enhancing or suppressing a sound signal from a specific direction. There is no disclosure or suggestion of means for improving the clarity of speech and means for suppressing the occurrence of masking. Furthermore, a specific configuration for realizing the technology, for example, a data communication means and high-level calculation for realizing a reduction in size and a reduction in power consumption necessary for the user to wear on both ears. There is no disclosure or suggestion regarding the means for configuring the analysis processing unit to be performed.

  Patent Documents 2 and 3 describe a sound processing method for analyzing a sound component of an input sound and presenting different sound components to both ears in order to suppress the occurrence of a masking phenomenon. There is no disclosure or suggestion of data communication means between devices worn on both ears necessary to realize, means to improve the intelligibility of speech by performing high quality analysis under noise, and further advanced calculation There is no disclosure or suggestion regarding the constituent means of the analysis processing unit for performing, for example, the constituent means for realizing the miniaturization and low power consumption necessary for the user to wear and use both ears.

  Patent Document 4 describes a technique for performing data communication between devices worn in both ears using in-vivo communication (biological communication) technology. The sound source direction estimation means to be performed using the data is disclosed. , Means to enhance or suppress the sound signal from a specific direction to improve the user's listening, means to suppress the occurrence of masking phenomenon, means to improve the intelligibility of speech under noise, There is no disclosure or suggestion regarding the configuration means of the analysis processing unit for performing simple calculations, for example, the configuration means for realizing the downsizing and low power consumption necessary for the user to wear on both ears .

  In order to solve the above-described problems, the present invention has the following configuration with respect to a hearing device such as a hearing aid that is worn on both ears and listens to voice and ambient noise.

  In a pair of listening devices to be worn on both ears, each listening device has a microphone for receiving a sound signal, an analysis processing unit for analyzing the sound signal received by the microphone, and an analysis result by the analysis processing unit on the other side. A data transmission unit to be transmitted to the other listening device, a data receiving unit to receive data from the other listening device, an analysis result by the analysis processing unit and an analysis result received by the data reception unit to determine a sound source of a specific sound signal A sound source direction estimating unit that estimates a direction, a sound signal processing unit that emphasizes and / or suppresses a sound signal from a specific direction estimated by the sound source direction estimating unit, and the sound signal processing unit that is emphasized and / or suppressed An acoustic signal radiating unit that emits a sound signal is included. As a result, a pair of hearing devices attached to both ears can enhance and / or suppress sound signals from a specific direction while exchanging data with each other.

  Further, the pair of listening devices to be worn on both ears includes a pair of listening devices and a dedicated signal processing device, and each of the pair of listening devices receives a sound signal and a sound received by the microphone. The signal processing unit includes a data transmission unit that transmits a signal to a signal processing dedicated device, a data reception unit that receives a sound signal from the signal processing dedicated device, and an acoustic signal emitting unit that emits the sound signal received by the data receiving unit. The dedicated device includes a data receiving unit that receives sound data from each of the pair of listening devices, an analysis processing unit that analyzes each of the sound signals received by the data receiving unit, and a specific result from an analysis result by the analysis processing unit. A sound source direction estimating unit for estimating the direction of the sound source of the sound signal, and a sound signal processing for enhancing and / or suppressing a sound signal from a specific direction estimated by the sound source direction estimating unit. And part was a sound signal the sound signal processing section for emphasizing and / or suppression to include data transmission unit to be transmitted to each of the pair of hearing device. As a result, a pair of hearing devices attached to both ears exchange data with each other, and while using a signal processing technique that requires a large amount of computation, a pair of hearing devices can be reduced in size and power consumption can be reduced. It becomes possible to enhance and / or suppress the sound signal from the direction of.

  Further, the listening device for wearing on both ears comprises a pair of listening devices and a signal processing dedicated device, each of the pair of listening devices receiving a sound signal and a sound signal received by the microphone. The analysis processing unit to analyze, the data transmission unit for transmitting the analysis result of the analysis processing unit to the signal processing dedicated device, the data receiving unit for receiving the sound signal from the signal processing dedicated device, and the data receiving unit An acoustic signal radiating unit that emits a sound signal, and the signal processing dedicated device includes a data receiving unit that receives an analysis result from each of the pair of listening devices, and a specific sound signal from the analysis result received by the data receiving unit. A sound source direction estimating unit for estimating the direction of the sound source, a sound signal processing unit for enhancing and / or suppressing a sound signal from a specific direction estimated by the sound source direction estimating unit, and the sound signal A sound signal processing section is stressed and / or suppression and to include data transmission unit to be transmitted to each of the pair of hearing device. As a result, a pair of hearing devices worn on both ears exchange data with each other, and a pair of hearing devices and a dedicated signal processing device efficiently share the computation, and while using a signal processing technique that requires a large amount of computation. In addition, the pair of hearing devices can be reduced in size and power consumption, and sound signals from a specific direction can be enhanced and / or suppressed.

