JP6036141B2 - Sound processor - Google Patents

Description

  The present invention relates to a technique for suppressing a reverberation component of an acoustic signal.

  Techniques for suppressing reverberation components included in acoustic signals have been conventionally proposed. For example, in Patent Document 1, a prediction filter coefficient of a reverberation component is estimated by using a probability model of a prediction filter coefficient that estimates a reverberation component included in an acoustic signal, and a reverberation component is calculated using a prediction filter after estimation. Techniques for suppressing are disclosed. Non-Patent Document 1 discloses a technique for suppressing a reverberation component by estimating an inverse filter of a transfer function from a sound source to a sound collection point and applying the estimated inverse filter to an acoustic signal. .

JP 2009-212599 A

K. Furuya, et al. "Robust speech dereverberation using multichannel blind deconvolution with spectral subtraction", IEEE Transantions on Audio, Speech, and Language Processing, vol. 15, no. 5, p.1579-1591, 2007

  However, under the techniques of Patent Document 1 and Non-Patent Document 1, when the suppression strength of the reverberation component is increased, there may be a problem that the sound quality after the reverberation suppression is deteriorated. In view of the above circumstances, an object of the present invention is to suppress deterioration in sound quality due to reverberation suppression.

  In order to solve the above problems, an acoustic processing device according to the present invention includes a harmonic separation unit that separates an acoustic signal into a harmonic component and a non-harmonic component, and a first reverberation component that suppresses the reverberation component of the harmonic component. Suppression means, and second reverberation suppression means for suppressing the reverberation component of the non-harmonic component with a suppression intensity different from the suppression intensity of the reverberation component by the first reverberation suppression means. With the above configuration, the reverberation component of the acoustic signal is suppressed with different suppression strengths for the harmonic component and the non-harmonic component. Therefore, compared with the case where the reverberation component is suppressed without separating the acoustic signal into the harmonic component and the non-harmonic component, an effective reverberation suppression effect is realized while suppressing deterioration in sound quality due to the reverberation suppression. It is possible. For example, when it is assumed that a decrease in sound quality due to reverberation suppression tends to be manifested by non-harmonic components as compared to harmonic components, the suppression strength of the reverberation components by the first reverberation suppression unit is the second. A configuration that exceeds the suppression strength of the reverberation component by the reverberation suppression means is preferable. Further, by adopting a configuration in which the suppression strength of the reverberation component by the second reverberation suppression unit exceeds the suppression strength of the reverberation component by the first reverberation suppression unit, for example, a characteristic acoustic effect (effect) is given to the acoustic signal. It is possible.

  The acoustic processing apparatus according to a preferred aspect of the present invention includes a synthesis processing unit that synthesizes a harmonic component after processing by the first dereverberation unit and a non-harmonic component after processing by the second dereverberation suppression unit. According to the above configuration, it is possible to generate an acoustic signal in which both the reverberation component of the harmonic component and the reverberation component of the non-harmonic component are suppressed.

  In a preferred aspect of the present invention, the first dereverberation suppression means is a moving average of the harmonic component intensities to which the first moving average coefficient (for example, the average number NH1 or the smoothing coefficient αH1) is applied. First intensity averaging means (for example, intensity averaging unit 42) for calculating intensity QH1 (k, m)) and a first adjustment value (for example, adjustment value GH (k, m) for suppressing the reverberation component of the harmonic component. ) In accordance with the first average intensity, and first suppression for causing the first adjustment value calculated by the first adjustment value calculation means to act on the harmonic component. Processing means (for example, suppression processing unit 46), and the second dereverberation suppressing means applies a second moving average coefficient (for example, average number HP1 or smoothing coefficient αP1) different from the first moving average coefficient. The second average intensity (for example, average intensity QP1 (k, m)) is a moving average of the wave component intensity. The second intensity averaging means (for example, the intensity averaging unit 52) to be determined and the second adjustment value (for example, the adjustment value GP (k, m)) for suppressing the reverberation component of the inharmonic component according to the second average intensity Second adjustment value calculation means (for example, adjustment value calculation section 54) to be calculated, and second suppression processing means (for example, suppression processing section) that causes the second adjustment value calculated by the second adjustment value calculation means to act on the subharmonic component. 56). In the above configuration, the first moving average coefficient applied to the moving average by the first intensity averaging means and the second moving average coefficient applied to the moving average by the second intensity averaging means are individually selected, The suppression strength of the reverberation component by the first reverberation suppression unit and the suppression strength of the reverberation component by the second reverberation suppression unit can be easily made different from each other. For example, in the configuration for calculating the simple moving average of the intensities of the harmonic component and the inharmonic component, the number (time length) of the intensities targeted for the moving average is suitable as the first moving average coefficient and the second moving average coefficient. In the configuration for calculating the exponential moving average of the intensity of the harmonic component and the non-harmonic component, the smoothing coefficient is suitable as the first moving average coefficient and the second moving average coefficient. However, the configuration and method in which the first dereverberation suppression unit and the second dereverberation suppression unit suppress the reverberation component are not limited to the above examples, and a known dereverberation technique can be arbitrarily adopted.

