JP2012213114A - Acoustic signal processing apparatus and acoustic signal processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To adjust an amplification characteristic more appropriately.SOLUTION: An acoustic signal processing apparatus 100 includes a noise level analysis section 90 for analyzing an ambient noise level, an audiometric section 40 for outputting a variable volume sound and measuring a user's auditory characteristic according to the amount of variation of the volume, and a signal processing section 60a for generating, in accordance with the noise level and the auditory characteristic data measured by the audiometric section 40, an output signal characteristic that, in an intermediate sound pressure level region in an input signal range from a minimum sound pressure level to a maximum sound pressure level, excluding a predetermined range of a low sound pressure level region in the minimum sound pressure level and a predetermined range of a high sound pressure level region in the maximum sound pressure level, increases gradually to a first sound pressure level in the intermediate sound pressure level region from the low sound pressure level region side and then decreases gradually from the first sound pressure level toward the high sound pressure level region side, and performing signal processing according to the output signal characteristic.

Description

  The present invention relates to an acoustic signal processing apparatus and an acoustic signal processing method for performing signal processing according to a hearing characteristic of a user.

  In normal headphone or earphone playback, the sound image is localized in the head, so out-of-head localization headphones and out-of-head localization signal processing for localizing the sound image outside the head are already known. As shown in Patent Document 1, for example, this technique has a plurality of head-related transfer functions in the direction to be localized in advance, convolves with the audio signal to be localized, and further convolves the inverse characteristics of the transfer function between the headphones and the eardrum. Can be realized. If the audio signal is binaural audio, that is, the audio signal recorded by setting the microphone between the eardrum position of both ears or the eardrum position near the ear canal entrance, convolve the inverse characteristics of the transfer function between the headphones and the microphone. To achieve out-of-head localization.

  However, the head-related transfer function, the transfer function between the headphone and the eardrum, or the transfer function between the headphone and the microphone varies depending on the shape of the individual's head and auricle, and the localization effect is low in the transfer function of others. In addition, hearing ability declines as you get older, so the correct signal is not transmitted and you can't hear any sound.

  In view of this, the present applicant has proposed a technique for obtaining a reproduced sound that is audibly perceptible according to the individual hearing characteristics (see Patent Document 2). In this prior art, the audio signal is divided into a plurality of frequency bands, and the sound pressure level of each frequency band is set to a low sound pressure level region in a predetermined range from the minimum sound pressure level and a high sound pressure in a predetermined range from the maximum sound pressure level. Three areas are provided: a level area and an intermediate sound pressure level area therebetween. Then, the sound pressure level is gradually increased from the low sound pressure level region side to the intermediate point in the intermediate sound pressure level region, and is amplified with an amplification characteristic that sequentially decreases the sound pressure level from the intermediate point to the high sound pressure level region side, The amplified audio signals of a plurality of frequency bands are added.

Japanese Patent No. 2924539 JP 2009-077378 A

  According to the technique disclosed in Patent Document 2, it is possible to provide a sound that takes into account the individual's hearing characteristics, but it is possible to obtain a reproduction sound that is more suitable for the individual's hearing characteristics and is easy to listen to and has a natural orientation. It is hoped that. That is, as aging progresses, the importance of out-of-head localization headphones that take into account individual hearing characteristics and hearing aids with a natural sense of localization are increasing.

  The present invention has been made to solve such a problem, and an object thereof is to provide an acoustic signal processing device and an acoustic signal processing method capable of better adjusting the amplification characteristics.

  In order to solve the above-mentioned problem, an acoustic signal processing device (100) according to an embodiment of the present invention can change a sound level of a sound level and a noise level analysis unit (90) for analyzing a surrounding noise level, The minimum sound of the input signal according to the hearing measurement unit (40) that measures the hearing characteristic of the user based on the amount of change in volume, and the hearing characteristic data measured by the noise level and the hearing measurement unit (40) An intermediate sound pressure level region excluding a low sound pressure level region in a predetermined range from the minimum sound pressure level and a high sound pressure level region in a predetermined range from the maximum sound pressure level in a range from a pressure level to a maximum sound pressure level, An output signal characteristic that is sequentially increased from the low sound pressure level region side to the first sound pressure level in the intermediate sound pressure level region and gradually decreased from the first sound pressure level to the high sound pressure level region side. It generates, characterized in that it comprises a signal processing unit for performing signal processing (60a) in response to the output signal characteristics.

  In the acoustic signal processing device (100), the noise level analysis unit (90) performs pitch extraction based on the ambient sound, and when the pitch is extracted, the sound pressure level of the ambient sound is determined from the conversation sound. When the sound pressure level is calculated and the pitch is not extracted, the sound pressure level of the ambient sound is calculated as the sound pressure level of the noise, and the sound pressure level of the conversation sound calculated by the predetermined number of pitch extractions The first sound pressure level may be set based on an average value of sound pressure levels of the noise.

  In the acoustic signal processing device (100), the hearing measurement unit (40) measures an optimum listening level (Best) of a user, and the signal processing unit (60a) determines the intermediate sound based on the optimum listening level. A change point (P2) between the pressure level region and the high sound pressure level region may be set.

