JP2008233672A - Masking sound generation apparatus, masking sound generation method, program, and recording medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a technique for generating a masking sound having sound characteristics most suitable for masking sound characteristic of a sound to be masked. <P>SOLUTION: In a masking sound generation apparatus according to the present invention, a storage means thereof stores a scramble sound signal having been processed in advance so that a meaning as a language can not be decided. In operation mode 1, a CPU analyzes sound characteristics of a noise generated in a sound space, reads a scramble sound signal similar to the analysis result out of the storage means, and outputs it as a masking sound. In operation mode 2, when a user specifies a scramble sound signal directly or when information associated with properties of a person using the sound space etc., is received, the CPU selects and reads a scramble sound signal out of the storage means according to the input contents and outputs it as a masking sound. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a technique for generating a masking sound.

  It is generally known that when a certain sound (target sound) is heard and there is another sound (masking sound) with acoustic characteristics (frequency characteristics, etc.) close to the target sound, the target sound becomes difficult to hear. This is called the masking effect. The masking effect is rooted in human auditory characteristics, and becomes more prominent as the masking sound frequency is closer to the target sound frequency and the masking sound volume level is higher relative to the target sound volume level. It is known to become.

Various acoustic techniques using this masking effect have been proposed in the past, and examples thereof include techniques disclosed in Patent Documents 1 and 2. Japanese Patent Application Laid-Open No. 2004-151620 discloses a technique for generating a masking sound by making an acquired sound meaningless by dividing an acquired sound into predetermined frames and playing back in reverse in each frame. Patent Document 2 discloses a technique for generating a masking sound by making a sound meaningless by dividing a sound signal into a plurality of segments and changing the order of the plurality of segments.
Japanese Patent Application No. 2006-242344 JP-T-2005-554061

According to the techniques described in Patent Documents 1 and 2, since a masking sound is generated in real time from the collected sound, high performance is required for processing the sound signal.
The present invention has been made in view of the above problems, and it is an object of the present invention to provide a technique for generating a masking sound having an acoustic characteristic most suitable for masking an acoustic characteristic of a sound to be masked.

  The masking sound generation apparatus according to the present invention stores a plurality of scrambled sound signals in which the time series of the sound signal is changed by dividing the sound signal into sections of a predetermined time length, and reconfiguring the sound signal. Storage means for storing each acoustic characteristic of the signal, acoustic characteristic analyzing means for collecting sound and analyzing the acoustic characteristic of the sound, acoustic characteristics analyzed by the acoustic characteristic analyzing means, and acoustics of the scrambled sound signal Output means for comparing a characteristic with a predetermined algorithm to determine a scrambled sound signal, reading the determined scrambled sound signal from the storage means, and outputting it.

  In the masking sound generation apparatus according to the present invention, in the above configuration, the output unit generates a scrambled sound signal read from the storage unit based on an acoustic characteristic of the sound analyzed by the acoustic characteristic analysis unit. Sound processing may be performed and output.

  Another configuration of the masking sound generation apparatus according to the present invention stores a plurality of scrambled sound signals in which the time series of the sound signal is changed by reconfiguring the sound signal into sections of a predetermined time length. Storage means for storing the acoustic characteristics of each of the scrambled sound signals, receiving means for receiving information on the acoustic characteristics of the sound to be masked from the operator, acoustic characteristics received by the receiving means and the scrambled sound signal A masking sound generating apparatus comprising: output means for comparing a sound characteristic with a predetermined algorithm to determine a scrambled sound signal, reading out the determined scrambled sound signal from the storage means, and outputting the signal.

  In the masking sound generation apparatus according to the present invention, in the above configuration, the output means includes a scrambled sound read from the storage means based on information on acoustic characteristics of the masked sound received by the receiving means. The signal may be subjected to acoustic processing and output.

  Further, another configuration of the masking sound generation device according to the present invention stores a plurality of scrambled sound signals whose time series is changed by dividing the sound signal into sections of a predetermined time length and reconfiguring the sound signal. Storing means, receiving means for receiving an instruction signal designating any of the scrambled sound signals stored in the storage means from an operator, and storing the scrambled sound signal indicated by the instruction signal received by the receiving means. Output means for reading out from the means and outputting.

  A masking sound generation apparatus according to the present invention is a scrambled sound signal in which the time series of each section is changed by receiving a sound signal and processing the sound signal into predetermined sections in any one of the above configurations And scramble means for generating and storing it in the storage means.

  The masking sound generation apparatus according to the present invention further includes receiving means for receiving information on acoustic characteristics of a space where the scrambled sound signal is emitted from an operator in any one of the above configurations, and the output means includes: Based on the information about the acoustic characteristics of the space received by the receiving means, the scrambled sound signal read from the storage means may be subjected to acoustic processing and output.

  The masking sound generation method according to the present invention stores a plurality of scrambled sound signals in which the time series of the sound signal is changed by dividing the sound signal into sections of a predetermined time length and reconfiguring the sound signal, A storage step of storing each acoustic characteristic of the scrambled sound signal; an acoustic characteristic analysis step of collecting sound and analyzing the acoustic characteristic of the sound; and the acoustic characteristic and the scrambled sound analyzed in the acoustic characteristic analysis step An output step of comparing the acoustic characteristics of the signal with a predetermined algorithm to determine a scrambled sound signal, and reading out the determined scrambled sound signal from the storage device and outputting it;

  Another configuration of the masking sound generation method according to the present invention is that a plurality of scrambled sound signals in which the time series of the sound signal is changed are stored in a storage device by reconfiguring the sound signal by dividing the sound signal into sections of a predetermined time length. A storage step for storing and storing each acoustic characteristic of the scrambled sound signal; a receiving stage for receiving information on the acoustic characteristic of the sound to be masked from an operator; and the acoustic characteristic received in the receiving stage and the scrambled An output step of comparing the acoustic characteristics of the sound signal with a predetermined algorithm to determine a scrambled sound signal, and reading out the determined scrambled sound signal from the storage device and outputting it;

  Further, another configuration of the masking sound generation method according to the present invention is a storage device that stores a scrambled sound signal in which the time series of the sound signal is changed by reconfiguring the sound signal by dividing the sound signal into sections of a predetermined time length. A plurality of storage stages, a reception stage for receiving an instruction signal designating one of the scrambled sound signals stored in the storage stage from an operator, and a scrambled sound signal indicated by the instruction signal received in the reception stage. And an output stage for reading out from the storage device and outputting it.