  Further, in a pair of listening devices to be worn on both ears, each listening device has a microphone that receives a sound signal, an analysis processing unit that analyzes the sound signal received by the microphone, and an analysis result by the analysis processing unit Is transmitted to the other hearing device, a data receiving unit that receives data from the other hearing device, an analysis result by the analysis processing unit, and a masking in the sound signal from the analysis result received by the data receiving unit. A masking amount calculating unit for calculating the amount, a signal separating unit for separating the masker frequency component and the masky frequency component from the masking amount calculated by the masking amount calculating unit, and the separated masker frequency component and / or the masky frequency It includes a sound signal synthesizer that synthesizes the component into a sound signal and an acoustic signal radiating unit that emits the synthesized sound signal. It was. As a result, a pair of hearing devices attached to both ears exchange data with each other, and the occurrence of the masking phenomenon can be suppressed.

  Further, the listening device for wearing on both ears comprises a pair of listening devices and a signal processing dedicated device, each of the pair of listening devices receiving a sound signal and a sound signal received by the microphone. A signal processing unit including a data transmission unit for transmitting to a signal processing dedicated device, a data receiving unit for receiving a sound signal from the signal processing dedicated device, and an acoustic signal emitting unit for emitting the sound signal received by the data communication unit. A data receiving unit for receiving sound data from each of the pair of listening devices, an analysis processing unit for analyzing the sound signals received by the data receiving unit, and masking in the sound signal from the analysis result by the analysis processing unit The masking amount calculation unit for calculating the amount, the masker frequency component calculated from the masking amount calculation unit, and the masker frequency component and the masky frequency A signal separating unit for separating the minutes, a sound signal synthesizing unit for synthesizing the separated masker frequency components and / or masky frequency components into sound signals, and a sound signal and masky frequency components synthesized from the masker frequency components And a data transmission unit for transmitting the sound signal synthesized from each of the pair of listening devices. As a result, a pair of hearing devices attached to both ears exchange data with each other, and while using a signal processing method that requires a large amount of computation, a pair of hearing devices can be reduced in size and power consumption can be reduced. Can be suppressed.

  Further, the listening device for wearing on both ears comprises a pair of listening devices and a signal processing dedicated device, each of the pair of listening devices receiving a sound signal and a sound signal received by the microphone. An analysis processing unit to analyze, a data transmission unit that transmits an analysis result of the analysis processing unit to a signal processing dedicated device, a data receiving unit that receives analysis data from the signal processing dedicated device, and an analysis received by the data receiving unit A sound signal synthesizer that synthesizes data into a sound signal and an acoustic signal radiating unit that emits the synthesized sound signal. The signal processing dedicated device receives data analysis results from each of the pair of listening devices. A masking amount calculating unit that calculates a masking amount in the sound signal from an analysis result received by the data receiving unit, and a mask calculated by the masking amount calculating unit. A signal separating unit that separates the frequency component of the masker and the frequency component of the maskee from the king amount, and a data transmission unit that transmits the separated frequency component of the masker and / or the frequency component of the masky to each of the pair of listening devices I did it. As a result, a pair of hearing devices worn on both ears exchange data with each other, and a pair of hearing devices and a dedicated signal processing device efficiently share the computation, and while using a signal processing technique that requires a large amount of computation. In addition, the pair of hearing devices can be reduced in size and power consumption, and the masking phenomenon can be suppressed.

  The analysis by the analysis processing unit is configured by Wavelet conversion. As a result, sound signal enhancement and / or suppression in a specific direction following a time-frequency change and suppression of the occurrence of a masking phenomenon, which cannot be achieved by normal frequency analysis, can be performed.

  The analysis by the analysis processing unit is configured by discrete cosine transform. As a result, frequency analysis is efficiently performed on the input sound signal, and sound signal enhancement and / or suppression from a specific direction and generation of a masking phenomenon can be efficiently performed.