  In a preferred aspect of the present invention, the second adjustment value calculation means is configured so that the suppression strength of the reverberation component in the late reverberation section of the non-harmonic component exceeds the suppression strength of the reverberation component outside the late reverberation section. An adjustment value is calculated for each unit period of the acoustic signal. In the above configuration, the fluctuation in volume in the later reverberation section is suppressed by increasing the suppression strength of the reverberation component in the later reverberation section. Therefore, there is an advantage that deterioration of sound quality after reverberation suppression can be suppressed. In addition, the specific example of the above aspect is later mentioned, for example as 3rd Embodiment.

  The acoustic processing device according to each of the above aspects is realized by hardware (electronic circuit) such as a DSP (Digital Signal Processor) dedicated to processing of an acoustic signal, or a general-purpose calculation such as a CPU (Central Processing Unit). This is also realized by cooperation between the processing device and the program. The program according to the present invention includes a harmonic separation process that separates an acoustic signal into a harmonic component and a non-harmonic component, a first dereverberation process that suppresses a reverberation component of the harmonic component, and a first dereverberation process. A computer executes a second reverberation suppression process for suppressing a reverberation component of a non-harmonic component with a suppression strength different from the suppression strength of the reverberation component. According to the above program, the same operation and effect as the sound processing apparatus according to the present invention are realized.

  The program of the present invention is provided in a form stored in a computer-readable recording medium and installed in the computer. The recording medium is, for example, a non-transitory recording medium, and an optical recording medium (optical disk) such as a CD-ROM is a good example, but a known arbitrary one such as a semiconductor recording medium or a magnetic recording medium This type of recording medium can be included. For example, the program of the present invention can be provided in the form of distribution via a communication network and installed in a computer.

1 is a block diagram of a sound processing apparatus according to a first embodiment of the present invention. It is explanatory drawing of the relationship between the average intensity | strength of an acoustic signal, and an adjustment value. It is a block diagram of a 1st dereverberation part and a 2nd dereverberation part. It is a flowchart of the operation | movement which the adjustment value calculation part of the 2nd dereverberation part in 3rd Embodiment performs.

<First Embodiment>
FIG. 1 is a block diagram of a sound processing apparatus 100 according to the first embodiment of the present invention. As shown in FIG. 1, a signal supply device 12 and a sound emitting device 14 are connected to the sound processing device 100. The signal supply device 12 supplies the acoustic signal SX to the acoustic processing device 100.

  The acoustic signal SX is a numerical series indicating an acoustic time waveform obtained by mixing harmonic components and non-harmonic components. The harmonic component means a harmonic acoustic component such as a performance sound of a musical instrument such as a stringed instrument or a wind instrument or a human vocal sound, and a non-harmonic component means a performance sound of a percussion instrument or various noises (for example, air conditioning equipment). Non-harmonic acoustic components such as operating sounds and environmental sounds such as crowded sounds in crowds. A reverberation component derived from reflection or scattering in the acoustic space is added to the sound indicated by the acoustic signal SX. Specifically, there is an acoustic signal SX obtained by adding a reverberation effect to an existing sound such as a recorded sound or a synthesized sound, or an acoustic signal SX of an acoustic actually recorded in an acoustic space having a reverberation effect. The signal is supplied from the signal supply device 12 to the sound processing device 100. For example, a sound collection device that collects ambient sound to generate an acoustic signal SX, a playback device that acquires the acoustic signal SX from a portable or built-in recording medium and supplies the acoustic signal SX to the acoustic processing device 100, or a communication network A communication device that receives the sound signal SX from the sound signal and supplies the sound signal SX to the sound processing device 100 may be employed as the signal supply device 12.