  The program according to the present invention stores a plurality of scrambled sound signals in which the time series of the sound signal is changed by dividing the sound signal into sections of a predetermined time length and reconfiguring each of the scrambled sound signals. Storage means for storing the acoustic characteristics, acoustic characteristic analysis means for collecting sound and analyzing the acoustic characteristics of the sound, acoustic characteristics analyzed by the acoustic characteristic analysis means, and acoustic characteristics of the scrambled sound signal A scrambled sound signal is determined by comparison with a predetermined algorithm, and the determined scrambled sound signal is read from the storage means and functions as an output means for outputting.

  Another configuration of the program according to the present invention is to store a plurality of scrambled sound signals in which the time series of the sound signal is changed by reconfiguring the computer by dividing the sound signal into sections of a predetermined time length. Storage means for storing the acoustic characteristics of each of the scrambled sound signals, receiving means for receiving information on the acoustic characteristics of the sound to be masked from the operator, acoustic characteristics received by the receiving means and the scrambled sound signal The scrambled sound signal is determined by comparing the acoustic characteristics with a predetermined algorithm, and the determined scrambled sound signal is read from the storage means and functions as output means for outputting.

  In another configuration of the program according to the present invention, the computer stores a plurality of scrambled sound signals in which the time series of the sound signal is changed by reconfiguring the sound signal into sections of a predetermined time length. Storing means, receiving means for receiving an instruction signal designating any of the scrambled sound signals stored in the storage means from an operator, and storing the scrambled sound signal indicated by the instruction signal received by the receiving means. It functions as an output means for reading out from the means and outputting.

  The computer-readable recording medium according to the present invention stores a plurality of scrambled sound signals in which the time series of each section is changed by dividing the sound signal into predetermined sections, and selects each scrambled data. It is memorized so that it can be read out automatically.

  With the masking sound generation device, the masking sound generation method, the program, or the recording medium according to the present invention, it is possible to generate a masking sound having an acoustic characteristic most suitable for masking the acoustic characteristic of the sound to be masked.

Hereinafter, embodiments of the present invention will be described with reference to the drawings.
(A: Configuration)
(A-1: Overall configuration)
FIG. 1 is a diagram showing a configuration of a sound masking system 1 according to the present invention. As shown in FIG. 1, a microphone 30 is suspended from the ceiling and installed in the acoustic space 20A. A speaker 40 is suspended from the ceiling and installed in the acoustic space 20B.

The microphone 30 picks up sound in the acoustic space 20 </ b> A (audible sound such as human speech or air-conditioning operation sound), converts it into an analog sound signal, and outputs the analog sound signal to the masking sound generator 10.
The speaker 40 receives an analog sound signal from the masking sound generator 10 and reproduces it in the acoustic space 20B.

(A-2: Configuration of the masking sound generator 10)
Next, the configuration of the masking sound generation apparatus 10 will be described with reference to FIG. The masking sound generator 10 generates a sound signal representing a masking sound (masker). The masking sound is emitted in the acoustic space 20B, and the contents of the conversation in the acoustic space 20A are difficult to be heard by other users of the acoustic space 20B (security protection). The sound that can be heard from 20A is prevented from disturbing the conversation or disturbing the work concentration (noise masking).

  A CPU (Central Processing Unit) 100 executes various programs stored in the storage unit 200 to perform operations characteristic of the present invention and control operations of each unit of the masking sound generation apparatus 10.

  The audio input unit 300 includes an analog / digital (hereinafter abbreviated as “A / D”) converter 310 and an input terminal 320. The microphone 30 is connected to the input terminal 320, and the sound signal generated by the microphone 30 is input to the A / D converter 310 via the input terminal 320. The A / D converter 310 performs A / D conversion on the sound signal received from the microphone 30 and outputs a digital sound signal to the CPU 100.

  The audio output unit 400 includes a D / A converter 410, an amplifier 420, and an output terminal 430. The D / A converter 410 converts the sound signal received from the CPU 100 into an analog sound signal by performing D / A conversion. The amplifier 420 adjusts the amplitude (master volume) of the sound signal received from the D / A converter 410 to an optimal value, and controls so that the masking effect is maximized. The amplification factor of the sound signal is controlled by the CPU 100 based on a signal from the operation unit 500 described later. The output terminal 430 is connected to the speaker 40, and the sound signal is output to the speaker 40 and emitted as a masking sound (masker) in the acoustic space 20B.

  The operation unit 500 is an input device having a touch panel, and outputs an operation content to the CPU 100 when the user of the masking sound generation device 10 is pressed. FIG. 3 is a diagram illustrating an appearance of the operation unit 500. The touch panel of the operation unit 500 includes an operation mode selection unit 510, a sound signal selection unit 520, a sex selection unit 530, an age selection unit 540, a language selection unit 550, an acoustic space selection unit 560, and a volume level selection unit 570.

  When a specific area on the touch panel is pressed by the user, the selected area is shaded as shown in the figure, and a signal indicating that the corresponding item is selected is displayed. It is output to the CPU 100. In the sound volume level selection unit 570, a larger sound volume level is associated with a larger number. Hereinafter, these signals are referred to as operation mode selection information, sound signal selection information, gender selection information, age selection information, language selection information, acoustic space selection information, and volume level selection information, respectively. Moreover, sex selection information, age selection information, language selection information, and acoustic space selection information are collectively referred to as condition setting information.

  In FIG. 2 again, the optical disk reproducing device 600 is a device for reading data recorded from the loaded optical disk. The read data is output to the CPU 100.

The storage unit 200 includes a ROM (Read Only Memory) 210 and a RAM (Random Access Memory) 220.
The ROM 210 stores a control program and data for causing the CPU 100 to execute functions characteristic of the present invention.
The RAM 220 has various storage areas and is used as a work area by the CPU 100. The RAM 220 has a sound signal storage area capable of storing each sound signal received from the sound input unit 300 for a predetermined time. The predetermined time is preferably as long as possible, and the masking sound generation device has high performance, but has an upper limit value depending on the capacity and performance of hardware resources. Therefore, in this embodiment, the predetermined time is set to 180 seconds. In addition, the RAM 220 stores various data such as parameters relating to the sound signal generation of the masking sound.
The units described above are connected via a bus 700 and exchange data with each other.

