JP5520456B2 - Binaural sound collection and playback system - Google Patents

Description

  The present invention relates to a technique for binaural sound pickup and reproduction.

Conventionally, it is known binaural sound pickup reproducing method as a method of recording and playback of ambient sound environment of the recording's also include directional information of the sound source.
This binaural sound collecting / reproducing method is a stereo sound collecting method for collecting sound by arranging microphones in the vicinity of or in the ear canal entrances of the left and right ears.
By reproducing the sound signal collected by the microphone near the entrance of the ear canal using headphones or the like, the sound field of the collected place is more faithfully reproduced.

In the binaural sound collection and reproduction method, sound is collected by a microphone system called a dummy head in which a microphone is installed in the vicinity of the entrance of the ear canal or in the ear canal. However, in this case, because the shape of the listener's head shape and a dummy head to listen to the picked-up audio signals different, when reproduction of the audio signal, is difficult to faithfully reproduce the sound field at the time of sound collection It is.

For this reason, it is conceivable as a technique for effectively realizing binaural that the sound signal picked up by the recording person's ear is reproduced by the recording person himself / herself as a listener (real head binaural sound collecting reproduction).
In this case, since a head-related transfer function (HRTF) at the time of sound collection and reproduction coincides in principle, a more faithful sound field reproduction is possible.

  As a technique related to the above-mentioned real head binaural sound collection reproduction, a technique is disclosed that uses an adaptive filter to identify (calculate) transfer characteristics including the characteristics of headphones and microphones, thereby reducing the deterioration of the binaural effect in the reproduced sound. (Patent Document 1).

  In this related technology, when playing back binaural signals picked up by a real head using headphones, the microphone used for picking up the sound (or a microphone with similar characteristics) is worn on the ear, and the headphones are also worn. Then, the characteristics of the microphone and the headphones are obtained by sequentially obtaining the characteristics opposite to the synthesized characteristics from the headphones to the microphone by the adaptive filter, and superimposing the obtained synthesized inverse characteristics on the normal audio signal as a binaural signal by the correction filter. to correct.

In addition, as a related technique for reproducing binaurally picked-up sound with a loudspeaker instead of headphones, a listener from a left loudspeaker installed on the front left side of the listener among the sounds reproduced from the loudspeaker. A method for canceling the sound that reaches the right ear (crosstalk) and the sound that reaches the left ear of the listener (crosstalk) from the right loudspeaker installed on the front right side of the listener is disclosed ( Patent Document 2).
JP 2005-318061 A JP 2000-278800 A

However, in the above-mentioned Patent Document 1, the correction band and the non-correction band are divided by a specific threshold, and in this case, it is difficult to faithfully reproduce the binaural signal that has been collected in the listener's ear. There is.
Also, when binaural playback of binaurally picked-up sound, it is necessary to wear headphones, and binaural playback using a loudspeaker is not assumed.

In Patent Document 2, when reproducing binaural sound pickup signal by the loudspeaker (reproduction speaker), since the listener's position and the reproduction speaker is fixed, when the position of the listener is changed, there is a disadvantage that can not be made to follow the change in the filter coefficients of its (characteristics).

[Object of invention]
An object of the present invention is to provide a binaural sound collecting / reproducing system that improves the inconvenience of the related technology and reproduces and reproduces binaural sound faithfully by a sound field at the time of sound collecting. .

In order to achieve the above object, a binaural sound collecting / reproducing system according to the present invention performs a predetermined correction on a recorded audio signal pre-recorded by a sound collecting microphone and supplies the corrected sound to a speaker, and is reproduced by the speaker. An adaptive filter that corrects a reproduced audio signal that is collected by the sound collecting microphone provided in advance in the auditory canal region of the listener and a corrected recorded audio signal that is output from the correction filter are acquired. And calculating a difference between the delay signal output means for outputting after delaying a preset time, the reproduced audio signal acquired via the adaptive filter and the recorded audio signal output from the delay signal output means, An adder with a residual data output function for sending the difference to the adaptive filter,
The adaptive filter, an inverse characteristic identification function of calculating a transfer characteristic opposite characteristics from the speaker to the sound collecting microphone based on the reproduced audio signal comprising picked up by the sound pickup microphone and said difference And the correction filter has a function of correcting the recorded audio signal based on the characteristic calculated by the adaptive filter,
Further, the adaptive filter is configured to have a filter characteristic identification function for identifying characteristics in the adaptive filter so that the difference is minimized based on the difference sent from the adder.

In addition , the binaural sound collecting and reproducing system according to the present invention includes a two-channel loudspeaker set in advance to the left and right corresponding to the left and right ears of the listener, and each loudspeaker for the recorded audio signal supplied to each loudspeaker. And a correction filter for performing a predetermined correction individually for each,
For the recorded audio signal supplied to each of the left and right loudspeakers via the correction filter, the voice reaching the right ear of the listener from the left loudspeaker and the left ear of the listener from the right loudspeaker It has a crosstalk cancellation part that performs correction to cancel the sound that reaches
The crosstalk cancellation unit
Corresponding to each of the left and right loudspeakers, correction is performed on the reproduced audio signal that is reproduced by the speaker and collected in advance by the sound collecting microphone provided in the auditory canal region of the listener. An adaptive filter, delayed signal output means for acquiring a recorded audio signal after correction output from the correction filter and outputting it after a predetermined time delay, a reproduced audio signal acquired via the adaptive filter, and the An adder having a residual data output function for calculating a difference with the recorded audio signal output from the delayed signal output means and sending the difference to the adaptive filter;
Inverse characteristic identification function for each adaptive filter to calculate a characteristic opposite to the transmission characteristic from the speaker to the sound collecting microphone based on the reproduced sound signal picked up by the sound collecting microphone and the difference. The correction filter has a function of correcting the recorded audio signal based on the characteristics calculated by the adaptive filter,
Furthermore, each adaptive filter has a configuration in which a filter characteristic identification function for identifying the characteristic in the adaptive filter is provided so that the difference is minimized based on the difference sent from the adder. .

Further , the binaural sound collecting / reproducing system according to the present invention includes a two-channel loudspeaker set in advance to the left and right corresponding to the left and right ears of the listener, and each loudspeaker for the recorded audio signal supplied to each loudspeaker. And a correction filter for performing a predetermined correction individually for each,
For the recorded audio signal supplied to each of the left and right loudspeakers via the correction filter, the voice reaching the right ear of the listener from the left loudspeaker and the left ear of the listener from the right loudspeaker It has a crosstalk cancellation part that performs correction to cancel the sound that reaches
The crosstalk cancellation unit
For the recorded audio signal, transmission characteristics opposite to the transmission characteristics between the left and right speakers and the left and right ears of the listener, and the left and right speakers and the right and left ears of the listener. A reproduction audio correction filter for performing correction to give a transfer characteristic between them,
Further, the reproduction sound correction filter in the reproduction sound correction filter based on the reproduction sound signal reproduced by each loudspeaker and previously picked up by the sound pickup microphone provided in the left and right ear canal regions of the listener and the recorded sound signal. A transfer characteristic is calculated, and a reproduction voice adaptive filter that sets the transfer characteristic in the reproduction voice correction filter is provided.