  In order to solve the above-mentioned problem, an acoustic signal processing method according to an embodiment of the present invention can vary a noise level analysis step for analyzing a surrounding noise level, a volume of a sound to be output, and based on a change amount of the volume. And a hearing measurement step for measuring a hearing characteristic of the user, and the sound signal in the range from the minimum sound pressure level to the maximum sound pressure level of the input signal according to the noise level and the hearing characteristic data measured in the hearing measurement step. An intermediate sound pressure level region excluding a low sound pressure level region in a predetermined range from a minimum sound pressure level and a high sound pressure level region in a predetermined range from the maximum sound pressure level, and the intermediate sound pressure level from the low sound pressure level region side Generating an output signal characteristic that sequentially increases to a first sound pressure level in a region and decreases gradually from the first sound pressure level to the high sound pressure level region side, Characterized in that it comprises a signal processing step of performing signal processing in accordance with the No. characteristics.

  ADVANTAGE OF THE INVENTION According to this invention, it is possible to provide the acoustic signal processing apparatus and acoustic signal processing method which can adjust an amplification characteristic better.

It is a block diagram which shows the acoustic signal processing apparatus in 1st Embodiment. It is a general-view figure of the earphone and microphone in a 1st embodiment. It is a flowchart which shows operation | movement of the audiometry part in 1st Embodiment. It is a general-view figure of the operation part in a 1st embodiment. It is a block diagram which shows the detail of the reproduction | regeneration signal processing part in 1st Embodiment. It is a figure which shows the characteristic of the gain curve in 1st Embodiment. It is a figure which shows an example of the minimum audible level which is the minimum audible level of a hearing characteristic data measured by the audiometry part in 1st Embodiment, and a reference. It is a figure which shows the characteristic of the gain curve in 1st Embodiment. It is a figure which shows the characteristic of the gain curve in 2nd Embodiment. It is a figure which shows the parameter of the gain curve in 2nd Embodiment. It is a block diagram which shows the acoustic signal processing apparatus in 3rd Embodiment. It is a block diagram which shows the detail of the reproduction | regeneration signal processing part in 3rd Embodiment. It is a flowchart which shows operation | movement of the noise level analysis part in 3rd Embodiment. It is a figure which shows the characteristic of the gain curve in other (or 4th) embodiment.

  Embodiments of an acoustic signal processing apparatus and method according to the present invention will be described below with reference to the accompanying drawings. Although each embodiment is described using a block diagram of a hardware configuration, a part of each embodiment may be configured as software. All parts that can be configured as software may be configured by software, and the use of hardware and software may be arbitrarily selected.

<First Embodiment>
FIG. 1 is a block diagram illustrating an acoustic signal processing apparatus 100 according to the first embodiment. The acoustic signal processing device 100 is an out-of-head localization headphone or the like that performs signal processing according to a user's hearing characteristics, and includes a microphone 10, an earphone 20, an impulse response measurement unit 30, an audio measurement unit 40, and an operation unit. 50, a reproduction signal processing unit 60, a data input / output unit 70, and a memory 80. As shown in FIG. 2, the microphone 10 and the earphone 20 have two left and right channels and are integrated. The impulse response measuring unit 30 outputs a pulse from the output terminal 1 and takes in a signal received by the microphone 10 via the switch 2a. The hearing measurement unit 40 outputs a measurement sound and measures the hearing characteristics of the user. The user operates the buttons of the operation unit 50 by determining whether or not the measurement sound emitted from the earphone 20 can be heard via the switch 3a. The operation unit 50 can vary the volume of the measurement sound to be heard and determine the level at that time. These operation contents are notified to the hearing measurement unit 40. The data input / output unit 70 receives the results obtained by the impulse response measurement unit 30 and the hearing measurement unit 40 and stores them in the memory 80 or passes the data to the reproduction signal processing unit 60. The reproduction signal processing unit 60 receives an audio signal from the microphone 10 via the switches 2b and 4a, and receives an audio signal from a DVD player, a TV tuner, or the like via the switch 4b. The reproduction signal processing unit 60 performs signal processing based on the results obtained by the hearing measurement or the impulse response measurement, and outputs an acoustic signal. As a result, a natural localized sound that is easy for the user to listen to is reproduced from the earphone via the switch 3b.

Hereinafter, the configuration of the acoustic signal processing apparatus 100 will be described in more detail. The impulse response measurement unit 30 measures a user's head-related transfer function. In order to audition a multi-channel audio disc such as a Blu-ray disc, it is necessary to measure the head-related transfer function between the seven speaker arrangements (L, R, C, SL, SR, SBL, SBR) and the audition position. . In this measurement, an amplifier and a speaker (not shown) are installed in a desired place in the room, and the output terminal 1 is connected to the amplifier. The user wears the earphone 20 firmly on both ears at the listening position. By doing so, the microphone 10 is set in the auricle. When the measurement is started, an impulse response is output from the impulse response measurement unit 30, and an impulse sound is output from an amplifier and a speaker (not shown). An impulse is a signal having a maximum amplitude at time t = t1, and a zero amplitude at t ≠ t1. The impulse response measurement unit 30 stores the left and right two-channel signals received by the microphone 10. The length of the stored signal may be changed depending on the size of the room and the reverberation time. Further, synchronous addition may be performed in order to improve the S / N ratio. The measurement method is not limited to this method, and a TSP method, a cross spectrum method, or the like can also be adopted. In this way, the left and right head related transfer functions H _Ll , H _Lr , H _Rl , H _Rr , H _Cl , H _Cr , H _SLl , H _SLr , H _SRl , H _SRr , H _SBLl , H in each speaker arrangement. _SBLr , H _SBRl , and H _SBRr are measured. Head-related transfer function data obtained by the measurement is stored in the memory 80 by the data input / output unit 70.

