JP2870421B2 - Hearing aid with speech speed conversion function - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a hearing aid for assisting a hearing function deteriorated due to aging or the like, and more particularly to a speech speed conversion processing for extending a captured voice signal on a time axis to slowly provide a wearer with a speech signal. The present invention relates to a hearing aid having a speech speed conversion function for clearly listening.

[0002]

2. Description of the Related Art Conventionally, hearing aids have been used to assist the hearing function of a person with reduced hearing. Incidentally, in such cases it is hearing loss is due to aging, (maximum speech speed that can identify speech) voice identification critical speed not decrease only the minimum audible signal level Noboru Ue and treble listening function decrease For such a person, it is desirable to process not only the sound amplification and frequency characteristics but also the time characteristics at the same time. For this reason, hearing aids have recently been proposed that can perform a speech speed conversion process of extending a fetched audio signal on a time axis so that the wearer of the hearing aid can slowly and clearly hear the hearing aid.

As a technique for extending an audio signal on a time axis, there is known a "high quality speech rate conversion type listening system" NHK Broadcasting Research Institute (Hearing Research Group H-92-5, Acoustical Society of Japan).
4: November 20, 1992).

[0004]

By the way [0007], when a person is to the conversation, such as measuring the <br/> utterances timing of the next voice while listening to the voice of self-emitted, to listen to the voice generated by the self Feedback processing is performed unconsciously.

[0005] However, in the above-mentioned conventional hearing aid having a speech rate conversion function, the voice uttered by the wearer himself is also subjected to speech rate conversion processing and then heard by the wearer. If you are wearing a hearing aid with a function,
Listening to your own voice delayed by the speech speed conversion process,
You will give feedback to your voice . However, when listening to the voice of their own that has been delayed has issued make a feedback to the next outgoing voice, there is a problem that normally will not be able to perform calling voice.

SUMMARY OF THE INVENTION It is an object of the present invention to provide a hearing aid which discriminates whether or not a fetched voice is a voice uttered by a wearer and outputs the voice without performing a speech speed conversion process when the voice is uttered by the wearer. It is in.

[0007]

[Means for Solving the Problems] of claim 1 invention is a hearing aid with a speech speed converting means for expanding the audio signal captured on the time axis, is mounted on the portion inserted into the ear hole
Solids that come into direct or indirect contact with the inner wall of the wearer's ear canal
A sound identification unit having a propagation sound detection unit, and identifying whether the captured audio signal is a sound emitted by the wearer,
A speech speed conversion prohibiting device for prohibiting the speech speed converting device when the voice discriminating device recognizes the voice of the wearer. The invention of claim 2 is
Speech rate conversion means for extending the captured audio signal on the time axis
In hearing aid with the outer ear of the wearer is attached to the ear
It has a speaker identification microphone that detects voice signals in the
Comparing the embedded audio signal with the audio signal in the ear canal
The audio signal captured by the
Voice identification means for identifying whether or not
When the step identifies the wearer's voice,
Means for inhibiting speech rate conversion means for inhibiting speech rate conversion means.
It is characterized by the following.

[0008]

In the hearing aid having a speech speed conversion function according to the present invention, whether or not the captured audio signal is a voice uttered by the wearer is identified by the audio identification means. The invention of claim 1
This identification is based on the audio signal propagating through the inner wall of the ear canal.
Detect by detecting the utterance of the wearer
U. According to the second aspect of the present invention, the identification is performed by using
Detects the audio signal inside and compares it with the captured audio signal.
By detecting the utterance of the wearer by comparing
U. In this identification, if the captured voice is identified as the voice uttered by the wearer, the captured voice signal is output without performing the speech speed conversion process.