(A-3; control program and data)
Next, the control program stored in the ROM 210 will be described. The CPU 100 executes various processes including the process described below by executing the control program.

  First, the “acoustic characteristic analysis process” will be described. The acoustic characteristic analysis process is a process of dividing an input sound signal into sections of a predetermined length and analyzing speech speed, formant, and frequency characteristics in each generated fragment (hereinafter referred to as a frame).

  First, the analysis of speech speed will be described. In the present embodiment, “speech speed (speech speed)” is the speed at which a voice is emitted, and is defined by the number of syllables per unit time or the like. A syllable is a group of phonemes having a certain voice length (for example, vowels), or a group that can follow a very short phoneme (for example, consonants) before and / or after a phoneme having a certain voice length. means. In the acoustic characteristic analysis process, the CPU 100 generates a time axis waveform of the sound signal for each frame of the received sound signal, and performs a smoothing process on the envelope (envelope) of the time axis waveform. Then, the peak position of the waveform constituting each syllable is detected for each frame from the smoothed waveform, and the number of peaks is measured. Thereafter, the number of syllables per unit time obtained by dividing the peak number as the syllable number and dividing the syllable number by the frame length is calculated as the speech speed. Here, the peak means a portion having the maximum level in the waveform constituting each syllable. Although the speech speed varies from frame to frame, the CPU 100 analyzes the speech speed at that time for each frame, calculates the average value of these values, and the standard deviation σ, which is the variation between the average values, and outputs it. To do.

  Next, formant analysis will be described. A formant is a mountain formed by concentrating energy in a specific frequency region on the spectrum envelope of speech. This is a frequency spectrum (overtone component distribution pattern) inherent to a human voice or the like, and has a feature that it does not depend on the pitch or strength of the voice. It is known that the sex, age, language used, etc. of a speaker can be read by analyzing formants. In the acoustic characteristic analysis process, the CPU 100 performs a Fourier transform on the waveform of each received sound signal in each frame. Then, the CPU 100 obtains the logarithm of the amplitude spectrum obtained by the Fourier transform, and inversely transforms it to generate the spectrum envelope of each frame. Then, the CPU 100 extracts the first formant frequency, the second formant frequency, and the third formant frequency from the obtained low frequency envelope frequency. In the present embodiment, the frequencies of the first to third formants are extracted, but any one or two of them, or the fourth and subsequent formants may be analyzed.

Next, frequency characteristic analysis will be described. The CPU 100 reads the received sound signal for each frame and generates spectrum data in the frequency domain of each frame by Fourier transform. From the generated spectrum data, the pitch of the sound represented by the sound signal can be read.
The above is the acoustic characteristic analysis process.

  Next, “reverse processing” of sound signals will be described. In the reverse processing, the CPU 100 once converts each frame of the received sound signal into a signal in the time axis region. Then, each frame of the sound signal is read from the reverse on the time axis, and each sound signal is converted into a new sound signal. This process is a process of reading out old data and generating a new sound signal in a temporal order opposite to the order in which the original sound signal was generated. From the sound signal generated by the reverse processing, the contents contained in the sound signal before the processing cannot be understood.

Next, the “windowing process” for each frame of the sound signal will be described. The windowing process is a process of converting the waveform of the connection portion so that the sound smoothly transitions when frames whose contents are not continuous are connected.
Specifically, the CPU 100 multiplies the sound signal of each frame by a “shaping function” composed of, for example, a trigonometric function so that the head of each frame rises smoothly, and the tail of each frame smoothly Shape it so that it falls. When a continuous sound signal is divided into a plurality of frames by acoustic processing and connected in an order different from that of the original sound signal, click noise may occur in the connected portion. Noise is removed.

Next, data stored in the ROM 210 will be described.
First, the “frame length selection table” will be described. FIG. 4 is a diagram showing an example of the frame length selection table. In the frame length selection table, the frame length is associated with the above-described speech speed range. For example, a frame length value of 0.10 [second] is associated with a speech speed of 7.5 or more and less than 12.5 [second- ¹ ]. Here, the length of one frame is set to be approximately the same as the time of one syllable when the speech speed is an intermediate value in the range of each speech speed. That is, at a speech speed of 10 [seconds ^-1 ], the utterance speed of one syllable is 0.10 seconds, which corresponds to a range of speech speeds of 7.5 to less than 12.5 including the speech speed of 10 [seconds ^-1 ]. The frame length was set to this one syllable speech time (0.10 seconds). This is because when one frame length is extremely shorter than one syllable, one syllable is divided into a plurality of frames, and each frame may be recognized as the original syllable even if reversely played back. This is because if the time is extremely longer than one syllable, each syllable in one frame may be recognized as it is even if each frame is randomly reconstructed.

Next, the “scrambled sound signal” will be described. A scrambled sound signal is a sound signal obtained by scrambled human speech (meaningless or unintelligible). Specifically, human speech is collected and corresponding waveform data is generated, divided into a plurality of frames every predetermined time (for example, 100 milliseconds), and these are combined in a different order from the original speech. This is a newly generated sound signal. In the present embodiment, a plurality of scrambled sound signals (scrambled sound signals 1, 2, 3,...) Are stored in the ROM 210 in an initial setting process described later. Humans cannot understand the meaning of language from this scrambled sound signal.
The ROM 210 also stores a white noise sound signal as an example of broadband noise in addition to a human voice signal. White noise is noise having a uniform power spectral density in the measurement frequency band.

  Next, the “scrambled sound signal selection table” will be described. As shown in FIG. 5, in the scrambled sound signal selection table, sound generator attribute information and acoustic characteristic information of the sound are stored for each scrambled sound signal number that can identify each scrambled sound signal stored in the ROM 210. Has been written. The sound generator attribute information includes the gender, age, language, and name of the person who pronounced the voice that is the source of the scrambled sound signal. For example, the scrambled sound signal 1 is generated from a sound that is blown by “Mr. A”, a 30-year-old Japanese male. The acoustic characteristic information includes data relating to the speech speed, formant, and frequency characteristic of the scrambled sound signal. In the formant and frequency characteristic items, a file name for uniquely identifying the formant and frequency characteristic data is written, and the data is separately written in the ROM 210.