  In order to obtain a natural feeling of localization, it is necessary to reproduce faithfully at a high frequency where the individual difference is the highest. Therefore, it is only necessary to faithfully reproduce the previously measured head-related transfer function. However, as a person grows older, hearing at a high frequency generally decreases. No matter how faithfully the hardware can be reproduced, if you can't hear the sound, the sense of orientation is lost.

  Therefore, the hearing measurement unit 40 performs the hearing measurement in a high frequency region with a large individual difference, here 1 kHz to 8 kHz. The hearing measurement unit 40 is not shown so that a signal of 0 dBFS (FS means full scale) is output from the earphone 20 at a sound pressure level of 100 dB (hereinafter, the sound pressure level is displayed as SPL after dB) during measurement. The volume of the amplifier is set. The value of 100 dBSPL is an example, and may be further increased according to the degree of hearing loss.

  FIG. 3 is a flowchart showing the operation of the hearing measurement unit 40. FIG. 4 is an overview diagram of the operation unit 50. Hereinafter, the operation during the hearing measurement will be described in detail with reference to these drawings.

  First, the hearing measurement unit 40 outputs a pure tone of a predetermined frequency from the earphone 20 at a volume (−XdBFS, here X = 30) smaller than full scale (0 dBFS) in step S1. For example, a 1 kHz pure sound is output first. The user changes the sound volume by pressing the up / down button 51 of the operation unit 50 until the sound volume output from the earphone 20 becomes a sound volume that is neither large nor small. The hearing measurement unit 40 determines whether or not the up / down button 51 is pressed based on a signal sent from the operation unit 50. (Step S3: Yes, proceed to Step S4).

  The user presses the OK button 52 of the operation unit 50 when the volume is just right by the operation of the up / down button 51. When the hearing measurement unit 40 acquires a signal indicating that the OK button 52 has been pressed from the operation unit 50 (step S2: YES), the sound pressure level at the time of signal acquisition is set as the user's optimum listening level Best (the sound pressure level ( The value is temporarily stored in an internal memory (not shown) with the value of dBSPL) (step S5). The optimum listening level of the user when outputting a pure sound of 1 kHz is Best1k.

  Subsequently, the hearing measurement unit 40 outputs a pure sound of 2 kHz in step S1, acquires and stores the optimum listening level Best2k of the user through steps S2 to S4. Similarly, the user's optimum listening level Best4k at 4 kHz and the user's optimum listening level Best8k at 8 kHz are acquired.

  Subsequently, the hearing measurement unit 40 obtains the minimum audible level of the user. The operation flow is the same as the operation for obtaining the optimum listening level Best described above. First, in step S1, a pure tone of 1 kHz is output. The user changes the volume by pressing the up / down button 51 of the operation unit 50 until the volume reaches a barely audible level. The hearing measurement unit 40 determines whether or not the up / down button 51 has been pressed based on a signal sent from the operation unit 50 (step S3: Yes, proceed to step S4). Then, the user presses the OK button 52 of the operation unit 50 when the sound level is reached. When the hearing measurement unit 40 acquires a signal indicating that the OK button 52 has been pressed from the operation unit 50 (step S2: YES), the hearing measurement unit 40 sets the sound pressure level at this time to the minimum audible level Min of the user. The sound pressure level (dBSPL) value is temporarily stored in an internal memory (not shown). The minimum audible level of the user when a 1 kHz pure tone is output is Min1k.

  Similarly, the hearing measurement unit 40 outputs a 2 kHz pure sound in step S1, acquires the minimum audible level Min2k that is the sound pressure level that the user can hear through step S2 to step S4, and temporarily stores it. In this manner, the minimum audible level Min4k and Min8k of the user is obtained for 4 kHz and 8 kHz.

  This completes the measurement. The Best1k, Best2k, Best4k, Best8k, Min1k, Min2k, Min4k, and Min8k acquired by the hearing acquisition unit 40 (hereinafter, these eight data are collectively referred to as “auditory characteristic data”) are sent to the data input / output unit 70. Stored in the memory 80.