  Further, when the analysis processing unit performs the analysis, the voice and noise included in the sound signal input to the analysis processing unit are separated, and the separated voice and / or noise is analyzed. This makes it possible to individually emphasize and / or suppress voice, which is a main component of communication, and noise, which is an interference sound, from the input sound signal, and efficiently suppress the occurrence of masking phenomenon. .

  Further, when analysis is performed by the analysis processing unit, the voice and noise included in the sound signal input to the analysis processing unit are separated based on Speech Absence Probability. As a result, voice from a specific direction, which is a major component of communication, and noise, which is a disturbing sound, can be emphasized and / or suppressed individually from the input sound signal, and the occurrence of masking can be suppressed. You can do it efficiently.

  Furthermore, the data transmission unit and the data reception unit are configured by in-vivo communication (biological communication). As a result, it is possible to implement a hearing device to be worn on both ears with a small-scale system without considering data communication wirelessly and radio frequency used for data transmission.

  According to the listening device of the present invention, a pair of listening devices attached to both ears can enhance and / or suppress a sound signal from a specific direction while exchanging data with each other. In addition, the sound information necessary for the user can be emphasized according to the sound environment, unnecessary sound information can be suppressed, and a comfortable sound can always be presented.

  The listening device of the present invention calculates the masking amount of sound information while a pair of listening devices attached to both ears exchange data with each other, and can separate the masker frequency component and the masky frequency component, so that the masking phenomenon Can be accurately suppressed, and a comfortable sound can always be presented according to the sound environment.

  Such sound processing often requires signal processing that requires a large amount of computation, but in order to use it comfortably by wearing it in both ears, it needs to be ultra-small and have low power consumption. . The listening device of the present invention can also include a dedicated processing device in addition to a pair of listening devices, and the listening device to be worn on both ears can be realized with ultra-small size and low power consumption.

  In addition, the listening device of the present invention can include Wavelet transform and discrete cosine transform instead of the commonly used FFT as a method for analyzing sound information. As a result, frequency analysis can be performed efficiently, and after considering changes in frequency information over time (time-frequency information), sound information necessary for the user is emphasized according to the sound environment. Necessary sound information can be suppressed, and the occurrence of masking can be accurately suppressed.

  In addition, when the analysis processing unit performs an analysis, the listening device of the present invention can separate voice and noise included in a sound signal input to the analysis processing unit, and perform specific signal processing on each of them. It is said. In general, the environment in which a hearing device is used is mixed with various types of noise along with the voice. However, according to this configuration, only the main voice is emphasized and noise and disturbing voice are suppressed. Can always provide a comfortable sound to the user.

  In the present invention, as a method for separating speech and noise, a method for separating based on Speech Absence Probability can be provided. According to this method, it is possible to efficiently and accurately separate voice and noise, emphasize only the main voice, suppress noise, and always provide a comfortable sound to the user.

  Furthermore, the present invention enables a configuration that uses in-vivo communication (biological communication) technology for data communication between both ears or with a processing-dedicated device. When performing data communication between both ears, a method using FM radio waves or the like is common, but this requires assignment of frequencies based on the Radio Law. Even if frequency allocation is obtained, it is difficult to realize an ultra-small and low-power consumption device. However, according to the present invention, both ears can be used without considering the radio frequency used for data transmission. Can be realized with a small-scale system.

  The best mode for carrying out the present invention will be described below in detail with reference to the drawings. In the following description, components having the same function are denoted by the same reference numerals, and repeated description thereof is omitted.

  FIG. 1 is a block diagram of a system according to an embodiment of the present invention, which is composed of a pair of a listening device 1 and a listening device 2 to be worn on both ears.

  The listening device 1 and the listening device 2 have the same configuration, the microphone 3 that picks up surrounding sounds, the analysis processing unit 4 that analyzes the sound information collected by the microphone 3, and the analysis processing unit 4 that analyzes the sound information. A data transmitting unit 5 for transmitting the analyzed result to the other listening device, a data receiving unit 6 for receiving the data transmitted by the data transmitting unit 5, and an analysis result and data receiving unit 6 of the analysis processing unit 4 From the received analysis result of the other listening device, using the sound source direction estimating unit 7 that estimates the direction of the sound source of the sound information existing around it and the various sound source information estimated by the sound source direction estimating unit 7, A sound signal processing unit 8 that emphasizes a necessary signal (speech in this case) and suppresses an unnecessary signal (noise in this case) and a speaker (earphone) 9 that is an acoustic signal radiating unit are included.