Accordingly, the hearing aid outputs the voice signal without performing the speech speed conversion processing when the captured voice signal is a voice uttered by the wearer, and outputs the voice signal when the captured voice signal is not the voice uttered by the wearer. The audio signal is output after performing speed conversion processing. Therefore, the hearing aid wearer listens to his or her own voice that has not been subjected to speech rate conversion and gives feedback to the next generation, so that the next voice can be made normally. In addition, since the user hears the voice uttered by the other party whose speech speed has been converted, the user can hear the voice slowly and clearly.

[0010]

FIG. 1 is a block diagram showing the configuration of a hearing aid having a speech speed conversion function according to an embodiment of the present invention. This hearing aid determines whether or not the audio signal input from the microphone 1 is a signal of a voice uttered by the wearer, and in the case of a voice uttered by the wearer, amplifies and outputs only the amplified signal. It is easy to hear the wearer by also be output from the audio signal in temporal extension in the case of voice a person has issued other than.
When a speech signal is temporally expanded, it is not possible to obtain a natural voice simply by extending the signal uniformly. This is because, when a person speaks slowly, the sounding time of the vowel slightly increases as the sounding interval, that is, the silence period becomes longer, but the sounding time of the consonant hardly changes. Therefore,
In this hearing aid, a silent section and a vowel section (voiced section) are expanded, and a consonant section (unvoiced section) is output without being expanded, thereby realizing natural speech expansion. It should be noted that the expansion ratio can be set individually for the silent section and the vowel section.

In FIG. 1, the hearing aid comprises a microphone 1, an anti-aliasing filter 2, an A / D conversion circuit 3
, An input unit comprising a DSP circuit 12, a memory circuit 13, a D / A conversion circuit 7, a low-pass filter 8, an equalizer 9, an amplifier 10, an output unit comprising a receiver 11, a clock generator 4, and a sampling clock generator. It comprises a clock circuit comprising a circuit 5 and a speaker identification section 14 for identifying whether or not the audio signal received by the microphone 1 is a voice uttered by the wearer.

The microphone 1 takes in air vibration as an audio signal, converts it into an analog electric signal, and outputs it. The anti-aliasing filter 2 is a so-called low-pass filter. In order to prevent distortion from occurring due to A / D conversion of the A / D conversion circuit 3, the microphone 1
Which is の of the sampling frequency (f = 16 kHz) of the A / D conversion circuit 3 from the analog signal input from
Cuts high frequencies above kHz. A / D conversion circuit 3
Converts the analog signal whose high-frequency range has been cut by the anti-aliasing filter 2 into a digital signal. A / D
The sampling frequency of the conversion circuit 3 is 16 as described above.
kHz, and the sampling clock of this frequency is provided from the sampling clock generation circuit 5. The sampling clock generation circuit 5 includes a clock generator 4
The high-speed clock signal (several tens of megahertz) generated by the above is divided to generate the sampling clock. A high-speed clock signal generated by the clock generator is supplied as an operation clock of the DSP circuit 12.

The signal processing section 6 is a circuit for performing a speech speed conversion process such as a decompression process on the digital signal input from the A / D conversion circuit 3, and cuts out or cuts out the audio signal by a microprogram. Processing such as expansion of a voiced sound portion of the audio signal is performed. D / A conversion circuit 7
Converts the digital signal processed by the signal processing unit 6 into an analog signal. The low-pass filter 8 has a cut-off frequency of 8 kHz and removes aliasing noise from the analog signal output from the D / A conversion circuit 7. This analog signal is input to the equalizer 9, and the frequency characteristics are converted according to the hearing of the wearer, such as by boosting the signal level in the middle and high frequency range. The signal output from the equalizer 9 is amplified by the amplifier 10 into a signal having sufficient power, and then input to the receiver 11. This processing may be performed by the DSP circuit 12 without providing the equalizer 9. The receiver 11 is mounted on the ear of the wearer, converts the input analog signal into air vibration, and emits the air vibration to the ear canal of the wearer.