  The sound processing device 100 is a time domain sound signal in which the reverberation component (late reverberation component) of the sound signal SX supplied by the signal supply device 12 is suppressed (that is, sound in which the direct sound component and the initial reflection component of the sound signal SX are emphasized). Signal) SY is a signal processing device (reverberation suppression device). The sound emitting device 14 (for example, a speaker or headphones) emits a sound wave according to the acoustic signal SY generated by the acoustic processing device 100. The illustration of a D / A converter that converts the acoustic signal SY from digital to analog and an amplifier that amplifies the acoustic signal SY are omitted for the sake of convenience.

  As shown in FIG. 1, the sound processing device 100 is realized by a computer system including an arithmetic processing device 22 and a storage device 24. The storage device 24 stores a program executed by the arithmetic processing device 22 and various data used by the arithmetic processing device 22. A known recording medium such as a semiconductor recording medium or a magnetic recording medium or a combination of a plurality of types of recording media can be arbitrarily employed as the storage device 24. A configuration in which the acoustic signal SY is stored in the storage device 24 (therefore, the signal supply device 12 can be omitted) is also suitable.

  The arithmetic processing device 22 executes a program stored in the storage device 24, thereby generating a plurality of functions (frequency analysis unit 32, harmonic separation unit 34, first reverberation unit 34) for generating the acoustic signal SY from the acoustic signal SX. The suppression unit 40, the second dereverberation suppression unit 50, the synthesis processing unit 62, and the waveform generation unit 64) are realized. A configuration in which each function of the arithmetic processing device 22 is distributed to a plurality of devices, or a configuration in which a dedicated signal processing circuit shares a part of the functions of the arithmetic processing device 22 may be employed.

  The frequency analysis unit 32 sequentially generates a spectrum (complex spectrum) X (k, m) of the acoustic signal SX for each unit period (frame) on the time axis. The symbol k is a variable that indicates an arbitrary frequency (frequency band) on the frequency axis, and the symbol m is a variable that indicates an arbitrary unit period on the time axis. For generation of the spectrum X (k, m), for example, a known frequency analysis such as a short-time Fourier transform can be arbitrarily employed. Note that a filter bank including a plurality of bandpass filters having different passbands can be used as the frequency analysis unit 32.

  The harmonic separation unit 34 separates the acoustic signal SX into a harmonic component and a non-harmonic component. Specifically, the harmonic separation unit 34 generates a harmonic component spectrum XH (k, m) and a non-harmonic component spectrum XP (k, m) from the spectrum X (k, m) of the acoustic signal SX. Generated sequentially for each unit period. For example, the harmonic separation unit 34 calculates the cepstrum of the acoustic signal SX for each unit period by logarithmic discrete Fourier transform of the spectrum X (k, m) of the acoustic signal SX, and the cepstrum of each unit period is calculated as the acoustic signal SX. Low-order region corresponding to the general structure of the amplitude spectrum | X (k, m) |, and a fine periodic structure of the amplitude spectrum | X (k, m) | (ie, fundamental component and multiple harmonic components) Is divided into higher-order regions corresponding to harmonic structures arranged at equal intervals on the frequency axis), so that the harmonic component spectrum XH (k, m) and the non-harmonic component spectrum XP (k, m ). In addition, a well-known technique is arbitrarily employ | adopted for isolation | separation of a harmonic component and a non-harmonic component. For example, the harmonic component of the acoustic signal SX is continuous in the time axis direction while the non-harmonic component is continuous in the frequency axis direction. An HPSS (Harmonic and Percussive Sound Separation) technique for separating non-harmonic components can also be applied. As shown in the spectrum XH (k, m) and spectrum XP (k, m), subscript H is added to elements related to harmonic components (Harmonic) and related to non-harmonic components (Percussive). A subscript P is added to the element to be processed.