  Here, it is desirable that the data transmission unit 5 and the data reception unit 6 are configured by wireless devices in consideration of user convenience. However, the general wireless technology consumes a large amount of power, and the shape of the device is too large to be worn in the ear.

  Therefore, in the present embodiment, in-vivo communication (biological communication) technology is used for the data transmission unit 5 and the data reception unit 6. The in-vivo communication (biological communication) technique is a technique for performing communication using a human body as a transmission path, and is a technique capable of performing large-capacity data communication with low power consumption.

  For estimating the direction of the sound source, for example, “(1) means for inputting acoustic signals generated from a plurality of sound sources from the left and right sound receiving units, and (2) both left and right input, as described in JP-A-2004-325284. Means for dividing the signal for each frequency band, (3) means for obtaining an IPD for each frequency band from the cross spectrum of the left and right input signals, and an ILD from the level difference of the power spectrum, and (4) IPD and / or for all frequency bands Alternatively, a means for estimating a sound source direction candidate for each frequency band by comparing the ILD with that of the database, (5) a direction having a high appearance frequency among the sound source directions obtained for each frequency band, Means for estimating the sound source direction, (6) means for separating the sound source by mainly extracting the frequency band of the specific sound source direction based on the sound source direction information estimated from the above, The use of such a sound source separation system "consisting of, can be easily realized.

  Further, in this embodiment, wavelet conversion is used for analysis in the analysis processing unit 4. In general, FFT is often used for such frequency analysis. In FFT, the frequency resolution is inversely proportional to the size of the window function. Furthermore, in the case of FFT, it can only be analyzed with a uniform width on the frequency axis. If the size of this window function is set large, the frequency resolution is improved, but the information about the time axis, that is, the time resolution is lowered. On the other hand, if the size of the window function is set small, the time resolution is improved, but the frequency resolution is decreased. That is, the FFT cannot achieve both frequency resolution and time resolution.

  Sound information that exists in the real environment changes from moment to moment. Human auditory senses the surrounding sound environment and recognizes necessary information such as voice while adapting to this change over time based on information input from both ears. It is also known that the frequency analysis mechanism in the human auditory peripheral system can be simulated by a filter bank called an auditory filter. The bandwidth of the auditory filter changes in accordance with the frequency of the input sound signal (the bandwidth becomes wider as the frequency goes higher). This means that the frequency selectivity (frequency resolution) changes depending on the frequency of the input sound signal.

  Therefore, a linear analysis method such as FFT has redundant data in the high frequency region and lacks resolution in the low frequency region. In the FFT, the analysis results are arranged linearly on the frequency axis.

  Although FFT is an excellent frequency analysis technique, it is difficult to achieve both accurate tracking of input sound signals over time and accuracy of frequency analysis (frequency resolution). Furthermore, since the frequency range that can be analyzed is equal in the FFT, it is difficult to perform an analysis that takes into account the auditory filter according to the frequency of the input sound signal. ) And the surrounding environment, the sound source direction may not be sufficiently estimated.

  Wavelet is “ripple wave” and is a technique for analyzing the frequency-time axis of a signal by enlarging or reducing the basic function of one mother wavelet. In Wavelet transform, the analysis frequency width can be set to be a logarithmic width on the frequency axis, so avoid the degradation of resolution in the low frequency region and overanalysis in the high frequency region, which was a problem in FFT. be able to. Although this method requires a large calculation, it solves the problems of the FFT described above, achieves both accurate tracking of the sound environment over time and accuracy of frequency analysis, and further according to the frequency of the input sound signal. Since the analysis can be performed according to the auditory filter characteristics, a high-quality listening device can be realized.

  For example, FIG. 7 shows a configuration example of the sound source separation system described in Japanese Patent Application Laid-Open No. 2004-325284. By converting the FFT of this configuration example into Wavelet transform and IFFT into inverse Wavelet transform, the frequency-time axis of the signal Analysis is possible, and a high-quality listening device that can follow changes with time of surrounding sounds has been realized.

  The sound signal processing unit 8 discriminates the sound source estimated by the sound source direction estimation unit 7 from the sound source corresponding to the sound and the noise corresponding to the noise, emphasizes the sound, and suppresses the noise. Specifically, it is realized by amplifying the intensity of the component from the direction corresponding to the voice and attenuating the intensity of the component from the direction corresponding to the noise on the frequency axis.