When an audio signal is input from the microphone 1, the signal is converted to a digital signal by an A / D conversion circuit 3 via an anti-aliasing filter 2 and input to a DSP circuit 11. At the same time, the speaker identification unit 14 identifies whether the voice signal is a voice uttered by the wearer. When the speaker identification unit 14 identifies that the voice is not a voice uttered by the wearer, that is, a voice uttered by a person other than the wearer, the DSP circuit 12 temporally expands the input digital signal. This decompression process will be described in detail in the description of the flowchart of FIG. The expanded digital signal is converted to an analog signal by a D / A conversion circuit 7, subjected to aliasing noise removal, equalization, and amplification by a low-pass filter 8, an equalizer 9, and an amplifier 10, and then emitted from a receiver 11.

On the other hand, if the speaker identification unit 14 determines that the input voice signal is a voice uttered by the wearer himself,
The DSP circuit 12 outputs the input digital signal to the D / A conversion circuit 7 without decompressing the input digital signal. As a result, the voice uttered by the wearer is equalized and amplified, but is emitted from the receiver 11 without performing temporal expansion.

FIG. 2 is a flowchart showing signal processing of a hearing aid according to an embodiment of the present invention. Microphone 1
The audio signal fetched in step (1) is cut in the high-frequency range by the anti-aliasing filter 2 and converted into a digital signal by the A / D conversion circuit 3 (n1). This digital signal is
The data is sequentially taken into the memory circuit 13 via the DSP circuit 12 (n2). The memory circuit 13 stores N data defined as one frame, and the speaker identification unit 1
In step 4, it is determined whether the speaker is a wearer (n3).
If it is determined that the speaker is not the wearer, the memory circuit 1
3 is subjected to speech speed conversion processing (n4 to n12), and the digital data subjected to the speech speed conversion processing is output to the D / A conversion circuit 7. On the other hand, if the speaker is identified as the wearer, the speech rate conversion processing is not performed, and the input digital signal is output to the D / A conversion circuit 7.

Here, the speech speed conversion processing of n4 to n12 will be described. First, it is determined whether this frame is a voice section or a silent section. Therefore, the memory circuit 1
3. The short-time average audio power E1 of the frame data (N sampling data) stored in No. 3 is obtained (n
4). The short-time average audio power E1 is a value obtained by dividing the sum of squares of each data in the frame section by 1 / N. The short-time average audio power E1 is compared with a preset threshold value P1, and when E1> P1, it is determined as a voice section, and when E1 <P1, it is determined as a silent section (n5).

If it is determined in step n5 that the data is in the voice section, the frequency fz of this data is calculated (n6). The frequency fz is extracted from the number of zero crossings of the frame data. Then, the frequency is analyzed to determine whether the sound is a voiced sound or an unvoiced sound (n7). Here, voiced sounds are mainly vowels, and unvoiced sounds are mainly consonants. For this reason, the frequency included in the voiced sound and the frequency included in the unvoiced sound are generally high in unvoiced sound and low in voiced sound. Therefore,
When the number of zero crossings is small (when the frequency is low), it is a voiced sound, and when the number of zero crossings is large (when the frequency is high).
It can be identified as unvoiced. In this embodiment, the frequency fz obtained from the number of zero crossings is compared with a preset threshold frequency fz1, and if fz> fz1, it is identified as an unvoiced sound section, and if fz <fz1, it is identified as a voiced sound section. did.

If it is determined in n7 that the section is a voiced sound section, the frame data is cut out with an appropriate window width, and an autocorrelation function is obtained (n8). Then, a pitch period is obtained from the obtained autocorrelation function (n9), and a zero crossing point having a positive differential coefficient closest to the start point of the section is set as a start point, and one of the start points of the next pitch section after the pitch section length is determined. The expansion waveform is cut out using the immediately preceding data as the end point (n10). Then, the waveform is repeated by connecting data of the pitch section length from the start point of the extraction as the next data (n1).
1). On the other hand, if it is determined in n7 that the section is a voiceless sound section, it is output without performing the processing of n8 to n11. That is,
The expansion processing is performed on the vowel section at a preset magnification, and the expansion processing is not performed on the consonant section.