The present invention than the structure described above, according to this, an adaptive filter for calculating an inverse characteristic of the transfer characteristic opposite from the speaker to the sound collection microphone, recording audio signal reproduced by the loudspeaker By adopting a configuration including a correction filter that performs a correction process using the calculated inverse characteristics, the sound field at the time of sound collection can be faithfully reproduced in the ear canal region where the sound collection microphone is attached. it can.

[Embodiment 1]
Hereinafter, a first embodiment of the present invention, with reference to FIGS. 1 to 10 will be described the configuration contents.

Binaural sound pickup reproducing system of this embodiment 1, as shown in FIG. 1, a binaural reproduction apparatus (hereinafter referred to as "reproducing apparatus") 1 for reproducing audio signals binaural recording, supplied from the reproducing apparatus 1 A headphone speaker (hereinafter referred to as “headphones”) 3 that converts an audio signal into sound and an audio signal output from the headphones 3 (3R, 3L) are mounted near the ear canal in the left and right ears of the listener 5. A microphone 2 (2R, 2L) for sound is provided.

Furthermore, headphones 3R, 3L left and right ears their respective to the mounted microphone 2R of the listener 5, it is mounted 2L to to cover each microphone 2R to the right playback system R of the reproducing apparatus 1, the microphone 2L Are connected to the left reproduction system L of the reproduction apparatus 1 and input the collected sound to the reproduction apparatus 1.

The microphone (sound collecting microphone) 2 is composed of 2R attached to the listener's 5 right ear and 2L attached to the left ear.
As shown in FIGS. 2A and 2B, the microphone 2R is inserted into an ear hook 21 to be put on the pinna of the listener 5, a microphone unit 22 for collecting sound, and an ear canal of the listener 5. The probe 23 is configured. Note that the microphone 2L has the same configuration as the microphone 2R.

Here, binaural recording, which is a technique for recording and storing the audio played back in this embodiment (
Real head recording) will be explained.

First, microphones 2R and 2L are mounted in the vicinity of the ear canal of each of the right ear and the left ear of a recording person 4 (see FIG. 3) who performs binaural recording at a voice recording site.
Here, each of the microphones 2R attached to the right ear of the recording person 4 can listen to the sound signal (live sound) reaching the both ears of the recording person 4 without being disturbed. It shall have a probe type that does not block the road. FIG. 3 illustrates a case where the microphone 2R is attached to the right ear canal.

In this state, the sound at the sound pickup site is picked up by the microphones 2R and 2L, and the picked up sound is recorded in stereo on the recording device 6 (see FIG. 4) .
The recording device 6 may be a digital device or an analog device as long as it is a device capable of collecting stereo sound.

  Here, the audio signals collected by the microphones 2R and 2L are input to the recording device 6 as shown in FIG. 4 as a right channel audio signal (SR) and a left channel audio signal (SL), respectively. The data is recorded in a data storage unit 7 such as a disk medium or a memory card provided in the recording device 6.

The channel audio signals SL and SR to be recorded on the recording device 6 are input to a recording processing unit 8 provided in the recording device 6 in advance, and are compressed and encoded as necessary, and music in a preset format. It is recorded in the data storage unit 7 as data.

3 and 4 exemplify the case where the microphone 2R is attached to the right ear of the recording person 4, but the left ear of the recording person 5 is similarly attached to the microphone 2L. And

Next, the configuration of the playback device 1 will be described.
The playback device 1 is connected to the headphones 3 (3R, 3L), and plays back the input audio signals SR and SL recorded in the data storage unit 7.
As shown in FIG. 5, the right reproduction system R of the reproduction apparatus 1 includes a correction filter 12R, a delay element 13R, an adaptive processing unit 14R, and an adder 15R. Similarly to the case of FIG. 5 , the left reproduction system L of the reproduction apparatus 1 includes a correction filter 12L, a delay element 13L, an adaptive processing unit 14L, and an adder 15L (not shown) .

On the other hand, the reproduction processing unit 11 executes a real head recorded binaural recorded speech signals read from the (audio signals SL and SR) data storage unit 7, expansion decoding soil necessary processing as shown in FIG. 5 The digital audio data DR and DL are output to the correction filters (12R, 12L) (digital audio data output function).
Here, the configuration content of the right channel reproduction system R of the reproduction apparatus 1 will be described in detail with reference to FIG .

In FIG. 5, the correction filter 12R performs a correction process on the supplied digital audio data (hereinafter referred to as “binaural sound collection signal”) DR according to a preset characteristic (filter coefficient) to obtain an audio signal s (k ) To the headphone 3R and the delay element 13R (
Audio correction supply function).

  The correction filter 12R is constituted by, for example, an FIR filter or the like, and its filter characteristic is updated by changing and setting the filter coefficient by an adaptive filter 14R described later.

Further, the correction filter 12R has a superimposition processing function for performing processing (convolution processing) for superimposing the above characteristics on the binaural sound pickup signal (DR) picked up by the real head.
Accordingly, the correction filter 12 (12R, 12L) can be if the audio signal binaural sound pickup is headphones listening, the correction of the characteristics of the until ear headphones in the voice signal to the flat.

Incidentally, the correction filter 12R is, when the execution of the voice correction supplying function, the inverse characteristic to be described later, corrects the binaural sound pickup audio signal (filtering). Here, the correction filter 12R performs a process of giving an inverse characteristic based on a transfer function from the headphone 3R to the microphone 2R to the binaural sound pickup signal (DR).

Next, the delay element 13R performs a delay process on the input voice data s (k) from the correction filter 12R for a preset time t, and the delayed voice data d (k−
t) is output to the adder 15R (delayed data transmission function).
Here, it is assumed that a time value including a sound propagation time from the headphones 3R to the microphone 2R is set in advance as the delay time t in the delay data transmission function.

The adaptive filter 14R performs a correction process on the reproduced audio data ( reproduced audio signal ) x (k) supplied from the microphone 2R based on a preset filter coefficient, and the corrected audio data is A reproduction signal correction output function for outputting to the adder 15 as an output y (k) is provided.

Further, the adaptive filter 14R changes the characteristic (filter coefficient) of the adaptive filter 14R itself so that the output y (k) approximates the desired signal (here, the delayed speech data d (k -t )). It has a characteristic update control function for updating. Here, the adaptive filter 14R minimizes the difference e (k) output from the adder 15R based on the difference e (k) and the reproduced audio data x (k) sent from the adder 15R as described later. Thus, the filter coefficient is self-set (inverse characteristic identification processing function).

By the way, the characteristics of the reproduced signal (reproduced audio data x (k)) collected by the microphone 2R are the characteristics of the headphone 3R and the position of the microphone 2R from the position of the headphone 3R ( and the position of the microphone 2L from the position of the headphone 3L). ) And the characteristics of the microphone 2R (hereinafter referred to as “composite characteristics”).

  Therefore, if a characteristic that approximates this synthesis characteristic and is opposite to the synthesis characteristic (reverse characteristic: synthesis inverse characteristic) is obtained, the characteristics of the headphones 3R, from the position of the headphones 3R to the position of the microphone 2R are determined by the synthesis inverse characteristic. And the characteristics of the microphone 2R (referred to as “characteristics from the headphones 3 to the microphone 2”) can be corrected.