  In this example, wavelet conversion is performed and the sound source direction is estimated and then voice and noise are discriminated. However, in an environment where many people are talking at once, such as a cocktail party, There is a possibility that speech component (speech to be heard), disturbing speech component, and noise are mixed, and the direction estimation accuracy of the main speech component is lowered.

  In this case, as shown in FIG. 8, the speech and noise are separated before the wavelet transform, and the sound other than the speech is deleted before the sound source direction estimation, so that only the plurality of speech components are present. Thus, a more effective listening device can be configured by estimating the sound source direction, emphasizing the main speech component, and suppressing the disturbing speech component. The separated noise components are synthesized by the sound signal processing unit. If a suppression process is added at this time, the noise component also serves as a noise suppression system.

  Furthermore, it is effective to use Speech Absence Probability to separate this voice and noise. Speech Absence Probability compares the frequency spectrum of noise with the frequency spectrum where noise and voice are mixed, and the probability that the voice is included in the frequency spectrum to be analyzed at that time (exactly, included) This is a method of estimating the frequency and the output at the time based on the probability. Therefore, as shown in FIG. 8, it is analyzed whether or not the signal input to the microphone of each listening device is voice at the first stage, only the voice is extracted, and only the sound in the desired sound source direction is presented to the listener By doing so, only necessary audio information can be presented to the listener more efficiently.

    FIG. 2 is a block diagram of a system according to the second embodiment of the present invention, which is composed of a pair of listening devices 10, a listening device 11 and a processing dedicated device 12 for wearing on both ears.

  The listening device 10 and the listening device 11 have the same configuration, the microphone 3 that picks up surrounding sounds, the data transmission unit 5 that sends the sound information collected by the microphone 3 to the processing dedicated device 12, and the processing dedicated The data receiving unit 6 that receives sound data processed by the device 12 and a speaker (earphone) 9, the processing dedicated device 12, the data receiving unit 6 that receives the sound information collected by the microphone 3, An analysis processing unit 4 that analyzes sound information from each of the two listening devices, and a sound source direction estimation unit that estimates the direction of the sound source of the sound information existing in the vicinity from the analysis result analyzed by the analysis processing unit 4 7 and a sound signal processing unit that uses the various sound source information estimated by the sound source direction estimating unit 7 to emphasize a necessary signal (speech in this case) and suppress unnecessary signals (noise in this case). 8 , And a data transmission unit 5 for transmitting the sound information processed by the previous period sound signal processor 8 to the hearing device 10 and the hearing device 11.

  In the present embodiment, all processing that requires complicated and large-scale computation such as Wavelet transformation is performed by the processing dedicated device 12. The listening device 10 and the listening device 11 are attached to the user's ear, like the listening device 1 and the listening device 2, but the processing dedicated device 12 is in a place where biological communication is possible (such as in a pocket). As long as it is provided, advanced signal processing as in the present invention can be realized without impairing user convenience.

  FIG. 3 is a block diagram of a system according to the third embodiment of the present invention, which is composed of a pair of listening devices 13, a listening device 14, and a processing dedicated device 15 for wearing on both ears.

  The listening device 13 and the listening device 14 have the same configuration, the microphone 3 that picks up surrounding sounds, the analysis processing unit 4 that analyzes sound information picked up by the microphone 3, and the analysis processing unit 4 that analyzes the sound information. The processing dedicated device 15 is composed of a data transmitting unit 5 that transmits the analysis result to the processing dedicated device 15, a data receiving unit 6 that receives sound data processed by the processing dedicated device 15, and a speaker (earphone) 9. The data receiving unit 6 that receives the analysis result analyzed by the analysis processing unit 4, and the sound source direction estimating unit that estimates the direction of the sound source of the sound information existing around the data receiving unit 6 from the analysis results from the two listening devices 7 and a sound signal processing unit that uses the various sound source information estimated by the sound source direction estimating unit 7 to emphasize a necessary signal (speech in this case) and suppress unnecessary signals (noise in this case). 8 and the previous term And a data transmitting unit 5 for transmitting the sound information processed by the signal processing unit 8 to the hearing device 13 and the hearing device 14.