If a silent section is identified at n5, a silent section having a preset length is inserted (n12).

By performing the above processing, the hearing aid converts the voice uttered by another person into a speech speed and outputs it, and outputs the voice uttered by the wearer without performing the speech speed conversion process. Therefore,
The wearer of the hearing aid can make the next utterance normally, and can hear the voice uttered by another person slowly and clearly.

Here, the process of identifying whether the speaker at n3 is a wearer will be described in detail.

First, a method of identifying whether a speaker is a wearer using solid-borne sound will be described.

FIG. 3 is a diagram showing a configuration of a speaker identification unit for identifying a speaker by using a solid-borne sound. Speaker identification unit 14
Is a bone conduction microphone 21 for detecting vibration or mechanical change of an object in direct contact, such as an acceleration sensor or a pressure sensor;
An amplifier 22 for amplifying the output of the bone conduction microphone 21, a comparison data storage circuit 23 for storing comparison data, and a comparator 24 for comparing the output of the amplifier 22 with the comparison data
And

The bone conduction microphone 21 is mounted on a portion of the hearing aid that is to be inserted into the ear canal. In the state mounted on the wearer, the bone conduction microphone 2 1 is mounted so as to be in contact through the structure directly or held in the ear canal inner wall of the wearer. Bone conduction microphone 21 detects the vibration of the ear canal inner wall caused by outgoing voice wearer, and outputs this into an electric signal. This signal is amplified by the amplifier 22. The amplification factor of the amplifier 22 is set in advance. The comparator 24 compares the signal amplified by the amplifier 22 with the comparison data stored in the comparison data storage circuit 23. When the signal amplified by the amplifier 22 in this comparison is larger than the comparison data, the comparator 24 outputs a signal indicating that the speaker is the wearer, and otherwise, the speaker is not the wearer. Is output. When the input signal is a signal indicating that the speaker is the wearer, the signal processing unit 6 performs the D / A conversion circuit 7 without performing the speech speed conversion process on the audio signal captured by the microphone 1. Output to

FIG. 4 is a diagram showing a configuration for discriminating whether or not a speaker is a wearer from a spatially propagating sound. In this circuit, a speaker identification microphone 31 for detecting a voice uttered by a wearer is used. The speaker identification microphone 31 is, for example, an earphone type, which is inserted into the ear of the wearer so as to detect the voice of the wearer loudly. Is detected. This circuit has the following configuration so as not to mistake this as the voice of the wearer.

In FIG. 4, the audio signal captured by the speaker identification microphone 31 is amplified by the amplifier 33, and the voice signal captured by the microphone 1 is amplified by the amplifier 32. The comparator 34 compares the signal B output from the amplifier 32 with the signal A output from the amplifier 33. The amplification factors of the amplifiers 32 and 33 are individually determined according to the positions where the microphone 1 and the speaker identification microphone 31 are attached. Comparator 34 determines when signal A is greater than signal B,
A signal indicating that the speaker is the wearer is output.

When this signal is input, the DSP circuit 12 prohibits the speech speed conversion processing for the voice signal fetched from the microphone 1.

With the above configuration, the voice uttered by a person other than the wearer becomes louder, so that even if the speaker identification microphone 31 takes in a large volume, the voice is not erroneously recognized as the voice of the wearer.