In addition, since the composite inverse characteristic described above includes a characteristic opposite to the acoustic transmission characteristic from the headphones 3R to the microphone 2R, sound is transmitted with a flat transmission characteristic from the position of the headphones 3R to the position of the microphone 2R. Will be. That is, it is possible to correct the spatial transfer characteristics that are influenced by the shape of the ear that is different for each person when using ear-covering headphones or the like.

For this reason, in the correction filter 12R, this combined inverse characteristic is used for binaurally picked up sound (
By superimposing on s (k)), the synthesis characteristic is corrected, and the binaural sound is reproduced faithfully in the ear of the listener 5.

The adder 15R performs a process of adding the output y (k) input as the negative component and the delayed audio data d (k −t ) input as the positive component (addition processing function).
Here, a process of subtracting the output y (k) that is the voice data subjected to the filtering process from the delayed voice data d (k −t ) is performed.
The adder 15R sends residual data (difference) e (k) as an addition (subtraction) result to the adaptive filter 14R (residual data output function).

Here, it is assumed that the adaptive processing unit 16R is configured by the adaptive filter 14R and the addition processing unit 12R. The adaptive processing section 16R, the signal from the microphone 2R (reproduced audio data) x (k) and signal from the delay element 13R and the (delayed audio data) d (k -t) is input, the adaptive filter 14R , such that the output y (k) is approximated signal to desired and that delay the audio data d (k -t), that is, as the power of the difference e (k) is small, the characteristics of the adaptive filter 14R itself Works to fix.
Thereby, the adaptive filter 14R can identify the reverse characteristic from the headphones 3R to the microphone 2R.

  Here, when the LMS (Least Mean Square) method or the learning identification method (Normalize Least Mean Square), which are general adaptive algorithms, is applied to the inverse characteristic identification processing function in the adaptive filter 14R, the identification process of the inverse characteristic is performed. It takes time, and it is difficult to perform correction that follows changes in the wearing state of the headphones.

Therefore, in the first embodiment, the adaptive filters 14R and 14L for identifying the reverse characteristics from the headphone to the microphone are configured by the high-speed H∞ filter (FHF: high-speed calculation filter) disclosed in Japanese Patent No. 4067269. To do. This makes it possible to identify the reverse characteristic at high speed.

Further, by sequentially performing the identification of inverse characteristic with FHF, even when the variation of the resulting characteristics to variations in the mounted state of the headphone 3 by the listener 5 occurs, correction dynamically follow this variation filter The characteristics of 12R can be updated quickly.
In the first embodiment, the right channel reproduction system R has been described, but the left channel reproduction system L also includes a correction filter 12R, a delay element 13R, an adaptive filter 14R, and an adder 15R of the right channel reproduction system R, respectively. The correction filter 12L, the delay element 13L, the adaptive filter 14L, and the adder 15L are configured to correspond to each other and include the headphone 3L and the microphone 2L.

By the way, the transfer characteristic (transfer characteristic between the headphone and the microphone) from the headphone 3R to the microphone 2R or from the headphone 3L to the microphone 2L can be identified by, for example, the configuration shown in FIG.
Here, as in the present embodiment (FIG. 5), the microphone 2 (2R, 2L) and the headphone 3 (3R, 3L) are attached to the listener.

The recorded sound (right channel) recorded by the real head is supplied to the headphone 3R and the preset adaptive filter X as shown in FIG. The recorded audio signal is reproduced by the headphones 3R, and the audio signal picked up by the microphone 2R and the audio signal input through the adaptive filter X are added, and this addition result (difference) e is used as the adaptive filter X. input.
In FIG. 6, the adaptive filter X identifies the filter characteristics of the adaptive filter so that the input addition result e is minimized. Thereby, the adaptive filter X can identify the transfer characteristic from the headphones 3R to the microphone 2R.

  Here, the time waveform of the transfer characteristic (synthesis characteristic) from the headphones (3R) to the microphone (2R) identified by the adaptive filter X is shown in FIG. FIG. 7B shows frequency characteristics obtained by Fourier transforming the time waveform of the transfer characteristics shown in FIG.

Next, FIG. 8 shows a configuration for identifying the reverse characteristic of the transfer characteristic (the transfer characteristic between the headphone and the microphone) from the headphone 3R to the microphone 2R or from the headphone 3L to the microphone 2L.
FIG. 8 is a simplified configuration of the first embodiment, and the adaptive filter Y corresponds to the adaptive filter 14R (FIG. 5).

As in the case of FIGS. 1 and 2 , a microphone 2 (2R, 2L) and a headphone 3 (3R, 3L) are attached to the listener as shown in FIG.

Here, the recorded sound (right channel) recorded by the real head is supplied to the headphone 3R and a preset delay element D (corresponding to the delay element 13R).
The recorded audio signal is reproduced by the headphone 3R, and the audio signal picked up by the microphone 2R is input to the adding unit via the adaptive filter Y. The adaptive filter Y performs a correction process on the audio signal based on preset filter coefficients.
Similarly to the delay element 13R of the first embodiment, the delay element D performs a delay process on the input recorded audio signal for a preset time and uses the delayed audio data subjected to the delay process as an adder. Output.

Similar to the adder 15R, the adder performs addition processing on the input signals and outputs the addition result.
Here, the audio signal from the adaptive filter Y and the audio signal input via the delay element D are added by the adder, and the addition result e is output to the adaptive filter Y.

The adaptive filter Y identifies the filter characteristics of the adaptive filter Y so that the addition result e from the adding unit is minimized.
Thereby, the adaptive filter Y can identify the reverse characteristic (reverse transfer characteristic) of the transfer characteristic from the headphones 3R to the microphone 2R.

  Here, the time waveform of the reverse transfer characteristic from the headphone to the microphone identified by the adaptive filter Y is shown in FIG. Further, FIG. 9B shows frequency characteristics obtained by performing Fourier transform on the time waveform of the transfer characteristics shown in FIG. Here, it can be seen that the frequency characteristic of FIG. 9B is a characteristic of a shape obtained by folding the frequency characteristic of FIG. 7B up and down.

Further, by performing a process (convolution process) for convolving the time waveform of FIG. 7 and the time waveform of FIG. 9, a time waveform in the form of a pulse is obtained as shown in FIG. FIG. 10B shows frequency characteristics obtained by performing Fourier transform on the time waveform of FIG.
The shape of this frequency characteristic is substantially flat, and it can be confirmed that the time waveform of the reverse transfer characteristic shown in FIG. 9 is the reverse characteristic of the transfer characteristic shown in FIG.
Thus, it can be seen that the configuration of the first embodiment allows the adaptive filter 14R corresponding to the adaptive filter Y to effectively identify the reverse transfer characteristic of the transfer characteristic from the headphone 3R to the microphone 2R.

As described above, it is possible to identify the transfer characteristic from the headphones to the microphone with the adaptive filter X in the configuration shown in FIG. 6, and then obtain the reverse transfer characteristic from the identified transfer characteristic.
However, in this case, there is a disadvantage that the amount of processing for obtaining the reverse transfer characteristic from the transfer characteristic becomes enormous. Further, the inverse characteristic obtained from the identified transfer characteristic (FIG. 7A) has a disadvantage that the transfer characteristic cannot be made flat when the convolution process is performed.

In contrast, in the first embodiment, since the adaptive filter 14R, 14 L are is configured to identify the inverse transfer characteristic from the headphone to the microphone directly, the inverse transfer characteristic for performing effectively the convolution processing, It can be obtained quickly and reliably.