  The first reverberation suppression unit 40 in FIG. 1 suppresses the reverberation component of the harmonic component after separation by the harmonic separation unit 34. Specifically, the first reverberation suppressing unit 40 sequentially generates a spectrum YH (k, m) of an acoustic component in which a reverberation component is suppressed from a harmonic component spectrum XH (k, m) for each unit period. On the other hand, the second reverberation suppression unit 50 suppresses the reverberation component of the non-harmonic component after separation by the harmonic separation unit 34. Specifically, the second reverberation suppressing unit 50 sequentially generates the spectrum YP (k, m) of the acoustic component in which the reverberation component is suppressed from the spectrum XP (k, m) of the non-harmonic component for each unit period. . Note that specific configurations and operations of the first dereverberation suppressing unit 40 and the second dereverberation suppressing unit 50 will be described later.

  1 synthesizes the harmonic component after processing by the first dereverberation suppression unit 40 and the non-harmonic component after processing by the second dereverberation suppression unit 50. Specifically, the synthesis processing unit 62 generates the spectrum YH (k, m) of the harmonic component after dereverberation by the first dereverberation unit 40 and the non-harmonic component after dereverberation by the second dereverberation unit 50. The spectrum Y (k, m) of the acoustic signal SY is generated by adding (or weighted sum) with the spectrum YP (k, m) (Y (k, m) = YH (k, m) + YP (k, m)).

  The waveform generation unit 64 generates a time domain acoustic signal SY from the spectrum Y (k, m) generated by the synthesis processing unit 62 for each unit period. That is, the waveform generation unit 64 converts the spectrum Y (k, m) of each unit period into, for example, a time domain signal by short-time Fourier inverse transform, and interconnects the unit periods that are adjacent to each other. Generate SY. As understood from the above description, in the acoustic signal SY, the reverberation component (the reverberation component of the harmonic component and the reverberation component of the non-harmonic component) of the acoustic signal SX is suppressed. The acoustic signal SY generated by the waveform generator 64 is supplied to the sound emitting device 14 and is emitted as a sound wave.

<Specific examples of reverberation suppression>
Prior to explaining the reverberation suppression of the harmonic component by the first reverberation suppression unit 40 and the reverberation suppression of the non-harmonic component by the second reverberation suppression unit 50, the reverberation component is suppressed from the spectrum X (k, m) of the acoustic signal SX. Processing to be performed will be described.

Attention is paid to the average intensity Q1 (k, m) and average intensity Q2 (k, m) of the power | X (k, m) | ² per unit period of the acoustic signal SX. The average intensity Q1 (k, m) is defined by the following formula (1A), and the power over the unit period of N1 pieces (N1 is a natural number of 2 or more) | X (k, m) | ² The moving average (simple moving average). On the other hand, the average intensity Q2 (k, m) is defined by the following formula (1B) of the power | X (k, m) | ² over N2 (N2 is a natural number) unit periods. It is a moving average. The number N1 of unit periods added to the calculation of the average intensity Q1 (k, m) exceeds the number N2 of unit periods added to the calculation of the average intensity Q2 (k, m) (N1> N2).

Part (B) of FIG. 2 is a graph depicting changes over time of the average intensity Q1 (k, m) and the average intensity Q2 (k, m) at an arbitrary frequency. As shown in part (A) of FIG. 2, the average intensity Q1 (k, m) assuming that the acoustic signal SX is a room impulse response in which the power | X (k, m) | ² (power density) exponentially decays in time. ) And average intensity Q2 (k, m) are shown in part (B) of FIG.

As can be seen from part (B) of FIG. 2, the average intensity Q1 (k, m) and average intensity Q2 (k, m) follow the power | X (k, m) | ² of the acoustic signal SX. Changes over time. However, since the average number N1 applied to the calculation of the average intensity Q1 (k, m) exceeds the average number N2 applied to the calculation of the average intensity Q2 (k, m), the average intensity Q1 (k, m) is In addition, it follows the time change of the power | X (k, m) | ² of the acoustic signal SX with lower followability (change rate) than the average intensity Q2 (k, m). Specifically, as shown in part (B) of FIG. 2, the average intensity Q2 (k, m) exceeds the average intensity Q1 (k, m) in the section immediately after the time t0 of the start of the indoor impulse response. Increase with rate of change. The average intensity Q1 (k, m) and the average intensity Q2 (k, m) reach peaks at different points on the time axis, and the average intensity Q2 (k, m) is the average intensity Q1 (k, m). ) Decrease at a rate of change exceeding.