In FIG. 4, the signal A and the signal B are compared by an analog circuit. However, the comparison can be performed by using the function of the DSP circuit 12. In this case, the output signal of the speaker identification microphone 31 is A / D converted and input to the DSP circuit 12, and the operation shown in FIG. 5 is performed. First, the signal A received by the speaker identification microphone 31 and the signal B received by the microphone 1 are amplified by predetermined coefficients (n21, n
22) Compare the amplified signal A (n23). The coefficients for the signals A and B are individually determined according to the positions where the microphone 1 and the speaker identification microphone 31 are attached. When the signal A is larger than the signal B as a result of the comparison, it is determined that the speaker is the wearer (n24),
Otherwise, it is determined that the speaker is not the wearer (n2
5). In this way, even if the occurrence of a person other than the wearer is extremely large, the speaker can be accurately identified. The signal processing section 6 prohibits the above-mentioned speech speed conversion processing based on this signal.

Next, a description will be given of a hearing aid which stores the characteristics of the voice emitted by the wearer and determines whether or not this signal is the voice of the wearer based on the characteristics of the input audio signal.

FIG. 6 is a block diagram of the hearing aid. The difference between this embodiment and the hearing aid shown in FIG. 1 is that the speaker identification unit 14 is not provided and the wearer's voice feature storage area 13 a for storing the features of the voice emitted by the wearer in the memory circuit 13. In that it has

The operation of the hearing aid differs from that of the above embodiment only in the processing for identifying whether the speaker is a wearer, and the other processing (speech rate conversion processing, etc.) is the same. is there. Here, only the process for identifying whether the speaker is the wearer will be described, and the description of the other operations will be omitted.

FIG. 7 is a flowchart showing the speaker identification processing of this embodiment. The wearer of the hearing aid stores in advance the features of the voice uttered by the hearing aid in the wearer's voice feature storage area 13a (see FIG. 7A). In order to store the characteristics of the voice uttered by the user in the wearer's voice characteristic storage area 13a, the hearing aid is switched to the registration mode, and the microphone 1 captures the user's own voice (n31). This audio signal passes through the anti-aliasing filter 2, and A
The signal is converted into a digital signal by the / D conversion circuit 3 and sent to the signal processing unit 6. In the signal processing section 6, the DSP circuit 12 performs a linear prediction analysis of this signal (n32), and extracts the features of the voice uttered by the wearer (n33). Then, the extracted voice features are stored in the wearer voice feature storage area 13a (n34).

In the normal mode (when in use) (see FIG. 7B), when the hearing aid captures an audio signal with the microphone 1 (n41), the hearing aid converts the captured audio signal into a signal along the same path as described above. Input to the processing unit 6. In the signal processing unit 6, the DSP circuit 12 performs a linear prediction analysis of the signal (n42), and extracts the features of the audio signal (n4).
3). Then, the voice feature stored in the wearer voice feature storage area 13a is compared with the voice feature extracted in n43, and the similarity is evaluated. If the evaluation point in this evaluation is larger than a preset value, the speaker is identified as the wearer, and if the evaluation point is smaller than the preset value, the speaker is Identify that you are not a person.

It should be noted, the speaker identification section 1 4 to a switch, the wearer himself may be caused to enter an identification signal. For example, if the signal processing unit 6 is configured to inhibit the speech speed conversion processing while the switch is on, the wearer realizes the function of the present invention by turning on this switch only while he / she is speaking. can do. Conversely, if the signal processing unit 6 is configured to inhibit the speech speed conversion processing while the switch is off, the wearer need only turn off this switch while he or she is speaking. Note that this switch is not a main switch for stopping the operation of the hearing aid, but a switch for permitting / prohibiting the speech speed conversion process, and the voice input from the microphone 1 even while the speech speed conversion process is prohibited. The signal is amplified and the receiver 1
1 is output.

As described above, when the captured audio signal is a voice uttered by the wearer, the hearing aid outputs this audio signal without performing the speech speed conversion process, and the captured audio signal is generated by the wearer. If it is not a voice, the voice signal is output after performing speech speed conversion processing. Therefore, the wearer of the hearing aid
Since the user listens to his / her own voice that has not been subjected to speech speed conversion and gives feedback to the next generation, the next utterance can be performed normally. In addition, since the user hears the voice uttered by the other party whose speech speed has been converted, the user can hear the voice slowly and clearly.