[Description of Operation of First Embodiment]
Here, as described above, the case where the listener 5 wears the microphones 2R and 2L and the headphones 3R and 3L and reproduces the binaural signal recorded as described above will be described.

First, in FIG. 1, the reproduction processing unit 11 reads the audio data stored in the data storage unit 7, and supplies the right channel audio data DR and the left channel audio data DL to the correction filters 12R and 12L , respectively. To do.

Hereinafter, an operation flow in the right playback system in the first embodiment (FIG. 5).
The correction filter 12R performs a correction process using a preset filter coefficient (correction characteristic) on the sent audio data DR, and sends it to the headphones 3R and the delay element 13R as an audio signal s (k). .

  The audio signal s (k) supplied to the headphones 3R is D / A converted by a D / A (Digital / Analog) converter and supplied as an analog audio signal.

  In the headphone 3R, the audio signal s (k) sent from the correction filter 12R is reproduced (sounded) as the right channel sound, and the microphone 2R collects the right channel sound reproduced by the headphone 3R.

The adaptive filter 14R performs correction processing on the reproduced audio data x (k) collected by the microphone 2R based on a preset filter coefficient, and outputs it to the adder 12 as an output signal y (k). Output.
The reproduced audio data x (k) is A / D converted by an A / D (Analog / Digital) converter provided in advance in the reproducing apparatus 1 and supplied as a digital audio signal.

  The delay element 13 delays the audio signal s (k) from the correction filter 12R for a preset time, and adds the delayed audio data d (kt) subjected to the delay process to the adder 15R. Output to.

The adder 15R calculates a difference between the output y (k) corrected by the adaptive filter and the output signal d (kt) of the delay element, and sends the calculated difference e (k) to the adaptive filter 14R. The
In the adaptive filter 14R, the characteristics are self-set so as to minimize the difference e (k) based on the fed difference e (k) and the signal x (k). The identified inverse characteristic is notified to the correction filter 12R.

  Thereby, in the adaptive filter 14R, an inverse characteristic opposite to the synthesized characteristic (synthetic characteristic) from the headphone 3R to the microphone 2R attached to the ear of the listener 5 (here, the right ear) is identified.

  Next, the correction filter 12R performs the correction process by performing the convolution process on the audio data DR from the reproduction processing unit 11 using the identified inverse characteristic as the characteristic of the correction filter 2R itself, and supplies the correction result to the headphone 3R. .

In the first embodiment, the operations of the right channel reproduction system, the headphones 3R, and the microphone 2R have been described. However, the corresponding left channel reproduction system (correction filter 12L, delay element 13L, adaptive filter 14L, adder 15L), The headphones 3L and the microphone 2L operate in the same manner as described above.

As described above, in the first embodiment, the binaural sound pickup (sound) signal that has been binaurally picked up is corrected so as to flatten the characteristics between the headphones and the ears, and the binaural sound pickup sound is faithful. Reproduced in the listener's ear.

In addition, when binaural reproduction is performed according to the first embodiment, the correction characteristics of the correction filter can be updated sequentially, so that changes in characteristics due to changes in the wearing state of headphones during reproduction are also followed. Thus, the optimum characteristics are identified, and the playback sound can be filtered.

[Embodiment 2]
Next, Embodiment 2 according to the present invention will be described with reference to FIGS.
Here, the same reference numerals are used for the same parts as those in the first embodiment.
The second embodiment, in place of the headphone of the first embodiment, loudspeakers 200L in the front left of the listener 5, a point in which a loudspeaker 200R in the front right, the crosstalk cancellation unit in the playback device 100 (crosstalk (Cancellation process) The point provided with 50 is different from the case of the first embodiment.

Here, binaural sound recording (real head recording) stereo audio signal (recording signal)
When playing with the listener pair of stereo loudspeakers (hereinafter referred to as "loudspeaker") 200L, 200 installed in front of the R a, as shown in FIG. 11, the sound that reaches the right ear from the loudspeaker 200L There is a signal, an audio signal reaching the left ear from the loudspeaker 200R. These signals are called crosstalk.

Real head From been From signals (left and right channel audio signals), when reproducing by a pair of loudspeakers 200L, 200 R, the listener 5 side of listening to this, spatial sound under the influence of crosstalk It spreads and the sense of orientation is reduced. For this reason, as shown in FIG. 12, it is necessary to perform a crosstalk canceling process (crosstalk canceling process) on the reproduced left and right channel audio signals (x1 and x2).

The crosstalk cancellation process cancels a signal (crosstalk) that reaches the listener 5 from the left speaker y1 to the listener's right ear z2 by a signal from the right speaker y2. Similarly, in this crosstalk cancellation process, a signal (crosstalk) reaching the listener's left ear z1 from the right speaker y2 is also canceled by the signal from the left speaker.

Here, the schematic block diagram of this Embodiment 2 is shown in FIG.
Similarly to the first embodiment, microphones 2L and 2R are provided in both ears of the listener 5, respectively, a left loudspeaker (hereinafter referred to as “speaker”) 200L and a right loudspeaker (hereinafter referred to as “speaker”) in front of the listener 5. 200R is provided, and the playback apparatus 100 that provides audio signals to the speakers 200 L and 200 R is provided.

Reproducing apparatus 100 is connected to speaker 200 L Contact and 200 R, the data storage unit inputs recorded in the (same in the storage unit 7 of FIG. 1) audio signal SR, and SL reproduction processing unit 11 for reproducing ( (Not shown), a correction filter A for correcting the left channel audio signal DL input from the reproduction processing unit 11, and a correction filter for correcting the right channel audio signal DR similarly input from the reproduction processing unit 11. B.

In addition to the correction filters A and B, the reproduction apparatus 100 according to the second embodiment includes correction filters C and D in a crosstalk cancellation unit 50 described later. As shown in FIG. 13, the characteristics of each of these four filters are represented by transfer functions G11, G22, G12, and G21 between the left speaker (200L) and the right speaker (200R), respectively, and both ears of the listener. To be determined.

  Here, the transfer function between the left speaker and the left ear of the listener 5 is G11, the transfer function between the left speaker and the right ear of the listener 5 is G12, and between the right speaker and the listener's left ear. Is G21, and the transfer function between the right speaker and the right ear of the listener 5 is G22. Thereby, each transfer function changes when the positional relationship between the left and right speakers and the left and right ears of the listener 5 changes. 1 / G11 and 1 / G22 indicate inverse transfer functions of G11 and G22, respectively.

Here, 1 / G11 and 1 / G22 perform reproduction system compensation, and G12 / G22 and G21 / G11 are subtracted from the signal of the other channel, respectively. Thereby, a feedback loop is formed, and this configuration is called a feedback type crosstalk canceling device.
With such a configuration, by processing the reproduced left and right channel audio signals, the signal from the left speaker reaches only the left ear, and the signal from the right speaker reaches only the right ear.

However, when crosstalk cancellation (Fig. 12) is generally listener filter fixed characteristics assuming that are in the middle of the left and right speakers is set is mounted. However, in this case, when the listener moves or changes direction, the crosstalk cannot be canceled properly when the positional relationship between the speaker and the listener changes, and the localization of the audio to be listened to The effect (binaural effect) is lost.