FIG. 5 is a block diagram showing details of the reproduction signal processing unit 60. The reproduction signal processing unit 60 generates an acoustic signal that is easy to listen to according to the hearing characteristic of the user using the hearing characteristic data. That is, an audio output signal from a BD player or the like is input from the input terminal 4 to the reproduction signal processing unit 60 via the switch 4b. At this time, if the audio signal is a stream such as Dolby Digital, it is demodulated by a decoder (not shown) and separated into a 7.1-channel signal. The signals of the seven channels excluding the WF channel are input to the convolution unit 61. On the other hand, the data input / output unit 70 receives the head-related transfer function data H _Ll , H _Lr , H _Rl , H _Rr , H _Cl , H _Cr , H _SLl , H _SLr , H _SRl , H _SRr , H _SBLl , H _SBLr , H _SBRl , and H _SBRr are read and passed to the convolution unit 61. The convolution unit 61 convolves the head transfer function of both ears with each signal such as L signal and H _Ll , H _Lr , R signal and H _Rl , H _Rr ,. Is output to the adder 62. The adder 62 adds the left-ear channels and the right-ear channels to a 2-channel signal. The demodulated WF channel signal is amplified by the amplifier 63, branched, and added to the left and right channel signals by the adder 64a. Next, the band dividing unit 65 divides the left and right two channels into five bands, a low band of 750 Hz or less, an octave band having a center frequency of 1 kHz, 2 kHz, 4 kHz, and a high band of 8 kHz or more. The sound pressure level amplifying unit 66 amplifies the sound pressure level based on a gain curve obtained from the user's hearing characteristic data. Further, the adder 67 adds the signals of the five bands and outputs the signals as two left and right channel signals again.

  The sound pressure level amplification unit 66 receives the hearing characteristic data stored in the memory 80 from the data input / output unit 70, and performs dynamic range compression characterized by a gain curve shown in FIG. In this dynamic range compression, for the sound pressure level in each frequency band, a low sound pressure level region R1 in a predetermined range from a minimum sound pressure level to a sound pressure level (first sound pressure level) greater than the minimum sound pressure level, A high sound pressure level region R2 in a predetermined range from a maximum sound pressure level to a sound pressure level smaller than the maximum sound pressure level (second sound pressure level), in the meantime (from the first sound pressure level to the second sound pressure level) 3) of the intermediate sound pressure level region R3.

  The gain curve in FIG. 6 is preferably a non-amplified region where the input signal is not amplified in the low sound pressure level region R1 and the high sound pressure level region R2. By making the low sound pressure level region R1 and the high sound pressure level region R2 non-amplified regions, the musicality of the music is hardly impaired. In the intermediate sound pressure level region R3 between the low sound pressure level region R1 and the high sound pressure level region R2, the non-amplified state indicated by the broken line is an amplified region that is amplified by the characteristic indicated by the solid line. As shown in FIG. 6, in the intermediate sound pressure level region R3 that is an amplification region, an input is performed from a point P1 on the low sound pressure level region R1 side to an inflection point Pi that is an intermediate point in the intermediate sound pressure level region R3. A characteristic of increasing the amplification level sequentially as the sound pressure level of the signal increases, and decreasing the amplification level sequentially as the sound pressure level of the input signal increases from the inflection point Pi to the point P2 on the high sound pressure level region R2 side. It has become. FIG. 6 shows an example in which the amplification degree is linearly increased from the point P1 to the inflection point Pi and the amplification degree is linearly reduced from the inflection point Pi to the point P2.

  Here, in this embodiment, the gain curve parameters are set according to the user's hearing characteristic data. Specifically, the reproduction signal processing unit 60 sets the difference between the minimum audible level Min of the audio characteristic data measured by the audio measurement unit 40 and the minimum audible level Ref as a reference to the maximum gain Ga at the point Pi. For each frequency, the boundary between the high sound pressure level region R2 and the intermediate sound pressure level region R3 is set to the user's optimum listening level Best.

  FIG. 7 is a diagram illustrating an example of the optimum listening level Best, the minimum audible level Min, and the minimum audible level Ref as a reference of the hearing characteristic data measured by the hearing measurement unit 40. Here, the minimum audible level Ref exemplifies the minimum audible level Ref of a normal hearing person in the 40s.

  When the minimum audible level Min1k of the user at 1 kHz is 40 dBSPL, the sound of 20 dBSPL (minimum audible level Ref1k) that should be heard by a normal hearing person cannot be heard unless it is increased by +20 dB. Therefore, a gain curve that outputs a signal obtained by amplifying a −80 dBFS input signal by +20 dB shown in FIG. 8A is created. Here, since this embodiment is a system of 100 dBSPL = 0 dBFS, 20 dBSPL is −80 dBFS. When amplified by 20 dB, the output signal becomes −60 dBFS (40 dBSPL = −60 dBFS).

  That is, the inflection point Pi is a point where the input signal is −80 dBFS and the maximum gain Ga = 20 dB. From FIG. 7, since the optimum listening level Best1k at 1 kHz is 70 dBSPL, when a sound of 70 dBSPL or higher is input, it is not amplified, that is, it is amplified by 0 dB. Therefore, the point P2 which is the boundary between the intermediate sound pressure level region R3 and the high sound pressure level region R2 in the intermediate sound pressure level region R3 is −30 dBFS. In this way, the gain curve in the 1 kHz band is determined as shown in FIG. Similarly for the other bands, the gain curve of the 2 kHz band is determined as shown in FIG. 8B, the gain curve of the 4 kHz band is determined as shown in FIG. 8C, and the gain curve of the 8 kHz band is shown in FIG. It is determined as in d). In the 8 kHz band, as shown in FIG. 7, since the user's minimum audible level Min8k is the same as the normal hearing person's minimum audible level Ref8k, the maximum gain Ga = 0 dB is amplified, and the gain curve is a straight line. Since an elderly person tends to cause a supplementary phenomenon at a high frequency, the optimal listening level Best8k may be lower than other bands as in this example. This time, a point P1 (hereinafter simply referred to as “P1 point” or “P1”) at the end of the intermediate sound pressure level region R3 on the low sound pressure level region R1 side is set to a minimum value of −90 dBFS of 16-bit quantization. The point P2 (hereinafter simply referred to as “P2 point” or “P2”) at the end of the high sound pressure level region R2 side is determined by the optimum listening level Best obtained by the hearing measurement, but “just right volume” is determined. When it is difficult for the user to perform, for example, the sound pressure level of each frequency that feels as large as 1 kHz and 70 dBSPL is obtained and set to P2, so that it is possible to deliver sound that better matches the hearing characteristics.