It should be noted that the voice identification means described in the claims corresponding to the embodiment can be limited as follows. The voice identification means is attached to a portion inserted into the ear hole,
A bone conduction microphone that directly or indirectly contacts the inner wall of the ear canal of the wearer, a comparison unit that compares the output of the bone conduction microphone with a preset threshold, and the output of the bone conduction microphone is the comparison unit. When the sound signal is larger than the threshold, the sound signal is identified as a sound signal emitted by the wearer.

The voice discriminating means is attached to a portion inserted into the ear canal, and compares a speaker discriminating microphone for converting air vibration into an electric signal with a signal picked up by the speaker discriminating microphone. And a means for identifying that the captured audio signal is a voice signal emitted by the wearer when the audio signal captured by the comparison means is smaller than the signal captured by the speaker identification microphone.

The voice identification means is a switch.

The voice discriminating means includes a user voice storing means for linearly analyzing a voice uttered by the wearer to extract and store a feature, and a feature extracting means for linearly analyzing a captured voice signal to extract a feature. And comparing the feature extracted by the feature extracting means with the feature stored in the wearer's voice storage means to identify whether or not the signal fetched from the similarity is a voice emitted by the wearer.

[0042]

As in the above, according to the present invention, according to the present invention, a hearing aid is not I line the speech speed conversion in the voice that the wearer has issued, perform the speech speed conversion for the voice of others is issued. Therefore, the wearer of the hearing aid can listen to his or her own voice that has not been subjected to the speech speed conversion, and can normally perform the next utterance. Also, you can slowly and clearly hear the voices of others.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a hearing aid having a speech speed conversion function according to an embodiment of the present invention.

FIG. 2 is a flowchart showing signal processing of a hearing aid according to an embodiment of the present invention;

FIG. 3 is a diagram illustrating a configuration of a speaker identification unit that identifies a speaker by using a solid-borne sound;

FIG. 4 is a diagram illustrating a configuration of a speaker identification unit that identifies a speaker using airborne sound;

FIG. 5 is a flowchart showing processing for identifying a speaker using airborne sound.

FIG. 6 is a block diagram showing a configuration of a hearing aid having a speech speed conversion function according to another embodiment of the present invention;

FIG. 7 is a flowchart showing a speaker identification process according to another embodiment of the present invention;

[Explanation of symbols]

1 microphone, 6 signal processor, 1 4 - speaker identification section 1 2-DSP circuit, 1 3 - memory circuit

──────────────────────────────────────────────────続 き Continued on the front page (58) Field surveyed (Int.Cl. ⁶ , DB name) H04R 25/00 G10L 3/00

Claims (2)

(57) [Claims] 1. A hearing aid provided with a speech speed conversion means for extending a fetched voice signal on a time axis, wherein the hearing aid is attached to a portion to be inserted into an ear canal and is attached to an inner wall of an ear canal of a wearer.
Equipped with solid-borne sound detection means that makes direct or indirect contact
A voice identification unit for identifying whether or not the captured audio signal is a voice uttered by the wearer; and prohibiting the speech speed conversion unit when the voice identification unit identifies the voice of the wearer. A hearing aid having a speech speed conversion function, comprising: speech speed conversion inhibiting means. 2. A method for extending a captured audio signal on a time axis.
In a hearing aid equipped with a speech speed converting means, a voice signal is detected in a user's ear canal, which is attached to an ear canal.
Has a speaker identification microphone, captures audio signals and the ear canal
Audio signal captured by comparing the
A sound that identifies whether the issue is from the wearer
The identification means and the voice identification means have identified the voice of the wearer.
Sometimes, a speech speed conversion prohibition hand that prohibits the speech speed conversion means.
Hearing aid with a speech speed conversion feature, characterized in that it comprises a stage, a
vessel.