Therefore, in the second embodiment, the microphones 2R and 2L of the present invention (generally referred to as “real head microphones”) are attached to both ears of the listener 5, thereby collecting the signal reproduced by the loudspeaker, The filter function is adaptively identified based on the signal supplied to the loudspeaker and the signal picked up by the real head microphone.
As a result, even if the listener moves and the positional relationship between the listener's ears and the loudspeaker changes, the filter function of the installed filter is adaptively updated, and optimal crosstalk cancellation is always achieved. It becomes possible to do.

  In this embodiment, in order to use an adaptive filter for crosstalk cancellation, G12 / G22 shown in FIG. 13 is decomposed into G12 and 1 / G22, and similarly G21 / G11 is decomposed into G21 and 1 / G11. (Figure 14).

This will be described in detail below.
As shown in FIG. 14, the binaural sound collecting and reproducing system of the present embodiment includes a binaural reproducing device (hereinafter referred to as “reproducing device”) 100 that reproduces an audio signal recorded in a real head, and an audio supplied from the reproducing device 100. The loudspeaker (hereinafter referred to as “speaker”) 200L and 200R for emitting a signal is provided. Further, as described above, in the vicinity of the auditory canal in the left and right ears of the listener 5, as in the first embodiment, the mounted microphones 2 (2L, 2R) are provided, and the microphones 200L and 200R are released. The microphones 2L and 2R pick up the sound to be sounded.

Further, as in the first embodiment, the playback device 100 reads the audio signal stored in the data storage unit 7 and corrects the left channel audio signal (DL) read by the playback processor 11. A signal correction unit 41 that is a filter that performs processing, a signal correction unit 42 that is a filter that performs correction processing on the right channel audio signal (corresponding to a “correction filter” respectively), and inputs through the signal correction units 41 and 42 and a crosstalk cancellation unit 50 for inputting an audio signal to by the speaker 200L performs correction processing on the audio signal, 200 R.

The signal correction unit 41 is a filter for correcting the left channel audio signal DL (correction filter A: FIG. 13), and is an inverse transfer function of the transfer function (G11) from the speaker 200L to the left ear of the listener 5. A filter coefficient (1 / G11) corresponding to is set.
The signal correction unit 41 gives the audio signal DL from the reproduction processing unit 11 a frequency characteristic based on the inverse function of the transfer function from the speaker 200L to the left ear of the listener 5. Thereby, the frequency characteristic of the sound from the speaker 200L to the left ear of the listener 5 can be corrected to be flat.
The filter coefficient (1 / G11) is calculated (identified) and set by a filter characteristic calculation unit 61 described later.

The signal correction unit 42 is a filter (correction filter B: FIG. 13) for correcting the right channel audio signal (hereinafter referred to as “audio signal”) DR, and is transmitted from the speaker 200R to the right ear of the listener 5. A filter coefficient (1 / G22) corresponding to the inverse transfer function of the function (G22) is set.
The filter coefficient (1 / G22) is calculated and set by a filter characteristic calculation unit 63 described later.
The signal correction unit 42 gives a frequency characteristic to the audio signal DR from the reproduction processing unit 11 based on an inverse function of a transfer function from the speaker 200R to the right ear of the listener 5. Thereby, the frequency characteristic of the sound from the speaker 200R to the right ear of the listener 5 can be corrected to be flat.

Crosstalk cancellation unit 50 performs addition processing respectively audio signals input via the signal correcting unit 41, the addition result to the speaker 200L, 200 respectively supplies the addition processing unit 31L to R, and 32R, Signal correction units 43, 44, 45, and 46 that perform a convolution process on the audio signal with a preset filter coefficient are provided.

Specifically, the crosstalk cancellation unit 50 has a CPU (Central Processing Unit).
And a control program which is realized by a storage unit such as a memory and a hard disk and controls signal processing for crosstalk cancellation.

The addition processing unit 31L described above includes the feedback signal b1 (k) corrected by the signal correction units 44 and 46 (corresponding to the correction filter D: FIG. 13), and the lch input signal corrected by the signal correction unit 41. lw (k) is input.
The addition processing unit 31L, the lch supply input signal lw (k) from the feedback signal b1 (k) is subtracted lch output signal lx (k) to the speaker 200L (input). Further, the addition processing unit 31L inputs the lch output signal lx (k) subjected to the addition processing to the signal correction unit 45 described later.

Similarly to the addition processing unit 31L , the addition processing unit 32R is a feedback signal b2 (k) corrected by the signal correction units 43 and 45 (corresponding to the correction filter C: FIG. 13).
The rch input signal rw (k) corrected by the signal correction unit 42 is input.
The addition processing unit 32R supplies (inputs) the rch output signal rx (k) obtained by subtracting the feedback signal b2 (k) from the rch input signal rw (k) to the speaker 200R. Further, the addition processing unit 32R inputs this rch output signal rx (k) to the signal correction unit 46 described later.

The signal correction unit 45 includes, for example, a FIR (Finite Impulse Response) filter and the like, and the audio signal lx (k) is input from the addition processing unit 31L.
In the signal correction unit 45, a filter coefficient corresponding to the inverse transfer function (1 / G22) of the transfer function (G22) from the speaker 200R to the right ear of the listener 5 is set.
The signal correction unit 45 gives frequency characteristics based on the inverse function (1 / G22) of the transfer function from the speaker 200R to the right ear of the listener 5 to the lch output signal lx (k) from the addition processing unit 31L. Is output to the signal correction unit 43 as an audio signal ly (k).

  The signal correction unit 45 is set with the same filter coefficient as the filter coefficient of the signal correction unit 42. The filter coefficient of the signal correction unit 45 is calculated and set by a filter characteristic calculation unit 61 described later. For this reason, the filter characteristic of the signal correction unit 45 is updated and set to be the same as the characteristic of the signal correction unit 42 based on the result of identification by the filter characteristic calculation unit 61.

  The signal correcting unit 43 performs a process of giving frequency characteristics based on the transfer function (G12) of the crosstalk from the speaker 200L to the right ear of the listener 5 to the audio signal ly (k) from the signal correcting unit 45. Do. The signal correction unit 43 outputs the feedback audio signal b2 (k) generated by this processing to the addition processing unit 32R (feedback audio signal transmission function).

Further, a filter coefficient corresponding to the transfer function (G12) in the crosstalk from the speaker 200L to the right ear of the listener 5 is set in the signal correction unit 43.
Further, the filter coefficient (G12) of the signal correction unit 43 is calculated and set by a filter characteristic calculation unit 63 described later. Therefore, the filter characteristic of the signal correction unit 43 is updated based on the result of identification by the filter characteristic calculation unit 63.

The addition processing unit 32R includes the feedback signal (minus component) b2 (k) corrected by the signal correction units 43 and 45 (corresponding to the correction filter C: FIG. 13), and the rch corrected by the signal correction unit 42. An input signal (plus component) rw (k) is input and an addition process is performed.
The addition processing unit 32R inputs the rch output signal rx (k) obtained by subtracting the feedback signal b2 from the rch input signal rw (k) to the speaker 200R and also to the signal correction unit 46 described later.