  As described above, according to the present embodiment, the gain curve parameters are set according to the minimum audible level Min and the optimum listening level Best of the user, so that the amplification characteristics can be adjusted better. The sound signal generated in this way is faithfully reproduced in consideration of not only the user's head and auricle but also the user's hearing, when the multi-channel speaker is reproduced in the room. It is easy to enjoy with natural orientation. Accordingly, the user is less likely to get tired of listening, and thus can enjoy listening for a long time.

  In this embodiment, the hearing measurement is performed by binaural listening, but may be performed separately on the left and right. In that case, dynamic range compression using different gain curves on the left and right is performed.

Second Embodiment
In the present embodiment, a case where the acoustic signal processing device 100 is used as a sound collector will be described. The basic configuration is the same as in FIG. The difference from the first embodiment is that (1) the contents of the hearing measurement, (2) the sound collected by the microphone 10 is input to the reproduction signal processing unit 60 and amplified to a finer sound pressure level for each band. (3) Characteristic correction of the earphone 20 is performed so that the sound image is localized out of the head and a natural sound field is obtained. Hereinafter, these three points will be described.

  When the measurement is started, the hearing measurement unit 40 outputs a guide voice for explaining the operation from the earphone 20 at about 70 dBSPL. The user changes the volume by pressing the up / down button 51 of the operation unit 50 until the volume is clearly audible. The hearing measurement unit 40 sends the change amount of the volume level to the data input / output unit 70. Next, measurement of the minimum audible level is started from the left ear. The level of pure sounds of 250 Hz, 500 Hz, 1 kHz, 2 kHz, 4 kHz, and 8 kHz may be varied randomly, or the levels may be lowered in order. In order to shorten the measurement time, the minimum audible level of a certain group may be placed on a normal distribution and presented by the binary tree method.

  When the user hears a sound, the user presses the OK button 52 of the operation unit 50. The hearing measurement unit 40 sends the minimum level at which the OK button 52 is pressed to the data input / output unit 70 as minimum audible level data. Similarly, the hearing ability measuring unit 40 measures the minimum audible level of the right ear, and acquires the minimum audible level data of the right ear. When the minimum audible level is measured, the maximum allowable level is measured. The user presses the up / down button 52 for each frequency and presses the OK button 52 at a level that feels large. This is also measured on the left and right. The hearing measurement unit 40 sends the maximum level at which the OK button 52 is pressed to the data input / output unit 70 as maximum allowable level data. The minimum audible level data and the maximum allowable level are stored in the memory 80. Here, the variable level of the guide voice is Lg, the minimum audible level is Min (i) dBFS (where i is the band number 0 (250 Hz) to 6 (8 kHz)), and the maximum allowable level is Max (i) dBFS (where i is a band number 0 to 6).

As the input / output characteristics of the dynamic range compression in the sound pressure level amplifier 66, as shown in FIG. 9, a gain curve is used so that the amplification in the high sound pressure level region R2 is suppressed to the sound pressure level P3 or less. The parameters in this case are set as shown in FIG. 10 based on the hearing measurement result. In FIG. 10, Refmin (i) is the minimum audible level of the normal hearing person, Ref 70 phon is a conversion of the 70 phon isosensitive curve data of the normal hearing person to the FS scale (in this system, 0 dBFS = 100 dBSPL by calibration), and i is the band number. is there. The size of 70 phon is just right for a normal hearing person, neither big nor small. The volume of the guide voice is adjusted to a level that is just right for the user. The gain curve shown in FIG. 9 is a curve having exponential characteristics in the high sound pressure level region R2 of P2 or higher. For input signals of P2 or higher, the sound pressure level of the output signal (hereinafter referred to as output sound pressure level) is compressed from P2 to the sound pressure level (hereinafter referred to as input sound pressure level) (hereinafter referred to as input sound pressure level) of P2 or higher. Without being performed, but is provided at the same level as the input sound pressure level, but when the input sound pressure level exceeds a predetermined input sound pressure level, the output sound pressure level approaches the P3 as much as possible as the input sound pressure level increases ( Amplification characteristics along y = A ^x where A is an asymptotic line and A is a positive number). That is, since the user sets the sound pressure level P3 to the maximum allowable level, the characteristic is set such that no further amplification is performed. As a simpler amplification characteristic, the input sound pressure levels P2 to P3 are provided without being compressed, and the input sound pressure level equal to or higher than P3 is a straight line that makes the output sound pressure level constant (P3). General amplification characteristics may be employed.