(B: Operation)
Next, the operation of this embodiment will be described.
(B-1: Initial setting process)
The CPU 100 performs an initial setting process before generating the masking sound. FIG. 6 is a flowchart showing a flow of processing performed by the CPU 100 in the initial setting processing.

  First, in step SA100, the CPU 100 receives a sound signal. Here, there are two methods for the CPU 100 to receive a sound signal. One is a method in which a user blows sound through the microphone 30 and the CPU 100 receives a sound signal through the sound input unit 300. The other method is a method of reading a sound signal from the optical disk on which the sound signal has been written by the optical disk reproducing device 600. In this case, the optical disk may be, for example, an optical disk sold as an off-the-shelf product, or may be one in which a user has previously written a sound signal on the optical disk.

  When the user finishes inputting the sound signal by any of the above methods, the sound generator attribute information (gender, age, language, and name of the person who pronounced the sound) regarding the sound signal is input via an input unit (not shown). Enter. The CPU 100 temporarily writes the received sound signal and sound generator attribute information in the RAM 220 in association with each other.

  In this operation example, the former method, that is, a method of inputting sound through the microphone 30, and the latter method, that is, a method of reading a sound signal from a storage medium such as an optical disk are used in combination. The sound signal input by the former method is as follows. Sound signals representing the pronunciation of “Mr. A”, a 30-year-old Japanese male, and “Mr. B”, a 25-year-old Japanese female, are input as sound signals that are the basis of the scrambled sound signals 1 and 2. . In addition, as a sound signal that is a source of the scrambled sound signal 3, a sound signal representing the pronunciation of “C group (5 people)” consisting of five Japanese men and women with an average age of 25 is input.

  The sound signal input by the latter method is as follows. A sound signal representing the pronunciation of a 10-year-old Japanese boy is input as the sound signal that is the basis of the scrambled sound signal 4. Further, a sound signal generated from the sound of a 30-year-old British male is input as the sound signal that is the basis of the scrambled sound signal 5.

  The sound signal to be input may be selected with reference to the frequency with which each user uses the acoustic space 20A and the type of language used in the acoustic space 20A. For example, when the acoustic space 20A is frequently used by “Mr. A”, “Mr. B”, and “C group”, or when meetings are frequently held in English, they are frequently used as described above. It is good to input the voice signal of the person who uses it and the sound signal of the language used.

  Next, in step SA110, the CPU 100 performs an acoustic characteristic analysis process. Specifically, the CPU 100 analyzes the speech speed, formant, and frequency characteristics in each sound signal written in the RAM 220, and associates the acoustic characteristic information, which is the analysis result, with the sound signal to be analyzed. Once written in the RAM 220.

  In step SA120, CPU 100 updates the scrambled sound signal selection table stored in ROM 210. Specifically, the CPU 100 reads sound generator attribute information and acoustic characteristics regarding each sound signal from the RAM 220 and writes them in the scrambled sound signal selection table. At this time, as shown in FIG. 5, the sound generator attribute information and the acoustic characteristics relating to the sound signals that are the basis of the scrambled sound signals 1, 2, 3, 4, and 5 are respectively scrambled sound signals 1, 2, 3, 4 , And 5 are written.

  In step SA130, the CPU 100 performs a sound signal scramble process. FIG. 7 is a flowchart showing the flow of the sound signal scramble process. FIG. 8 is a diagram showing a waveform of a sound signal accompanying the sound signal scramble process.

  In step SB100 in FIG. 7, the CPU 100 duplicates the sound signal written in the RAM 220. In this operation example, the CPU 100 duplicates the sound signal into three, and writes the duplicated sound signal in the RAM 220. Hereinafter, these sound signals are referred to as sound signals A, B, and C. Steps SB110 to SB150 described below are performed for each of the sound signals A, B, and C, and these sound signals are converted into different sound signals. The following processing may be executed simultaneously for three sound signals, or may be executed sequentially.

  In step SB110, the CPU 100 performs framing of the sound signal as follows. That is, the CPU 100 reads out information regarding the speech speed of the sound signal from the RAM 220. Then, the CPU 100 reads the frame length associated with the average value, the average value + σ, and the average value−σ in the frame length selection table stored in the ROM 210, and the sound signals A, B, And C are divided by the read frame lengths, and a frame generated as a result of the division is written in the RAM 220. Note that (a) -A, (a) -B, and (a) -C in FIG. 8 show a situation where the sound signals A, B, and C are divided by different frame lengths.

  In step SB120, CPU 100 performs the above-described reverse process for each frame of the sound signal written in RAM 220. As a result of the reverse processing, the frames of the sound signals A, B, and C are included in the frames as shown in FIGS. 8 (b) -A, (b) -B, and (b) -C, respectively. It is converted into data inverted in time.

In step SB130, a windowing process is performed on each frame. As a result, the waveform of the portion corresponding to the head and tail of each frame is shaped.
In step SB140, the CPU 100 randomly rearranges the order of the plurality of frames for each sound signal (see FIG. 8C).
In step SB150, the CPU 100 connects the sound signals of the rearranged frames to generate a new sound signal.
In step SB160, the CPU 100 mixes the sound signals A, B, and C separately processed in steps SB110 to 150, and generates a scrambled sound signal (see FIG. 8D).

  The scrambled sound signal generated by the above processing has the following characteristics. That is, in the generated scrambled sound signal, the fluctuation range of the volume level of the original sound signal becomes small and converges to an average volume level. This is because the original sound signal is divided into short frames and the frames are not only randomly rearranged, but also a plurality of sound signals that have undergone such processing are overlaid. . For this reason, the volume level of the scrambled sound signal is kept substantially constant, and the instability of the masking effect due to fluctuations in the volume level of the original sound signal is reduced.