The signal correction unit 46 includes, for example, an FIR (Finite Impulse Response) filter, and receives an audio signal (the rch output signal rx (k)) from the addition processing unit 32R.
Further, a filter coefficient corresponding to the inverse function (1 / G11) of the transfer function (G11) from the speaker 200L to the left ear of the listener 5 is set in the signal correction unit 46.
The signal correction unit 46 has a frequency characteristic based on an inverse function (1 / G11) of a transfer function from the speaker 200L to the left ear of the listener 5 with respect to the rch output signal rx (k) output from the addition processing unit 32R. Is output to the signal correction unit 44 as an rch audio signal ry (k).

  The signal correction unit 46 is set to the same filter coefficient as that of the signal correction unit 41. That is, the filter coefficient of the signal correction unit 46 is calculated and set by the filter characteristic calculation unit 61 described later. Therefore, the filter characteristic of the signal correction unit 46 is updated based on the result of identification by the filter characteristic calculation unit 63.

The signal correction unit 44 performs a process of giving a frequency characteristic based on the transfer function (G21) in the crosstalk from the speaker 200R to the left ear of the listener 5 to the audio signal ry (k) from the signal correction unit 46. Do. The signal correction unit 44 outputs the feedback audio signal b1 (k) generated by this processing to the addition processing unit 31L (feedback audio signal transmission function).

Here, in the signal correction unit 44, a filter coefficient (G21) corresponding to a transfer function in crosstalk from the speaker 200R to the left ear of the listener 5 is set in advance.
The filter coefficient of the signal correction unit 44 is calculated and set by a filter characteristic calculation unit 64 described later. For this reason, the filter characteristic of the signal correction unit 44 is updated by the result of identification by the filter characteristic calculation unit 64.

  Next, filter characteristic calculators 61 to 64 that are provided in the reproducing apparatus 100 and identify characteristics (transfer functions) of the signal correction units 41, 42, 43, 44, 45, and 46 will be described with reference to FIGS. Based on the explanation.

As shown in FIG. 15, the filter characteristic calculation unit 61 is connected to the reproduction processing unit 11 and the microphone 2L, and receives the left channel audio signal DL input from the reproduction processing unit 11 and the reproduction audio signal lp (x (x) from the microphone 2L. ) and calculates the filter coefficients of the signal correction unit 41, 4 6 based on.
The filter characteristic calculation unit 61 (corresponding to an adaptive filter) includes a delay unit 33, an adding unit 34, and an adaptive filter A as shown in FIG.
The delay means 33 delays the input real head sound collection left signal (DL) by a preset time (t) and outputs the delayed audio data ad (kt) to the addition means 34. (Delayed data transmission function).

The adding means 34 performs a process of adding the output ay (k) (minus component) input through the adaptive filter A and the delayed audio signal ad (kt) (plus component) (addition processing function). ).
Here, the adding means 34 performs a process of subtracting the output ay (k) from the delayed speech signal ad (kt ) , generates a residual signal ae (k) , and sends it to the adaptive filter A (residual signal). Output function).

The adaptive filter A includes a digital filter Aa and coefficient determination means Ab that determines the filter coefficient of the digital filter Aa .
The digital filter A a performs convolution processing on the reproduced audio data lp (k) input from the microphone 2L with a preset filter coefficient to generate an output ay (k). The digital filter A a outputs the generated output ay (k) to the adding means 34 (reproduction signal correction output function).

The coefficient determining means Ab reaches the left ear of the listener 5 from the speaker 200L based on the residual signal ae (k) output from the adding means 34 and the reproduced audio data lp (k) input from the microphone 2L. transfer characteristics (G11) opposite the inverse transfer characteristic of the audio signal: identify (1 / G11 inverse characteristic), in accordance with the reverse transfer characteristic (function), corrects the filter coefficient of the digital filter a a. This correction is performed so that the residual signal ae (k) input from the adding means 34 to the adaptive filter A is as small as possible.
Thereby, it is possible to perform an adaptive process for approximating the output ay (k) from the adaptive filter A to the delayed speech signal ad (kt).

Here, the filter coefficients of the corrected digital filter A a is the adaptive filter A, is notified to the signal correcting unit 41 and 46 (the identifying coefficient notification function). Thereby, the filter coefficients of the signal correction units 41 and 46 are set in accordance with the filter coefficient of the digital filter Aa (filter coefficient correction setting function).

As shown in FIG. 16, the filter characteristic calculation unit 62 includes an adding unit 35 and an adaptive filter B.
The adder 35 receives the output by (k) from the adaptive filter B and the reproduced audio data (microphone input signal) rp (k) input from the microphone 2R.
The adding means 35 performs a process of adding the microphone input signal (plus component) rp (k) and the output signal by (k) (minus component) (addition processing function).
Here, the residual signal be (k) is generated by subtracting the output signal by (k) from the input signal rp (k). The adding means 35 sends the residual signal be (k) to the adaptive filter B (residual signal output function).

Adaptive filter B includes a digital filter Ba, and coefficient determination section B b for determining a filter coefficient of the digital filter Ba.
The digital filter Ba performs convolution processing on the input real head sound collection left signal (DL) with a preset filter coefficient to generate an output by (k). The digital filter Ba outputs the generated output by (k) to the adding means 35 (recording signal correction output function).

The coefficient determining means B b calculates the transfer characteristic (G12) of the crosstalk reaching the right ear of the listener 5 from the speaker 200L based on the residual signal be (k) and DL output from the adding means 34. Then, the filter coefficient of the digital filter Ba is corrected according to this transfer characteristic. This correction is performed so that the residual signal be (k) input from the adding means 35 is as small as possible. Thereby, the coefficient determining means b can approximate the output by (k) to the reproduced audio data rp (k).

Here, the corrected filter coefficient of the digital filter Ba is notified to the signal correction unit 43 by the adaptive filter B. Thereby, the filter coefficient of the signal correction unit 43 is set to the filter coefficient identified by the adaptive filter B (digital filter Ba ) (filter coefficient correction setting function).

The filter characteristic calculation unit 63 is provided in connection with the reproduction processing unit 11 and the microphone 2R. The real head sound collection right signal DR input from the reproduction processing unit 11, the reproduction audio signal rp (k) from the microphone 2R, and Based on the above, the characteristics of the signal correction units 42 and 45 are identified and set.
Further, as shown in FIG. 17, the filter characteristic calculation unit 63 includes a delay unit 36, an addition unit 37, and an adaptive filter C.
The delay means 36 delays the input real head sound collection right signal (DR) by a preset time (t) and outputs the delayed audio data cd (kt) to the addition means 37. (Delayed data transmission function).

Further, the adding means 37 performs a process of adding the output cy (k) (minus component) input through the adaptive filter C and the delayed audio signal cd (kt) (plus component) (addition processing function). ).
Here, the adding means 34 performs a process of subtracting the output cy (k) from the delayed speech signal cd (kt ) to generate a residual signal ce (k) and send it to the adaptive filter C (residual signal). Output function).

The adaptive filter C includes a digital filter Ca and coefficient determination means Cb that determines the filter coefficient of the digital filter Ca.
The digital filter Ca performs a convolution process on the reproduced audio data rp (k) input from the microphone 2R with a preset filter coefficient to generate an output cy (k). The digital filter Ca outputs the generated output cy (k) to the adding means 37 (reproduction signal correction output function).