  As described above, since the average intensity Q1 (k, m) and the average intensity Q2 (k, m) change at different rates, the average intensity Q1 (k, m) and the average intensity Q2 (k, m) Is reversed at a specific time point tx on the time axis. The section from the time t0 to the time tx (the section where the average intensity Q2 (k, m) exceeds the average intensity Q1 (k, m)) SA corresponds to the section where the direct sound component and the initial reflection component of the room impulse response exist The section after the time point tx (the section where the average intensity Q1 (k, m) exceeds the average intensity Q2 (k, m)) SB corresponds to the section where the late reverberation component of the indoor impulse response exists.

Ratio of average intensity Q2 (k, m) to average intensity Q1 (k, m) (hereinafter referred to as “intensity ratio”) R (k, m) (R (k, m) = Q2 (k, m) / Q1 ( Assume an adjustment value G (k, m) calculated by the following equation (2) according to k, m)). The predetermined value Gmax is set to 1, for example, and the predetermined value Gmin is set to a numerical value lower than the predetermined value Gmax (for example, a numerical value not less than 0 and less than 1).

  The change in the adjustment value G (k, m) when the average intensity Q1 (k, m) and the average intensity Q2 (k, m) change as shown in the part (B) of FIG. 2 is the part (C) of FIG. Is shown in FIG. As understood from part (C) of FIG. 2, the intensity ratio R (k, m) exceeds the predetermined value Gmax in the section SA where the average intensity Q2 (k, m) exceeds the average intensity Q1 (k, m). The adjustment value G (k, m) is maintained at the predetermined value Gmax. In the section SB where the intensity ratio R (k, m) exceeds the predetermined value Gmin in the section SB where the average intensity Q2 (k, m) is lower than the average intensity Q1 (k, m), the adjustment value G (k, m ) Is set to the intensity ratio R (k, m) and decreases with time. In the section SB2 in which the intensity ratio R (k, m) is lower than the predetermined value Gmin in the section SB, the adjustment value G (k, m) is maintained at the predetermined value Gmin. That is, the adjustment value G (k, m) is set to a predetermined value (maximum value) Gmax in the section SA where the direct sound component and the initial reflection component exist, and is set to a predetermined value (minimum value) in the section SB where the late reverberation component exists. ) Decreases with time to Gmin. Therefore, by applying (typically multiplying) the adjustment value G (k, m) to the spectrum X (k, m) of the acoustic signal SX, the reverberation component of the acoustic signal SX is suppressed (direct sound component or initial value). It is possible to emphasize the reflection component. The first dereverberation suppression unit 40 and the second dereverberation suppression unit 50 of the first embodiment suppress the reverberation component using the above principle.

  As described above, the adjustment value G (k, m) for reverberation suppression is calculated according to the relative ratio between the average intensity Q1 (k, m) and the average intensity Q2 (k, m). As the difference between the intensity Q1 (k, m) and the average intensity Q2 (k, m) increases, the suppression intensity (degree of suppression effect) of the reverberation component due to the action of the adjustment value G (k, m) increases. That is, the reverberation component suppression strength tends to increase as the difference between the average number N1 of the average intensities Q1 (k, m) and the average number N2 of the average intensities Q2 (k, m) increases.

FIG. 3 is a block diagram illustrating a specific configuration of the first dereverberation suppressing unit 40 and the second dereverberation suppressing unit 50. As shown in FIG. 3, the first dereverberation suppression unit 40 includes an intensity averaging unit 42, an adjustment value calculation unit 44, and a suppression processing unit 46. The intensity average unit 42 calculates an average intensity QH1 (k, m) and an average intensity QH2 (k, m) as a moving average of the power | XH (k, m) | ² of the harmonic component of the acoustic signal SX. The average intensity QH1 (k, m) is a moving average of power | XH (k, m) | ² over a unit period of NH1 (NH1 is a natural number of 2 or more) as expressed by the following formula (3A): Simple moving average), and the average intensity QH2 (k, m) is represented by the following formula (3B), and the power over a unit period of NH2 (where NH2 is a natural number) | XH (k, m) | ² Is a moving average.