  In addition, since the frame length for dividing the sound signal is appropriately set according to the speech speed, the phonemes included in the original sound are appropriately divided and have a high masking effect. In addition, sound is rendered meaningless by dividing phonemes and performing reverse processing within the frame, thereby protecting the user's privacy and security. In addition, since the windowing process is performed at the joint of each frame, the generated scrambled sound signal is a smoothly connected sound signal.

In FIG. 6 again, the CPU 100 writes the generated scrambled sound signal in the ROM 210 in step SA140.
Further, the CPU 100 displays “name” associated with the scrambled sound signal of the number in the scrambled sound signal selection table on the right side of each option of the sound signal selection unit 520.

  Note that the ROM 210 also stores a sound signal representing white noise in advance. Therefore, when the initial setting process is completed, the ROM 210 is in a state where a scrambled sound signal and a white noise sound signal are stored as the sound signal that is the basis of the masking sound.

(B-2; Masking sound generation process)
Next, the masking sound generation process will be described. FIG. 9 is a flowchart showing the flow of the masking sound generation process.
When executing the masking sound generation process, the user of the masking sound generation apparatus 10 operates the operation mode selection unit 510 of the operation unit 500 to select one of the operation modes 1 and 2. The operation unit 500 outputs operation mode information indicating the selected operation mode to the CPU 100. Hereinafter, the masking sound generation process when each operation mode is selected by the user will be described.

(B-2-1; operation mode 1)
This operation mode is a mode in which an appropriate scrambled sound signal is automatically selected for generating a masking sound based on the acoustic characteristics of the sound in the acoustic space 20A.

In step SC100, CPU 100 receives the operation mode information.
In step SC110, CPU 100 determines whether or not the received operation mode information is 1. In this operation mode, since the operation mode information is “1”, the determination result in step SC110 is “Yes”, and the process in step SC120 is performed.

  In step SC120, the CPU 100 receives a sound signal representing a sound in the acoustic space 20A and performs an acoustic characteristic analysis process on the sound signal. Since this process is the same as the acoustic characteristic analysis process in the initial setting process, description thereof is omitted.

  In step SC130, the CPU 100 reads any one appropriate sound signal from the scrambled sound signal written in the ROM 210 based on the result of the acoustic characteristic analysis process in step SC120. That is, the CPU 100 compares the acoustic characteristics (speech speed, formant, and frequency characteristics) obtained as an analysis result in step SC120 with a scrambled sound signal selection table, and selects a scrambled sound signal having the most similar acoustic characteristics. .

  In step SC140, the CPU 100 outputs the read sound signal (data of 180 seconds in the present embodiment) as a masking sound. Since the scrambled sound signal is data for 180 seconds, the scrambled sound signal is repeatedly output in a loop after 180 seconds from the start of output. The volume level of the output scrambled sound signal is set to an optimum value according to the volume level input by the user through the volume level selection unit 570, and this process is executed as an interrupt process.

  In this operation mode, the acoustic characteristics of the sound in the acoustic space 20A are analyzed, and a scrambled sound signal having the most similar acoustic characteristics to the sound is selected from a number of scrambled sound signals stored in the ROM 210. As described above, the highest masking effect is exhibited when the masking sound is similar to the acoustic characteristics of the target sound. Therefore, the output masking sound has acoustic characteristics most suitable for masking the sound generated in the acoustic space 20A.

(B-2-2; operation mode 2)
Next, the masking sound generation process in the operation mode 2 will be described. This operation mode is a mode in which the masking sound is automatically selected according to the content of the user's instruction.

In step SA100, the CPU 100 receives the operation mode information.
In step SA110, CPU 100 determines whether or not the received operation mode information is 1. In this operation mode, since the operation mode information is “2”, the determination result in step SC110 is “No”, and the process in step SC150 is performed.

  Now, the user inputs parameters relating to the generation of the masking sound by one of the following methods. First, the first method will be described. The user refers to the “name” displayed on the right side of the sound signal selection unit 520 of the operation unit 500 and directly designates one of the sound signals. For example, when “Mr. A” utters in the acoustic space 20A, the user presses “1” in the sound signal selection unit 520, and presses “5” when a conference in English is held.

  Another method is a method in which the user selects a specific option for one or more of the sex selection unit 530, the age selection unit 540, the language selection unit 550, and the acoustic space selection unit 560. In this case, the CPU 100 selects a sound signal based on the selected information. For example, when “adult” “male” speaks “English” in “office”, as shown in FIG. 3, gender selection unit 530, age selection unit 540, language selection unit 550, and acoustic space Each item of the selection unit 560 is selected.

The operation unit 500 outputs sound signal selection information or condition setting information according to the above-described operation content.
In step SC150, CPU 100 receives sound signal selection information or condition setting information from operation unit 500.

  In step SC <b> 130, CPU 100 selects a sound signal based on sound signal selection information or condition setting information received from operation unit 500. That is, when the CPU 100 receives the sound signal selection information, the scrambled sound signal represented by the sound signal selection information is read from the ROM 210 and output as a masking sound. Further, when the CPU 100 receives the condition setting information, the information regarding the gender, age, language, and type of acoustic space written in the condition setting information is compared with a scrambled sound signal selection table, and a predetermined algorithm, For example, a scrambled sound signal that matches the setting condition such as a sound signal with the largest number of matched items, a sound signal recently selected from the past selection history, or a sound signal with the highest frequency of use is read. The predetermined algorithm may be arbitrarily set according to a user request.

  At this time, if “house” is written in the acoustic space selection information, the CPU 100 may select a white noise sound signal as the masking sound. This is because masking sound generated from random noise such as white noise is generally less likely to cause discomfort and discomfort than masking sound generated from human sounds. This is because masking with white noise, which has a low level of discomfort and incongruity, is desired in houses where priority is placed on sex. Needless to say, the sound signal of white noise may be given priority even in cases other than “house”.

  In step SC140, CPU 100 outputs either the selected scrambled sound signal or white noise sound signal. The volume level of the output scrambled sound signal is set to an optimum value according to the volume level input by the user through the volume level selection unit 570. This process is executed as an interrupt process.