The coefficient determining means Cb reaches the right ear of the listener 5 from the speaker 200R based on the residual signal ce (k) output from the adding means 37 and the reproduced audio data rp (k) input from the microphone 2R. A reverse transfer characteristic (1 / G22: reverse characteristic) opposite to the transfer characteristic (G22) of the audio signal is calculated, and the filter coefficient of the digital filter Ca is corrected according to the reverse characteristic. This correction is performed so that the residual signal ce (k) input from the adding means 37 to the adaptive filter C becomes as small as possible.
Thereby, it is possible to perform an adaptive process for approximating the output cy (k) from the adaptive filter C to the delayed speech signal cd (kt).

The corrected filter coefficient of the digital filter Ca is notified to the signal correction units 42 and 45 by the adaptive filter C (identification coefficient notification function). Thereby, the filter coefficients (characteristics) of the signal correction units 42 and 45 are set in accordance with the filter coefficient of the digital filter Ca (filter coefficient correction setting function).

As shown in FIG. 17, the filter characteristic calculation unit 64 includes at least an adding unit 38 and an adaptive filter D.
The adder 38 receives the output dy (k) from the adaptive filter D and the reproduced audio data (left microphone input signal) lp (k) input from the microphone 2L.

The adding means 38 performs a process of adding the left microphone input signal (plus component) lp (k) and the output signal dy (k) (minus component) (addition processing function).
Here, the residual signal de (k) is generated by subtracting the output signal dy (k) from the left microphone input signal lp (k). The adding means 38 sends the residual signal de (k)
Is sent to the adaptive filter D (residual signal output function).

The adaptive filter D includes a digital filter Da and coefficient determining means Db that determines a filter coefficient of the digital filter Da .
The digital filter Da performs convolution processing on the input real head sound collection right signal (DR) with a preset filter coefficient to generate an output dy (k). The digital filter Da outputs the generated output dy (k) to the adding means 38 (recording signal correction output function).

The coefficient determining means Db is based on the residual signal de (k) output from the adding means 38 and the real head sound collection right signal DR, and the transfer characteristic of the crosstalk reaching the left ear of the listener 5 from the speaker 200R. (G21) is calculated, and the filter coefficient of the digital filter Da is corrected according to this transfer characteristic. This correction is executed so that the residual signal de (k) input from the adding means 38 becomes as small as possible. As a result, the coefficient determining means Db can approximate the output dy (k) to the reproduced audio data lp (k).

Here, the corrected filter coefficient of the digital filter Da is notified to the signal correction unit 44 by the adaptive filter D. Thereby, the filter coefficient (characteristic) of the signal correction unit 44 is set in accordance with the filter coefficient of the adaptive filter D (digital filter Da ) (filter coefficient correction setting function).

  As described above, the filter characteristic calculation units 61 to 64 of the reproduction apparatus 100 use the recorded sound from the reproduction processing unit 11 and the sound collected by the microphone 2 to reflect the characteristics of the signal correction units 41 to 46 during real head reproduction. Can be identified dynamically based on the generated speech.

Further, the reproduction apparatus 100 of the second embodiment stops the identification when binaural reproduction is performed and the identification of each of the filter characteristic calculation units 61 to 64 is stable, and G12, 1 / G22, G21 identified at this time point , 1 / G11 filter coefficients may be the characteristics of the corresponding signal correction units 41 to 46.

In the second embodiment, the microphone 2 used for real head recording is worn by the listener 5 in the same manner as when recording, and the real head recorded sound from the stereo loudspeaker is listened to and supplied to the loudspeaker. The crosstalk between the loudspeaker and both ears can be dynamically canceled using the signal that is reproduced from the loudspeaker and collected by the microphone.

  Thereby, even if it is reproduction | regeneration by a loudspeaker, the reproduction | regeneration environment equivalent to the time of reproduction | regeneration with headphones (Embodiment 1) can be constructed | assembled in a listener's ear canal area | region. In addition, it is possible to update the characteristics dynamically following changes in the stereo loudspeaker to be used and changes in the listening position of the listener.

In the second embodiment, the filter characteristic calculation units 61 and 62 (
16) and the right side including the filter characteristic calculators 63 and 64 (FIG. 17) are described separately. However, when the left and right channels of the reproduced audio signal collected by the real head have a low correlation. May be set so that the identification processing is performed simultaneously on the left and right sides.

In the second embodiment, the filter coefficient calculation processing in the filter characteristic calculation units 61 to 64 (FIG. 15) includes a general adaptive algorithm LMS method (Least Mean Square) and a learning identification method (Normalize Least Mean Square). ), It takes time for identification, and it becomes difficult to follow the change in the position of the listener.
Therefore, in the second embodiment, the high-speed H∞ filter (FHF: high-speed calculation filter) disclosed in Japanese Patent No. 4067269 is used for the calculation processing of the filter coefficient.

  Also, the filter coefficients of the adaptive filters A (1 / G11), B (G12), C (1 / G22), and D (G21) are calculated at regular intervals using this FHF, and the calculated filter coefficients are By using it as a filter coefficient of the signal correction units 41 to 46, even when each transfer function changes due to movement of the listener 5, crosstalk cancellation processing is performed following the positional variation between the listener 5 and the speaker. It becomes possible.

In the second embodiment, since the crosstalk cancellation process is executed following the movement of the listener, the signal recorded by the real head microphone attached to the listener's left ear is transmitted only from the left speaker to the left ear. The signal recorded by the real head microphone attached to the right ear reaches only the right ear from the right speaker. For this reason, even when the audio signal recorded by the real head is reproduced by a pair of stereo loudspeakers, the sound field of the recording site in the auditory canal region of the listener is the same as when reproducing by the headphones of the first embodiment. The feeling is faithfully reproduced without losing the spatial spread or orientation.

Further, in the second embodiment, it is possible to sequentially identify without using a special measurement signal, and the binaural audio signal recorded in the real head can be reproduced more faithfully in the listener's ear. Furthermore, the speaker can be used without limiting the speaker and listening position used for reproduction.

It is a schematic block diagram illustrating an embodiment 1 that put the binaural sound pickup reproducing system according to the present invention. FIG. 2A is a front view of the microphone 2R, and FIG. 2B is a side view of the microphone 2R in the binaural sound collection and reproduction system disclosed in FIG. It is explanatory drawing which shows the state with which the recording person 4 mounted | wore with the microphone 2R in the binaural sound collection reproduction | regeneration system disclosed in FIG. It is an explanatory diagram showing a recording device in the binaural sound pickup reproducing system disclosed in FIG. It is a schematic block diagram which shows an example of an internal structure of the reproducing | regenerating apparatus in the binaural sound collection reproduction | regeneration system disclosed in FIG. It is explanatory drawing which shows one structural example for identifying the transmission characteristic from the headphones 3R to the microphone 2R in a binaural sound collection reproduction system. FIG. 7A shows the transfer characteristic identified by the adaptive filter X in the configuration example disclosed in FIG. 6, FIG. 7A shows the time waveform of the transfer characteristic identified by the adaptive filter X, and FIG. 7B shows the transfer characteristic shown in FIG. It is explanatory drawing which shows the frequency characteristic of the time waveform of A). It is explanatory drawing which shows a part of structure content in the binaural sound collection reproduction | regeneration system disclosed in FIG. FIG. 8A shows the transfer characteristic identified by the adaptive filter Y in the explanatory diagram disclosed in FIG. 8, FIG. 8A shows the time waveform of the transfer characteristic identified by the adaptive filter Y, and FIG. 8B shows the transfer characteristic shown in FIG. It is explanatory drawing which shows the frequency characteristic of the time waveform of A). 10A is a time waveform when the time waveform of FIG. 7A and the time waveform of FIG. 9A are convoluted, and FIG. 10B is the frequency of the time waveform of FIG. 10A. It is explanatory drawing which shows a characteristic. It is a figure which shows Embodiment 2 in the binaural sound collection reproduction | regeneration system by this invention, Comprising: It is a schematic explanatory drawing of the crosstalk which reaches | attains a listener from a stereo speaker. It is a schematic block diagram which shows the structure for canceling the crosstalk disclosed in FIG. It is a schematic block diagram which shows Embodiment 2 of the binaural sound collection reproduction | regeneration system by this invention. It is a schematic block diagram which shows a part of internal structure of the reproducing | regenerating apparatus in Embodiment 2 disclosed in FIG. FIG. 14 is a schematic block diagram illustrating another part of the internal configuration of the playback apparatus according to the second embodiment disclosed in FIG. 13. It is a schematic block diagram which shows the internal structure of the filter characteristic calculation part in the schematic block diagram disclosed in FIG. It is a schematic block diagram which shows the other internal structure of the filter characteristic calculation part in the schematic block diagram disclosed in FIG.