The average number NH1 of the formula (3A) exceeds the average number NH2 of the formula (3B) (NH1> NH2). That is, the average intensity QH1 (k, m) is a moving average over a long time compared to the average intensity QH2 (k, m). Therefore, the average intensity QH1 (k, m) power at | XH (k, m) | 2 of the constant τH1 time smoothing the average intensity QH2 (k, m) power at | XH (k, m) | 2 of The smoothing time constant τH2 is exceeded (τH1> τH2). That is, like the relationship between the average intensity Q1 (k, m) and the average intensity Q2 (k, m) described with reference to part (B) of FIG. 2, the average intensity QH1 (k, m) It follows the power | XH (k, m) | ² of the harmonic component with lower followability than the intensity QH2 (k, m).

The adjustment value calculation unit 44 in FIG. 3 calculates the adjustment value GH (k, m) according to the average intensity QH1 (k, m) and average intensity QH2 (k, m) of the harmonic component calculated by the intensity average unit 42. Each frequency is calculated sequentially for each unit period. Specifically, as expressed by the following formula (4), the intensity ratio RH (k, m) (RH (k, m) of the average intensity QH2 (k, m) to the average intensity QH1 (k, m) = QH2 (k, m) / QH1 (k, m)) is a numerical value between the predetermined value Gmax and the predetermined value Gmin (Gmin <RH (k, m) <Gmax), the adjustment value calculation unit 44 The adjustment value GH (k, m) is set to the intensity ratio RH (k, m). The adjustment value calculation unit 44 sets the adjustment value GH (k, m) to the predetermined value Gmax when the intensity ratio RH (k, m) exceeds the predetermined value Gmax, and the intensity ratio RH (k, m). Is less than the predetermined value Gmin, the adjustment value GH (k, m) is set to the predetermined value Gmin.

  As understood from the above description with reference to part (C) of FIG. 2, the adjustment value GH (k, m) calculated by the adjustment value calculation unit 44 suppresses the reverberation component of the harmonic component of the acoustic signal SX. It can be used as a coefficient (spectral gain). 3 suppresses the reverberation component of the harmonic component by applying the adjustment value GH (k, m) calculated by the adjustment value calculation unit 44 to the harmonic component of the acoustic signal SX. The reverberation component suppression is sequentially performed for each frequency for each unit period. Specifically, the suppression processing unit 46 adjusts the adjustment value GH corresponding to the unit period and frequency common to the spectrum X (k, m) for the harmonic component spectrum XH (k, m) of the acoustic signal SX. The spectrum YH (k, m) of the harmonic component after reverberation suppression is calculated by multiplying (k, m) (YH (k, m) = GH (k, m) · XH (k, m) ).

As understood from the above description, the formula (3B) applied to the calculation of the average number NH1 and the average intensity QH2 (k, m) of the formula (3A) applied to the calculation of the average intensity QH1 (k, m). ) Of the average intensity QH1 (k, m) and the average intensity QH2 (k, m) (the harmonic component power | XH () k, m) | ² ) increases, and as a result, the reverberation component suppression strength tends to increase.

On the other hand, as shown in FIG. 3, the second dereverberation suppression unit 50 includes an intensity averaging unit 52, an adjustment value calculation unit 54, and a suppression processing unit 56. Similar to the intensity average unit 42, the intensity average unit 52 is a moving average of the power | XP (k, m) | ² of the non-harmonic component of the acoustic signal SX, and the average intensity QP1 (k, m) and average intensity QP2 ( k, m). The average intensity QP1 (k, m) is a moving average of power | XP (k, m) | ² over a unit period of NP1 (NP1 is a natural number of 2 or more) as expressed by the following formula (5A). The average intensity QP2 (k, m) is a moving average of power | XP (k, m) | ² over NP2 (NP2 is a natural number) unit periods, as expressed by the following formula (5B). .