  In the first embodiment, the intermediate point Pi is set to the minimum audible level Ref of the normal hearing person. However, when the input sound is the microphone sound collection, it becomes very noisy if the sound of the minimum audible level is amplified. Therefore, in this embodiment, the initial value of Pi is for a quiet conversation sound (50 dBSPL), and is set to −50 dBFS.

  After such input / output characteristics are determined, the data input / output unit 70 reads out the inverse characteristics of the frequency characteristics of the earphone 20 stored in the memory 80 in advance, and the convolution unit 61 performs convolution processing. By this process, the characteristic unique to the earphone is canceled, so that the collected sound is corrected in consideration of the hearing characteristic of the user, and a faithful audio signal is delivered to the ear. As a result, it is easy to hear and a natural orientation is felt.

  Note that the data input / output unit 70 may be provided with a USB terminal or the like so that the data measured by the acoustic signal processing apparatus 100 can be used in another acoustic device. In this way, even a sound device such as a microphone 10 or a television or an audio amplifier without a measurement function can always provide an acoustic signal that matches the hearing characteristics of the user as long as it has playback signal processing.

<Third Embodiment>
By the way, according to a conventional acoustic signal processing apparatus, when a microphone recording sound such as a sound collector is targeted, noise is emphasized, and conversely, it is difficult to hear. Therefore, in the present embodiment, the following method is adopted to solve this problem.

  FIG. 11 is a block diagram illustrating an acoustic signal processing device 100 according to the third embodiment. The acoustic signal processing device 100 is a sound collector or the like that performs signal processing according to the hearing characteristics of the user, and includes a microphone 10, an earphone 20, an audio measurement unit 40, an operation unit 50, and a reproduction signal processing unit 60a. A data input / output unit 70, a memory 80, and a noise level analysis unit 90. The microphone 10 collects the current sound and branches it, sends one to the noise level analysis unit 90, and sends the other to the reproduction signal processing unit 60a. The noise level analysis unit 90 analyzes the noise level of so-called background noise (a sound other than the sound to be heard) and the noise level of the sound to be heard (here, the conversational sound of the other party). The analyzed result is sent to the data input / output unit 70 as noise level data and stored in the memory 80. The reproduction signal processing unit 60 a performs signal processing according to the noise level data stored in the memory 80. Other processing units are basically the same as those in the first or second embodiment.

  Hereinafter, the configuration of the acoustic signal processing apparatus 100 in the present embodiment will be described in more detail. First, calibration of the microphone 10 and the earphone 20 is necessary. That is, for the microphone 10, the microphone amplifier is adjusted so that the sound pressure level 100 dBSPL is 0 dBFS. Further, the earphone output sound pressure level when the 0 dBFS signal is reproduced is measured in advance. For example, a pure tone from 250 Hz to 8 kHz is reproduced and measured every octave. If this value is known, the minimum audible level of the user can be known from the sound pressure level. This operation may be performed at the time of factory shipment.

  The user measures his / her hearing ability with the acoustic signal processing apparatus 100. When the measurement is started, the data input / output unit 70 reads guide data explaining the operation method from the memory 80 and sends the guide data to the hearing measurement unit 40. This guide data is at a level of about −30 dBFS, and when it is listened to by the earphone 20, it is reproduced at 70 dBSPL. Although it is a normal level that is neither loud nor small for a normal hearing person, a hearing-impaired person may feel that it is a little small. The user can change the sound level of the guide data output from the earphone 20 by pressing the up / down button 51 of the operation unit 50. When the volume is easy to hear, the user presses the OK button 52 of the operation unit 50, and the hearing measurement unit 40 starts measurement when the OK button 52 is pressed. The level variation (from 70 dBSPL) of the guide data at this time is sent from the hearing measurement unit 40 to the data input / output unit 70. The hearing measurement unit 40 presents sound at various levels from 250 Hz to 8 kHz for each octave, and the user presses the OK button 52 of the operation unit 50 when listening. If it cannot be heard, the OK button 52 is not pressed. Of the levels presented randomly by the hearing measurement unit 40, the minimum value of the level at which the OK button 52 is pressed is the minimum audible level of the user. When the left ear measurement is complete, continue with the right ear measurement. Here, in order not to unnecessarily lengthen the measurement time, the hearing measurement unit 40 determines the level to be presented by the binary tree method using past data. That is, the result of measuring the minimum audible level in a certain group is placed on a normal distribution, and if the intermediate value is presented as the first presentation and heard, the intermediate value with the lower level is designated as the second presentation. On the other hand, when the user cannot listen to the first presented value, the intermediate value having a higher level than the first presented value is set as the second presentation. Then, the minimum audible level of one frequency can be determined by presenting the sound 3 to 4 times for an early person and 7 to 8 times for the person at the end of the population. When the minimum audible level is determined, the hearing measurement unit 40 sends measurement data to the data input / output unit 70.

  FIG. 12 is a block diagram showing details of the reproduction signal processing unit 60a in the third embodiment. The band dividing unit 65 divides the audio signal collected by the microphone 10 into bands. The sound pressure level amplification unit 66 performs dynamic range compression for each band. Dynamic range compression is adapted to the aging of human hearing characteristics. Amplification is gradually reduced because small sounds are amplified to a level where they can be heard, and loud sounds are supplemented.