  In this operation mode, a plurality of scrambled sound signals that most closely match the sound and the acoustic characteristics of the acoustic space 20A are stored in the ROM 210 based on information such as the characteristics of the sound in the acoustic space 20A and the type of the acoustic space 20A. Scrambled sound signal or white noise. In this case, the user can easily generate an optimum masking sound without knowing what kind of sound signal is stored in the ROM 210.

(C: Modification)
Although one embodiment of the present invention has been described above, it is needless to say that the embodiment may be modified as described below. Moreover, you may use combining the deformation | transformation described below.

(1) In the above embodiment, the case where the CPU 100 of the masking sound generation apparatus 10 executes many of the processes characteristic of the present invention has been described. However, a hardware module for performing each process is provided to perform the same process. You may make it do.

(2) In the above-described embodiment, the case has been described in which various processes (frame processing, reverse processing, windowing processing, and randomization processing) are all performed on the sound signal in the initial setting processing. However, it is not always necessary to perform all the processes described above, and it is sufficient that the sound signal is modified to such an extent that the meaning as a language cannot be understood by combining these processes.

(3) In the above embodiment, a case has been described in which a plurality of pieces of information (gender, age, language, speech speed, formant, frequency characteristics) regarding a sound signal are written in the scrambled sound signal selection table. However, in the acoustic characteristic analysis process, it is not always necessary to analyze all of the speech speed, formant, and frequency characteristic, and it is not necessary to write all these items in the initial setting process. Further, it is not necessary to write all of the sound generator attribute information. The CPU 100 may select the scrambled sound signal having the highest degree of matching within the range of the written item.

(4) In the said embodiment, an example of the method of the acoustic characteristic analysis process was demonstrated. However, the analysis method of each acoustic characteristic is not limited to the method described above, and any method may be used as long as a similar analysis result can be obtained.

(5) In the above embodiment, the process of analyzing the acoustic characteristics of the sound signal collected in the acoustic space 20A in the operation mode 1 has been described. However, the space where the masking sound is actually emitted is the acoustic space 20B, and an obstacle that changes the acoustic characteristics such as walls, that is, a sound insulation structure exists between the two acoustic spaces. Therefore, before performing the acoustic characteristic analysis process, the CPU 100 performs a filtering process simulating the sound insulation characteristic of the sound insulation structure on the target sound signal to give an acoustic effect when the sound signal passes through the wall. Then, an acoustic characteristic analysis process may be performed. As a result, the generated masking sound is generated from a sound signal simulating noise heard by the user of the acoustic space 20B, and therefore a higher masking effect can be expected.

(6) In the above embodiment, the case where the microphone 30 and the speaker 40 are provided in separate acoustic spaces has been described. However, the microphone 30 and the speaker 40 may be installed in the same acoustic space. For example, when the microphone 30 and the speaker 40 are installed in the acoustic space 20A, a masking sound is generated from the conversation contents of the user in the acoustic space 20A, and the masking sound is emitted in the acoustic space 20A. Both conversational content and masking sound can be heard. As a result, it becomes difficult for the user of the acoustic space 20B to understand the conversation content of the user of the acoustic space 20A. In this case, it is natural that the microphone 30 and the speaker 40 perform arrangement and signal processing so that no howling occurs.

(7) In the above embodiment, the case where the microphone 30 and the speaker 40 are installed in different acoustic spaces has been described. However, by placing the microphone 30 and the speaker 40 apart in the same space, a highly confidential conversation is made in the area near the microphone 30, and the masking sound is emitted to the user in the area near the speaker 40. The conversation content may not be heard.

(8) In the above embodiment, the case where the microphone 30 is installed in the acoustic space 20A and the speaker 40 is installed in the acoustic space 20B has been described. However, both the microphone 30 and the speaker 40 may be installed in each of a plurality of acoustic spaces, for example, the acoustic spaces 20A and 20B. In that case, the masking sound generation apparatus 10 has an input means, and when a user has a highly confidential conversation, the user inputs that fact via the input means, and the masking sound generation apparatus 10 receives the input. In such an acoustic space, the sound may be collected by the microphone 30 and the masking sound generated in the other acoustic space may be emitted.

(9) In the above embodiment, the CPU 100 duplicates the sound signal input in the sound signal scramble processing into three sound signals having different frame lengths, performs different sound signal processing on each sound signal, and thereafter These sound signals were mixed to generate a masking sound. However, the number of systems of sound signals to be handled is not limited to 3, but may be 1 or 2 or 4 or more, but the effect as a masking sound is higher as the number of systems is larger.

(10) In the above embodiment, the CPU 100 determines the average value, the average value + σ, and the average value −σ from the average value of the speech speed and the standard deviation σ that is temporal variation in the framing of the sound signal. The case has been described where the calculation and application to the framing process of each of the duplicated sound signals has been described. However, the values used are not limited to the average value and the average value ± σ. For example, standard error may be used instead of σ, or a predetermined value may be used instead of σ.
Further, in the frame length selection table, three frame lengths are associated with the speech speed, and the CPU 100 reads out the three frame lengths corresponding to the average value of the speech speed, and sets the read frame length as the frame length. It is sufficient to divide each sound signal into frames.

(11) In the above embodiment, the case where the duplicated sound signal is divided by different frame lengths has been described. However, a plurality of replicated sound signals may be divided by a common frame length. In that case, the CPU 100 may read the frame length corresponding to the average value of the speech speed, and divide each sound signal into frames using the read frame length.

(12) In the above embodiment, the case where white noise is used as random noise has been described. However, the type of random noise is not limited to white noise, but may be other sound sources such as pink noise (noise whose power spectral density is inversely proportional to frequency), or generated in advance from noise or vibration actually generated from air conditioning. You may use the sound signal which did.

(13) In the above embodiment, a case has been described in which an optical disk playback device is provided to write an existing sound signal to the ROM 210, and a sound signal written to the optical disk is written to the ROM 210. However, the device for taking in the sound signal from the outside is not limited to the optical disk reproducing device, and for example, the sound signal can be downloaded from a server via a communication network such as the Internet, or the masking sound generating device 10 can be connected to an external device. It is also possible to provide an I / O unit that mediates the connection of the audio signal and move a sound signal from the flash memory connected to the I / O unit to the ROM 210.

(14) In the above embodiment, the case where the operation modes 1 and 2 are selectable has been described. However, the processes shown in both operation modes need not be executable, and only one of them may be executed.