  In the present embodiment, analysis processing such as Wavelet conversion is performed individually by the listening device 13 and the listening device 14, and sound source direction estimation and sound signal processing are performed by the processing dedicated device 15. As a result, high-level arithmetic processing is distributed to the listening device and the processing-dedicated device, so that the calculation load on each device is reduced and the data transmission amount can be greatly reduced. In this embodiment, the data transmitted from the processing dedicated device 15 to the listening device 13 and the listening device 14 is sound data. However, the processing performed by the processing dedicated device 15 is only the sound source direction estimation, and the listening device 13 and Obviously, if the sound signal processing is performed by the listening device 14, the amount of data transmission can be further reduced.

  FIG. 4 is a block diagram of a system according to the fourth embodiment of the present invention, and includes a pair of listening devices 16 and a listening device 17 to be worn on both ears.

The listening device 16 and the listening device 17 have the same configuration, and the microphone 3 that picks up surrounding sounds, the analysis processing unit 4 that performs wavelet conversion of sound information collected by the microphone 3, and the analysis processing unit 4 A data transmission unit 5 that transmits the analyzed analysis result to the other listening device, a data reception unit 6 that receives the data transmitted by the data transmission unit 5, and an analysis result and data reception unit 6 of the analysis processing unit 4 Is input from the result obtained by the masking amount calculating unit 18 and the masking amount calculating unit 18 for calculating the masking amount for each frequency of the sound information existing in the periphery from the analysis result in the other listening device received at Separating the sound information into a masky frequency component and a masker frequency component, and a masky frequency component or masker frequency separated by the signal separator 19 A sound signal synthesis section 20 to the sound signal component synthesis to consist speaker (earphone) 9.

  Here, it is desirable that the data transmission unit 5 and the data reception unit 6 are configured by wireless devices in consideration of user convenience. However, the general wireless technology consumes a large amount of power, and the shape of the device is too large to be worn in the ear.

  Therefore, in the present embodiment, in-vivo communication (biological communication) technology is used for the data transmission unit 5 and the data reception unit 6. The in-vivo communication (biological communication) technique is a technique for performing communication using a human body as a transmission path, and is a technique capable of performing large-capacity data communication with low power consumption.

  For the calculation of the masking amount and the signal separation, for example, “Sound component detection unit described in JP-A-2005-268964 and sound component separation for separating the detected sound into a low-frequency sound component and a high-frequency sound component” An intensity comparison unit that compares the intensity of the separated sound component, a sound source direction detection unit that detects a sound source direction of the sound detected by the sound component detection unit, a comparison result in the intensity comparison unit, and a sound source direction detection And a selector that sorts the output destination of the high-frequency sound component and the low-frequency sound component based on the sound source direction detected by the unit, and the selector determines the sound that has been determined to have higher intensity by the intensity comparison unit. By outputting the component to the output unit for the right ear and the output unit for the left ear that corresponds to the sound source direction, the sound that enables the user to accurately perceive the sound source direction while suppressing the occurrence of masking phenomenon Processing apparatus, sound processing method And a sound processing program ”or“ a band where masking between frequency components occurs is detected from the frequency characteristics of the input signal, and the frequency band causing the masking and the frequency band to be masked are first. The sound processing apparatus can set the pass band of the low-pass filter and the high-pass filter so as to be divided and output to the sound output means and the second sound output means, thereby improving the clarity of the sound. Etc. can be easily realized.

  For example, FIG. 9 shows a configuration example of a sound processing device, a sound processing method, and a sound processing program described in Japanese Patent Application Laid-Open No. 2005-268964. This can be realized by using transformation. That is, from the analysis result on the frequency-time axis calculated by the Wavelet transform, the high frequency component and the low frequency component of the speech are separated, the intensity is compared with time, and FIG. 7 described in the first embodiment is used. The sound source direction can be estimated with high accuracy by the described method.

Thereby, a masker frequency component and a maskee frequency component are generated from the input sound, and output destinations of the masker frequency component and the maskee frequency component are distributed to the left and right listening devices 16 and 17 in consideration of the frequency-time axis. It becomes possible. Since the sound signal synthesizer 20 in each listening device synthesizes the masky frequency component or the masker frequency component , the user can recognize the input sound without missing the masky frequency component by the masking phenomenon. The occurrence of masking can be prevented by a sound processing device or the like.

  In this example, wavelet conversion is performed to calculate the amount of masking for each frequency and signal separation, but in an environment where many people are talking at a time under high noise or cocktail parties. In this case, main speech components (speech to be heard), disturbing speech components, and noise are mixed, and the accuracy of direction estimation of main speech components, calculation of masking amount, and signal separation may be reduced.