  In the first embodiment, since the minimum audible level is the barely audible volume, it is considered that it is sufficient to amplify the amplitude by the difference with respect to the minimum audible level of the normal hearing person, and the minimum audible level Ref of the normal hearing person and the minimum of the user The difference from the audible level Min was taken as the maximum gain Ga. However, when actually operated, the air conditioning noise and background noise, which were small, become very large, and the voice of the other party who wanted to hear the most and the background noise have the same sound pressure level. It was. Therefore, in order to amplify the vicinity of the sound pressure level of the other party's conversation sound as much as possible without amplifying the background noise, a setting method unique to the microphone sound different from the sound signal such as CD or broadcast is necessary.

  Therefore, in the present embodiment, in order to determine the intermediate points Pi and P1, the noise level analysis unit 90 analyzes the signal input from the microphone 10, and the sound pressure level of background noise and the sound of conversational sound for each frequency band. Find the pressure level. The noise level analysis unit 90 can determine whether or not the noise is background noise based on whether or not the pitch has been extracted. That is, if the pitch can be extracted, the signal section is a conversational sound, and if the pitch cannot be extracted, the section is only background noise.

  FIG. 13 is a flowchart showing the operation of the noise level analysis unit 90. The noise level analysis unit 90 performs frame processing. For example, one frame has 320 samples with respect to a sampling frequency of 32 kHz. This value is determined because the amplitude fluctuation width of the conversational voice is 10 milliseconds or less, but may be determined by looking at the amplitude fluctuation width for each band or may be determined from the characteristics of background noise. In addition, the noise level analysis unit 90 of this embodiment determines Pi by determining the average sound pressure level every five frames, but the number of frames for determining the average is not limited to this, and the sound of the sound on the spot is not limited. It may be changed according to the environment. Here, the purpose is to obtain Pi from the sound pressure level of the background noise and the conversation sound. For example, when the voice continues for a long time such as reading, the time for only the background noise is small. A relatively large number of frames is preferable so as to be secured.

  The noise level analysis unit 90 first performs initialization (step S11). Time is the averaged number of times, Lnsum [j], Lvsum [j] are the background noise in the band number j, and the total sound pressure level of the conversation sound, Ln [j], Lv [j] are the background noise, conversation The sum of the sound pressure levels in one frame in the sound band number j, Lnave [j], Lave [j] is the average level of background noise and conversational sound of band number j.

  In subsequent step S12, 1 is added to Time (for example, from Time = 0 to Time = 1), and pitch extraction of the Time (= 1) frame is performed (step S13). If pitch extraction is possible (step S14: YES), the section is regarded as a conversation sound and band division is performed (step S15). Next, the sound pressure level Lv [j] of the conversational sound of the frame is obtained for each band j (step S16), and the average level Lvave [j] of the conversational sound up to the Time (= 1) frame is obtained (step S17).

  On the other hand, when pitch extraction cannot be performed in step S14 (step S14: NO), the section is regarded as background noise, and also in this case, band division is performed (step S18). Similarly, the sound pressure level Ln [j] of the background noise of the frame is obtained for each band j (step S19), and the average level Lnave [j] of the background noise up to the Time (= 1) frame is obtained (step S20). If it is determined that Time is a multiple of 5 (Time% 5 = 0) (step S21: YES), the average of the background noise level and the conversational sound level is set as a point Pi of the band j (step S22). When such Pi is set as the inflection point Pi, the inflection point Pi is just between the conversation sound level and the background noise level. Therefore, the conversation sound with a low volume is firmly amplified, and the background noise is amplified with a smaller amplification factor or amplification factor 1 than the conversation sound. If Time is not a multiple of 5 (step S21: NO), the process returns to step S12, time is increased by 1 in step S12, and the process moves to pitch extraction for the next frame (step S13). In this way, the point Pi is updated at a rate of once every five times.

  Next, the setting of P1 will be described. The inflection point Pi was previously set to an intermediate value between the average level of background noise and the average level of conversation sound, so that background noise has a smaller amplification factor than that of conversation sound. However, when the background noise suddenly becomes louder or when the noise is fluctuating with a wide dynamic range, the sound is greatly amplified or not amplified every frame unit, so that it sounds like intermittent noise. Therefore, in the present embodiment, the level (about Pi-15 dB to Pi-20 dB) reaching the low sound pressure level region R1 with a gentle inclination from the intermediate point Pi is set to P1. If the amplification from the point P1 to the intermediate point Pi is not a gradual amplification, the noise volume may be too high or too low for each processing frame. This is because the change in amplification becomes intense from frame to frame, and the noise becomes intermittent rather than continuous. By setting the point P1 to a value about 15 dB to 20 dB smaller than the intermediate point Pi, noise amplification can be suppressed (cut) as much as possible, or even if noise amplification occurs, the amplification factor can be reduced. The effect of reducing the change in volume was obtained. Note that when the value is 20 dB or more, for example, 30 dB smaller than the intermediate point Pi (P1 = Pi-30), the noise itself is amplified.