(15) In the above embodiment, the case where the sound signal scramble process is performed in the initial setting process and the scramble sound signal is written in the ROM 210 in advance has been described. However, the CPU 100 may store the received sound signal in the ROM 210 without performing the sound signal scramble process, and output the masking sound while performing the sound signal scramble process in the masking sound generation process.
When the scrambled sound signal is stored on the optical disc, the sound signal scramble process may not be performed in the initial setting process.

(16) In the above-described embodiment, it has been described that a plurality of scrambled sound signals are generated, the sound signals are stored in the ROM 210, and are selected and used when generating a masking sound. Therefore, a storage medium storing “a set of a plurality of scrambled sound signals” in the above embodiment is created, and a sound signal read from the storage medium is selected and output by another sound signal playback device. You may do it.

(17) In the above embodiment, in the operation mode 1, the case where the CPU 100 selects the scrambled sound signal most similar to the acoustic characteristics of the received sound signal by referring to the scrambled sound signal selection table has been described. In the operation mode 2, the case where the CPU 100 refers to the scrambled sound signal selection table and selects the scrambled sound signal having the highest degree of coincidence with the various conditions input from the user has been described. However, in any of the above cases, the CPU 100 selects the scrambled sound signal in the scrambled sound signal selection table from among the scrambled sound signals that are not the highest in coincidence but the degree of coincidence exceeds a certain level. You may do it.

(18) In the above-described embodiment, the case has been described in which the scrambled sound signal or the white noise sound signal having the most similar acoustic characteristics is selected in the operation mode 1 based on the analysis result of the acoustic characteristic analysis process. However, a plurality of sound signals may be selected simultaneously. In this case, for example, in the operation mode 1, the operation unit 500 may be provided with an input unit for setting the number of sound signals to be selected. Then, the CPU 100 may select a predetermined number of sound signals in the order in which the acoustic characteristics are the same. In addition, when the sound signal is directly selected by the operator in the operation mode 2, a plurality of sound signals corresponding to the plurality of options pressed in the sound signal selection unit 520 may be selected. In this way, since a plurality of sound signals are output as a masking sound, it can be expected that effective masking is performed.

(19) In the above embodiment, various acoustic effects may be imparted to the output masking sound based on the content of the acoustic space selection information. For example, when the acoustic space selection information is “Hall”, the CPU 100 may add a reverberation effect to the read scrambled sound signal or white noise sound signal. In addition, as a method for imparting reverberation, a conventional technique such as superimposing a plurality of sound signals delayed for a predetermined time (convolution processing of reflected sound using an FIR filter) can be applied. Further, a difference may be provided in the reverberation time and the number of sound signals to be superimposed depending on the type of the selected acoustic space such as “conference room” or “hall”.
Moreover, you may perform timbre conversion by the process etc. which convolve a reflected sound as another acoustic effect. In the conference room, the sound is reflected by the walls and desks of the conference room or reverberated in the room, and is converted into a tone unique to the conference room. Therefore, when the acoustic space selection information is “conference room”, for example, the CPU 100 adjusts the waveform of the read scrambled sound signal or white noise sound signal, It may be converted into a timbre.
By performing the above acoustic processing, a masking sound with less discomfort is generated.

(20) In the above embodiment, the case where the type of room such as “meeting room”, “house”, “hall”, “office room”, etc. is written as an option in the acoustic space selection unit 560 has been described. However, for example, options indicating the acoustic characteristics of the room such as “a space in which sound is well reflected” and “anechoic room” may be provided. In short, the acoustic characteristic selection information may be information indicating the acoustic characteristics of the acoustic space.

(21) In the above embodiment, the case where the sound signal is selected based on the acoustic space selection information in the operation mode 2 has been described. However, the present invention is not limited to such a case, and input to the acoustic space selection unit 560 may be possible in any of the operation modes. By doing so, it is possible to perform various kinds of acoustic processing on the masking sound based on the acoustic characteristics of the acoustic space 20 as described in the modification (19).

(22) In the above-described embodiment, the case where the analysis result of the acoustic characteristic analysis process is used for selecting a scrambled sound signal or white noise in the operation mode 1 has been described. In that case, in the acoustic characteristic analysis process, the reverberation time and reflected sound characteristic (impulse response) in the acoustic space 20A are further measured, and the read sound signal is analyzed based on the analysis result of the acoustic characteristic analysis process. Various sound processings may be performed and output. For example, when the acoustic space 20A is a “hole”, the reverberation process may be performed on the read sound signal because the hall generally has a very long reverberation time.

(23) In the above embodiment, in the operation mode 2, when the condition setting information is input, the sound signal is read and output based on the condition setting information. In that case, various acoustic processes may be performed on the read sound signal based on the condition setting information. For example, when the gender selection information is “male”, the sound signal may be equalized to emphasize the low frequency component and convert it to a sound signal imitating a “male” voice. good. Further, when the age selection information is “dwarf”, the sound signal may be equalized to emphasize the frequency component having a high frequency and converted to a sound signal imitating the voice of “dwarf”. .

It is the figure which showed the structure of the acoustic space 20 in which the masking sound production | generation apparatus 10 was provided. 1 is a block diagram showing a configuration of a masking sound generation device 10. FIG. FIG. 3 is a diagram illustrating an appearance of an operation unit 500. It is the figure which showed an example of the frame length selection table. It is the figure which showed an example of the scramble sound signal selection table. It is the flowchart which showed the flow of the initial setting process. It is the flowchart which showed the flow of the sound signal scramble process. It is the figure which showed the waveform of the sound signal in a sound signal scramble process. It is the flowchart which showed the flow of the masking sound production | generation process.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Sound masking system, 10 ... Masking sound production | generation apparatus, 20A, 20B ... Acoustic space, 30 ... Microphone, 40 ... Speaker, 100 ... CPU, 200 ... Memory | storage part, 210 ... ROM, 220 ... RAM, 300 ... Audio | voice input part 310 ... A / D converter, 320 ... input terminal, 400 ... audio output unit, 410 ... D / A converter, 420 ... amplifier, 430 ... output terminal, 500 ... operation unit, 510 ... operation mode selection unit, 520 ... sound Signal selection unit, 530 ... gender selection unit, 540 ... age selection unit, 550 ... language selection unit, 560 ... acoustic space selection unit, 570 ... volume level selection unit, 600 ... optical disk playback device, 700 ... bus