In this case, as in the configuration example shown in FIG. 8, the speech and noise are separated before the wavelet transform, and the sounds other than the speech are suppressed before the calculation of the masking amount for each frequency and the signal separation. and, against the signal was only the sound component by performing calculation and signal separation of the masking quantity for each frequency concerning major sound components, it can be constructed more efficient hearing device. The separated noise components are synthesized by the sound signal processing unit. If a suppression process is added at this time, the noise component also serves as a noise suppression system.

  Furthermore, it is effective to use Speech Absence Probability to separate this voice and noise.

  FIG. 5 is a block diagram of a system according to the fifth embodiment of the present invention, which is composed of a pair of listening devices 21, a listening device 22 and a processing dedicated device 23 for wearing on both ears.

The listening device 21 and the listening device 22 have the same configuration, the microphone 3 that picks up surrounding sounds, the data transmission unit 5 that sends the sound information collected by the microphone 3 to the processing dedicated device 23, and the processing dedicated. The data receiving unit 6 that receives sound data processed by the device 23 and a speaker (earphone) 9, and the processing dedicated device 23 includes a data receiving unit 6 that receives sound information collected by the microphone 3, and An analysis processing unit 4 that performs wavelet conversion on sound information from the two listening devices, and a masking that calculates a masking amount for each frequency of sound information existing in the periphery from the analysis result analyzed by the analysis processing unit 4 A signal for separating input sound information into a masky frequency component and a masker frequency component from the result obtained by the amount calculation unit 18 and the masking amount calculation unit A separation unit 19, a sound signal synthesis unit 20 that synthesizes a masky frequency component and a masker frequency component separated by the signal separation unit 19, respectively, and a masky frequency component synthesized by the previous sound signal synthesis unit 20. And the data transmission unit 5 for transmitting the masker frequency component to the listening device 21 and the listening device 22, respectively.

  In the present embodiment, all processing that requires complicated and large-scale operations such as Wavelet transformation is performed by the processing dedicated device 23. The listening device 21 and the listening device 22 are attached to the user's ear in the same manner as the listening device 16 and the listening device 17, but the processing-dedicated device is provided in a place where biological communication is possible (such as in a pocket). As a result, it is possible to realize advanced signal processing as in the present invention without impairing the convenience for the user.

  FIG. 6 is a block diagram of a system according to the sixth embodiment of the present invention, and is composed of a pair of listening devices 24, a listening device 25 and a processing dedicated device 26 for wearing on both ears.

The listening device 24 and the listening device 25 have the same configuration, and the microphone 3 that picks up surrounding sounds, the analysis processing unit 4 that performs wavelet conversion of sound information collected by the microphone 3, and the analysis processing unit 4 The data transmitting unit 5 that transmits the analyzed analysis result to the processing dedicated device 26, the data receiving unit 6 that receives the analysis result analyzed by the processing dedicated device 26, and the analysis result received by the data receiving unit are combined. The sound signal synthesizer 20 and the speaker (earphone) 9 are made into a sound signal, and the processing dedicated device 26 includes a data receiving unit 6 that receives the analysis result analyzed by the analysis processing unit 4, and the two From the analysis result from the listening device, from the result obtained by the masking amount calculation unit 18 for calculating the masking amount for each frequency of the sound information existing in the surroundings, and the masking amount calculation unit, The signal separation unit 19 that separates the input sound information into the masky frequency component and the masker frequency component, and the masky frequency component and the masker frequency component separated by the previous signal separation unit 19 are transmitted to the listening device 24 and the listening device 25, respectively. The data transmission unit 5 includes:

  In the present embodiment, analysis processing such as wavelet conversion and synthesis of sound signals are performed individually by the listening device 24 and the listening device 25, and the calculation of the masking amount for each frequency and the separation of the masky frequency component and the masker frequency component are performed. This is performed by the processing dedicated device 26. As a result, high-level arithmetic processing is distributed to the listening device and the processing-dedicated device, so that the calculation load on each device is reduced and the data transmission amount can be greatly reduced.

  Note that with respect to the embodiment of the present application, the Wavelet transform is used for all the analysis processing, but it is naturally possible to perform this all with the discrete cosine transform.