  The noise level analysis unit 90 thus calculates the intermediate points Pi and P1 for each band, and sends these data to the data input / output unit 70. The data input / output unit 70 stores the data received from the noise level analysis unit 90 in the memory 80. The reproduction signal processing unit 60a divides the input audio signal by the band dividing unit 65, and the P1, Pi, Ga, P2, and P3 for each band received from the data input / output unit 70 are converted into the sound pressure level amplifying unit 66. The sound pressure level amplifying unit 66 amplifies the sound pressure in accordance with the gain curves characterized by these parameters, and then the adder 67 adds the bands to obtain the left and right two-channel signals. Is returned to the earphone 20.

  As described above, according to the present embodiment, since the gain curve parameters are set according to the ambient noise level and the conversational sound level, the amplification characteristic can be adjusted better. That is, it is possible to amplify the vicinity of the conversational sound of the other party as much as possible without amplifying the background noise. Thereby, since the user is less tired of listening, it is possible to fully enjoy long-term listening and conversation.

  Another setting method of the maximum gain Ga will be described with reference to FIG. In the first embodiment, the difference between the minimum audible level Min of the user and the minimum audible level Ref of the normal hearing person is the maximum gain Ga, but the maximum gain Ga may be set by the following method.

  As an example, when the minimum audible level Min of the user in a predetermined frequency band is −40 dBFS, the minimum audible level Ref of the normal hearing person is −70 dBFS, and the optimum listening level of the user is 70 dBSPL, first, the high sound pressure level region R2 The boundary point P2 is set to −30 dBFS. The point P1 has the same value as in the first embodiment. Next, the inflection point Pi is set to the average level of the conversation sound level and the background noise level, which is the method described above. In this embodiment, the average level is −65 dBFS. Subsequently, a temporary segment consisting of a line segment connecting the minimum audible level Min of the user and the minimum audible level Ref of the normal hearing person, and a line segment connecting the minimum audible level Min of the user and the point P2 (optimum listening level Best). A characteristic line is set, and the difference between the value of the output signal at the average level of the conversational sound level and the background noise level obtained from the temporary characteristic line is defined as a maximum gain Ga. That is, in this embodiment, the output signal on the temporary characteristic line with the input signal of −65 dBFS becomes −38 dBFS, and the difference of 27 dBFS becomes the maximum gain Ga.

  Each point thus obtained, inflection point Pi (input signal -65 dBFS, output signal -38 dBFS) and point P2 (both input signal and output signal are -30 dBFS), and point P1 (both input signal and output signal are -90 dBFS). The characteristics of the output signal may be adjusted based on a gain curve that connects

DESCRIPTION OF SYMBOLS 1 Output terminal 2 Microphone input terminal 3 Earphone connection terminal 4 Audio input terminal 10 Microphone 20 Earphone 30 Impulse response measurement part 40 Auditory measurement part 50 Operation part 60 Playback signal processing part 61 Convolution part 62 Adder part 63 Amplifier 64 Adder 65 Band division Unit 66 sound pressure level amplification unit 67 addition unit 70 data input / output unit 80 memory 90 noise level analysis unit

Claims (4)

A noise level analyzer that analyzes the ambient noise level;
An audio measurement unit that can vary the volume of the sound to be output and measures the user's audio characteristics based on the amount of change in the volume;
A low sound pressure level region from the minimum sound pressure level to a predetermined range in the range from the minimum sound pressure level to the maximum sound pressure level of the input signal according to the noise level and the hearing characteristic data measured by the hearing measurement unit. And the intermediate sound pressure level region excluding the maximum sound pressure level and the high sound pressure level region within a predetermined range, and gradually increasing from the low sound pressure level region side to the first sound pressure level in the intermediate sound pressure level region. A signal processing unit that generates an output signal characteristic that sequentially decreases from the first sound pressure level to the high sound pressure level region side, and that performs signal processing according to the output signal characteristic;
An acoustic signal processing device comprising: The noise level analysis unit performs pitch extraction based on the ambient sound, and when the pitch is extracted, calculates the sound pressure level of the ambient sound as the sound pressure level of the conversation sound, and the pitch is not extracted Calculates the sound pressure level of the ambient sound as the sound pressure level of the noise, and based on the average value of the sound pressure level of the conversation sound and the sound pressure level of the noise calculated by a predetermined number of pitch extraction The acoustic signal processing device according to claim 1, wherein the first sound pressure level is set. The hearing measurement unit measures the optimum listening level of the user,
The acoustic signal processing device according to claim 1, wherein the signal processing unit sets a change point between the intermediate sound pressure level region and the high sound pressure level region based on the optimum listening level. A noise level analysis step for analyzing the ambient noise level;
An audio measurement step that can vary the volume of the sound to be output and measures the user's audio characteristics based on the amount of change in the volume;
A low sound pressure level region from the minimum sound pressure level to a predetermined range in the range from the minimum sound pressure level to the maximum sound pressure level of the input signal according to the noise level and the hearing characteristic data measured in the hearing measurement step. And the intermediate sound pressure level region excluding the maximum sound pressure level and the high sound pressure level region within a predetermined range, and gradually increasing from the low sound pressure level region side to the first sound pressure level in the intermediate sound pressure level region. A signal processing step of generating an output signal characteristic that sequentially decreases from the first sound pressure level to the high sound pressure level region side, and performing signal processing according to the output signal characteristic;
An acoustic signal processing method comprising:

Images