Claims (14)

A memory for storing a plurality of scrambled sound signals in which the time series of the sound signal is changed and for storing acoustic characteristics of each of the scrambled sound signals by reconfiguring the sound signal by dividing it into sections of a predetermined time length Means,
Acoustic characteristic analysis means for collecting sound and analyzing acoustic characteristics of the sound;
The acoustic characteristic analyzed by the acoustic characteristic analyzing means and the acoustic characteristic of the scrambled sound signal are compared by a predetermined algorithm to determine a scrambled sound signal, and the determined scrambled sound signal is read from the storage means and output. And an output means. A masking sound generation apparatus comprising: an output means; A memory for storing a plurality of scrambled sound signals in which the time series of the sound signal is changed and for storing acoustic characteristics of each of the scrambled sound signals by reconfiguring the sound signal by dividing it into sections of a predetermined time length Means,
Receiving means for receiving information about the acoustic characteristics of the sound to be masked from the operator;
Output means for comparing the acoustic characteristics received by the receiving means and the acoustic characteristics of the scrambled sound signal by a predetermined algorithm to determine a scrambled sound signal, reading out the determined scrambled sound signal from the storage means and outputting it And a masking sound generating device. Storage means for storing a plurality of scrambled sound signals in which the time series of the sound signal is changed by dividing the sound signal into sections of a predetermined time length and reconfiguring;
Receiving means for receiving an instruction signal designating any of the scrambled sound signals stored in the storage means from an operator;
A masking sound generating apparatus comprising: output means for reading out and outputting the scrambled sound signal indicated by the instruction signal received by the receiving means from the storage means. Scrambling means for receiving a sound signal, processing the sound signal into predetermined sections, and generating a scrambled sound signal in which the time series of each section is changed, and storing the scrambled sound signal in the storage means. The masking sound generation apparatus according to claim 1, wherein the masking sound generation apparatus is a masking sound generation apparatus. The output means performs an acoustic process on the scrambled sound signal read from the storage means based on the acoustic characteristics of the sound analyzed by the acoustic characteristics analysis means, and outputs the result. Masking sound generator. The output means performs an acoustic process on the scrambled sound signal read from the storage means based on information about the acoustic characteristics of the masked sound received by the receiving means, and outputs the scrambled sound signal. The masking sound generator described in 1. Receiving means for receiving information on acoustic characteristics of a space where the scrambled sound signal is emitted from an operator;
5. The output unit according to claim 1, wherein the output unit performs an acoustic process on the scrambled sound signal read from the storage unit based on information about the acoustic characteristics of the space received by the receiving unit. A masking sound generator according to claim 1. By dividing and reconfiguring the sound signal into sections of a predetermined time length, a plurality of scrambled sound signals whose time series of the sound signal are changed are stored in a storage device, and the acoustic characteristics of each of the scrambled sound signals are A memory stage to memorize,
An acoustic characteristic analysis stage for collecting sound and analyzing the acoustic characteristics of the sound;
The acoustic characteristic analyzed in the acoustic characteristic analysis step and the acoustic characteristic of the scrambled sound signal are compared by a predetermined algorithm to determine a scrambled sound signal, and the determined scrambled sound signal is read from the storage device and output. A masking sound generation method comprising: an output stage. By dividing and reconfiguring the sound signal into sections of a predetermined time length, a plurality of scrambled sound signals whose time series of the sound signal are changed are stored in a storage device, and the acoustic characteristics of each of the scrambled sound signals are A memory stage to memorize,
A receiving stage for receiving information about the acoustic characteristics of the sound to be masked from the operator;
An output step of determining the scrambled sound signal by comparing the acoustic characteristic received in the receiving step with the acoustic characteristic of the scrambled sound signal by a predetermined algorithm, and reading and outputting the determined scrambled sound signal from the storage device And a masking sound generating method comprising: A storage step of storing a plurality of scrambled sound signals in which the time series of the sound signal is changed by dividing the sound signal into sections of a predetermined time length and reconfiguring the sound signal;
Receiving an instruction signal designating any of the scrambled sound signals stored in the storage step from an operator;
A masking sound generation method comprising: an output step of reading out and outputting the scrambled sound signal indicated by the instruction signal received in the receiving step from the storage device. Computer
A memory for storing a plurality of scrambled sound signals in which the time series of the sound signal is changed and for storing acoustic characteristics of each of the scrambled sound signals by reconfiguring the sound signal by dividing it into sections of a predetermined time length Means,
Acoustic characteristic analysis means for collecting sound and analyzing acoustic characteristics of the sound;
The acoustic characteristic analyzed by the acoustic characteristic analyzing means and the acoustic characteristic of the scrambled sound signal are compared by a predetermined algorithm to determine a scrambled sound signal, and the determined scrambled sound signal is read from the storage means and output. Program to function as output means. Computer
A memory for storing a plurality of scrambled sound signals in which the time series of the sound signal is changed and for storing acoustic characteristics of each of the scrambled sound signals by reconfiguring the sound signal by dividing it into sections of a predetermined time length Means,
Receiving means for receiving information about the acoustic characteristics of the sound to be masked from the operator;
Output means for comparing the acoustic characteristics received by the receiving means and the acoustic characteristics of the scrambled sound signal by a predetermined algorithm to determine a scrambled sound signal, reading out the determined scrambled sound signal from the storage means and outputting it Program to function as. Computer
Storage means for storing a plurality of scrambled sound signals in which the time series of the sound signal is changed by dividing the sound signal into sections of a predetermined time length and reconfiguring;
Receiving means for receiving an instruction signal designating any of the scrambled sound signals stored in the storage means from an operator;
A program for causing a scrambled sound signal indicated by an instruction signal received by the receiving means to function as an output means for reading out and outputting from the storage means.   By dividing the sound signal into predetermined sections and processing them, a plurality of scrambled sound signals in which the time series of each section is changed are stored, and the scrambled data is stored so that the respective scrambled data can be selectively read. recoding media.

Images