1,100 Binaural playback device 2L, 2R microphone (microphone for sound collection)
3L, 3R headphones (headphone speakers)
4 Recorder 5 Listener 6 Recording device 7 Data recording unit 11 Reproduction processing unit 12L, 12R Correction filter 13L, 13R Delay element (delay signal output means)
14L, 14R Adaptive filter 15L, 15R Adder (residual detection means)
200L, 200R Loudspeaker 50 Crosstalk cancellation unit 31L, 32R Addition processing unit 33, 36 Delay means (delay signal output means)
34, 35, 37, 38 Addition means (residual detection means)
41, 42 Signal correction unit (reproduction audio correction filter)
43, 44, 45, 46 Signal correction filter 61, 63 Filter characteristic calculation unit (reproduced voice adaptive filter)
62, 64 filter characteristic calculation unit
A, B, C, D Adaptive filter

Claims (8)

A correction filter that performs a predetermined correction on a recorded audio signal recorded in advance by a microphone for collecting sound and supplies the correction to a speaker;
An adaptive filter that corrects a reproduced audio signal reproduced by the speaker and collected in advance by the sound collecting microphone provided in the auditory canal region of the listener;
A delayed signal output means for acquiring a recorded audio signal after correction output from the correction filter and outputting the signal after a predetermined time delay;
Addition having a residual data output function for calculating a difference between the reproduced audio signal acquired via the adaptive filter and the recorded audio signal output from the delayed signal output means and sending the difference to the adaptive filter Equipped with
The adaptive filter, an inverse characteristic identification function of calculating a transfer characteristic opposite characteristics from the speaker to the sound collecting microphone based on the reproduced audio signal comprising picked up by the sound pickup microphone and said difference And the correction filter has a function of correcting the recorded audio signal based on the characteristic calculated by the adaptive filter,
Furthermore, the adaptive filter is provided with a filter characteristic identification function for identifying a characteristic in the adaptive filter so that the difference is minimized based on the difference sent from the adder. Playback system. The binaural sound collecting and reproducing system according to claim 1,
The speaker is provided with two channels for the left and right ears in accordance with the listener's ears, and the sound collecting microphone, correction filter, adaptive filter, delay signal output means, corresponding to each speaker of the two channels , And two adders ,
A binaural sound collecting / reproducing system, wherein the speaker is a headphone speaker to be attached to the auditory canal region of the listener. The binaural sound collecting and reproducing system according to claim 1,
The speaker is provided with two channels for the left and right ears in accordance with the listener's ears, and the sound collecting microphone, correction filter, adaptive filter, delay signal output means, corresponding to each speaker of the two channels , And two adders ,
A binaural sound collecting / reproducing system, wherein the speaker is constituted by a left loudspeaker and a right loudspeaker for the listener to listen through a space. A two-channel loudspeaker set to the left and right in advance corresponding to the left and right ears of the listener, and a correction filter for individually performing a predetermined correction for each loudspeaker on the recorded audio signal supplied to each loudspeaker As well as
For the recorded audio signal supplied to each of the left and right loudspeakers via the correction filter, the voice reaching the right ear of the listener from the left loudspeaker and the left ear of the listener from the right loudspeaker It has a crosstalk cancellation part that performs correction to cancel the sound that reaches
The crosstalk cancellation unit
Corresponding to each of the left and right loudspeakers, correction is performed on the reproduced audio signal that is reproduced by the speaker and collected in advance by the sound collecting microphone provided in the auditory canal region of the listener. An adaptive filter, delayed signal output means for acquiring a recorded audio signal after correction output from the correction filter and outputting it after a predetermined time delay, a reproduced audio signal acquired via the adaptive filter, and the An adder having a residual data output function for calculating a difference with the recorded audio signal output from the delayed signal output means and sending the difference to the adaptive filter;
Inverse characteristic identification function for each adaptive filter to calculate a characteristic opposite to the transmission characteristic from the speaker to the sound collecting microphone based on the reproduced sound signal picked up by the sound collecting microphone and the difference. The correction filter has a function of correcting the recorded audio signal based on the characteristics calculated by the adaptive filter,
Furthermore, each adaptive filter has a filter characteristic identification function for identifying a characteristic in the adaptive filter so that the difference is minimized based on the difference sent from the adder. Sound playback system. A two-channel loudspeaker set to the left and right in advance corresponding to the left and right ears of the listener, and a correction filter for individually performing a predetermined correction for each loudspeaker on the recorded audio signal supplied to each loudspeaker As well as
For the recorded audio signal supplied to each of the left and right loudspeakers via the correction filter, the voice reaching the right ear of the listener from the left loudspeaker and the left ear of the listener from the right loudspeaker It has a crosstalk cancellation part that performs correction to cancel the sound that reaches
The crosstalk cancellation unit
For the recorded audio signal, transmission characteristics opposite to the transmission characteristics between the left and right speakers and the left and right ears of the listener, and the left and right speakers and the right and left ears of the listener. A reproduction audio correction filter for performing correction to give a transfer characteristic between them,
Further, in the reproduction sound correction filter based on the reproduction sound signal reproduced by each loudspeaker and collected in advance by each sound pickup microphone provided in the left and right ear canal regions of the listener and the recorded sound signal A binaural sound collecting / reproducing system comprising: a reproduction sound adaptive filter that calculates the transfer characteristic and sets the transfer characteristic in the reproduction sound correction filter. In the binaural sound collecting and reproducing system according to any one of claims 1 to 4,
A binaural sound collecting / reproducing system using a high-speed calculation filter that calculates characteristics of the adaptive filter at high speed as the adaptive filter. In the binaural sound collecting and reproducing system according to claim 5,
A binaural sound collecting and reproducing system using a high-speed calculation filter that calculates each transfer characteristic in the reproduction audio adaptive filter at high speed as the reproduction audio adaptive filter. The binaural sound collecting and reproducing system according to any one of claims 1 to 7,
The binaural sound collecting / reproducing system, wherein the sound collecting microphone has a probe shape that can be inserted into the ear canal without blocking the ear canal.

Images