The block diagram of the system in the 1st Embodiment of this invention The block diagram of the system in the 2nd Embodiment of this invention The block diagram of the system in the 3rd Embodiment of this invention The block diagram of the system in the 4th Embodiment of this invention Block diagram of a system in the fifth embodiment of the present invention The system block diagram in the 6th Embodiment of this invention Configuration example of sound source separation system Configuration example of voice and noise separation system Example of a system that prevents the occurrence of masking and makes the user perceive the direction of the sound source

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Listening apparatus, 2 ... Listening apparatus, 3 ... Microphone, 4 ... Analysis processing part, 5 ... Data transmission part, 6 ... Data receiving part, 7 ... Sound source direction estimation part, 8 ... Sound signal processing part, 9 ... Speaker, DESCRIPTION OF SYMBOLS 10 ... Hearing device, 11 ... Hearing device, 12 ... Processing-only device, 13 ... Hearing device, 14 ... Hearing device, 15 ... Processing-only device, 13 ... Hearing device, 14 ... Hearing device, 15 ... Masking amount calculation part, 16 DESCRIPTION OF SYMBOLS ... Signal separation part, 17 ... Sound signal synthesis | combination part, 21 ... Hearing apparatus, 22 ... Hearing apparatus, 23 ... Dedicated processing apparatus, 24 ... Hearing apparatus, 25 ... Hearing apparatus, 26 ... Dedicated processing apparatus, 27 ... Voice / noise separation Department.

Claims (3)

A pair of listening devices to be worn on both ears, each of which listens to a microphone that receives a sound signal and voice and noise included in the sound signal received by this microphone in advance based on Speech Absence Probability An analysis processing unit that performs Wavelet transform on the separated speech and / or noise, a data transmission unit that transmits an analysis result by the analysis processing unit to the other listening device, and an analysis result from the other listening device. A data receiving unit for receiving, a masking amount calculating unit for calculating a masking amount in the sound signal from an analysis result by the analysis processing unit and an analysis result received by the data receiving unit, and a masking amount calculated by the masking amount calculating unit A signal separation unit that separates the frequency component of the masker and the frequency component of the masker, and the frequency component of the separated masker and And / or sound signal synthesis section for synthesizing the frequency components of the maskee sound signal, hearing device characterized by comprising a sound signal emission unit for emitting the synthesized sound signal. A listening device for wearing on both ears, comprising a pair of listening devices and a dedicated signal processing device, each of the pair of listening devices receiving a sound signal and a sound signal received by the microphone A signal processing unit comprising a data transmission unit for transmitting to a signal processing dedicated device, a data receiving unit for receiving a sound signal from the signal processing dedicated device, and an acoustic signal emitting unit for emitting a sound signal received by the data receiving unit. Is a data receiving unit that receives sound data from each of the pair of listening devices, and voice and noise included in the sound signal received by the data receiving unit in advance based on Speech Absence Probability, and and / or the analysis processing unit that performs each Wavelet transform on the noise, to calculate the masking amount in the sound signal from the analysis result by the analysis processing unit The masking amount calculation unit, the signal separation unit that separates the masker frequency component and the masky frequency component from the masking amount calculated by the masking amount calculation unit, and the separated masker frequency component and / or the masky frequency component, respectively. A sound signal synthesizing unit that synthesizes the signal, and a data transmission unit that transmits a sound signal synthesized from the frequency component of the masker and a sound signal synthesized from the frequency component of the maskee to each of the pair of listening devices. A characteristic listening device. A hearing device to be worn on both ears, comprising a pair of listening devices and a dedicated signal processing device, each of the pair of listening devices receiving a sound signal and a sound signal received by the microphone Based on Speech Absence Probability, the included speech and noise are separated in advance, an analysis processing unit that performs wavelet transform on the separated speech and / or noise, and the analysis result of the analysis processing unit is transmitted to a signal processing dedicated device. A data transmission unit, a data reception unit that receives analysis data from the signal processing dedicated device, a sound signal synthesis unit that synthesizes the analysis data received by the data reception unit into a sound signal, and an acoustic that emits the synthesized sound signal The signal radiating unit, the signal processing dedicated device includes a data receiving unit that receives an analysis result from each of the pair of listening devices, and a data receiving unit A masking amount calculation unit that calculates a masking amount in the sound signal from the received analysis result, and a signal separation unit that separates a masker frequency component and a masky frequency component from the masking amount calculated by the masking amount calculation unit are separated. A hearing device comprising: a data transmission unit that transmits a masker frequency component and / or a masky frequency component to each of the pair of listening devices.