The average number NP1 of the formula (5A) exceeds the average number NP2 of the formula (5B) (NP1> NP2). That is, the average intensity QP1 (k, m) is a moving average over a long time compared to the average intensity QP2 (k, m). Therefore, the average intensity QP1 (k, m) power at | XP (k, m) | When ^second smoothing constant τP1 the average intensity QP2 (k, m) power at | XP (k, m) | ² of It exceeds the smoothing time constant τP2 (τP1> τP2). That is, the average intensity QP1 (k, m) follows the power | XP (k, m) | ² of the non-harmonic component with lower followability than the average intensity QP2 (k, m).

The adjustment value calculation unit 54 in FIG. 3 represents the average intensity QP1 (k, m) of the non-harmonic component calculated by the intensity average unit 52, as expressed by the following formula (6), as with the adjustment value calculation unit 44. ) And the adjustment value GP (k, m) corresponding to the average intensity QP2 (k, m) are sequentially calculated for each frequency for each unit period. Specifically, the intensity ratio RP (k, m) (RP (k, m) = QP2 (k, m) / QP1 (k, m) of the average intensity QP2 (k, m) to the average intensity QP1 (k, m) According to m)), the adjustment value GP (k, m) is calculated by the following equation (6). The adjustment value GP (k, m) can be used as a coefficient (spectrum gain) for suppressing the reverberation component of the non-harmonic component of the acoustic signal SX.

  The suppression processing unit 56 suppresses the reverberation component of the non-harmonic component by applying the adjustment value GP (k, m) calculated by the adjustment value calculation unit 54 to the non-harmonic component of the acoustic signal SX. Specifically, the suppression processing unit 56 multiplies the spectrum XP (k, m) of the non-harmonic component of the acoustic signal SX by the adjustment value GP (k, m), so that the non-harmonic component after dereverberation is suppressed. The spectrum YP (k, m) is calculated (YP (k, m) = GP (k, m) · XP (k, m)).

As understood from the above description, the larger the difference (for example, the difference (NP1−NP2)) between the average number NP1 of the equation (5A) and the average number NP2 of the equation (5B), the average intensity QP1 (k, m ) And the average intensity QP2 (k, m) (the power of the non-harmonic component of the acoustic signal SX | difference in the tracking degree with respect to XP (k, m) | ² ) increases, resulting in suppression of the reverberation component There is a tendency for strength to increase.

  By the way, the reverberation component is a highly random acoustic component in which a large number of reflected sounds and scattered sounds arriving at the receiving point from multiple directions in the acoustic space are added, and the initial reflection component is a spatial component with respect to the receiving point. There is a tendency that the directivity is clear and the randomness is low compared to the reverberation component. On the other hand, the non-harmonic component is a transient acoustic component whose acoustic characteristics fluctuate irregularly in time and is highly random. The harmonic component is an acoustic component in which the acoustic characteristics are temporally sustained and is random. There is a tendency to be low. That is, the reverberation component and the inharmonic component have a common tendency in acoustic characteristics that the randomness is high, and the initial reflection component and the harmonic component have a tendency in acoustic characteristics that the randomness is low. Due to the commonality of the acoustic characteristics described above, in the separation processing by the harmonic separation unit 34, the reverberation component of the acoustic signal SX is highly likely to be selected as a non-harmonic component, and the initial reflection of the acoustic signal SX is performed. There is a general tendency that components are likely to be sorted into harmonic components.

  In addition, since non-harmonic components such as percussion instrument sounds are transient as described above, energy in a section other than immediately after sounding (the attack portion) is generally lower than harmonic components. Therefore, if the suppression strength of the reverberation component with respect to the non-harmonic component is too high, a decrease in sound quality after the reverberation suppression (for example, generation of musical noise) tends to be obvious. On the other hand, as described above, the harmonic component is rich in early reflection components and has higher energy than the non-harmonic component, so even if the suppression strength of the reverberation component is set high, the sound quality after reverberation suppression is reduced. There is a tendency that it is difficult to be manifested in comparison with non-harmonic components. In consideration of the above tendency, in the first embodiment, the first reverberation suppressing unit 40 and the second reverberation suppression are performed so that the suppression strength of the reverberation component with respect to the harmonic component exceeds the suppression strength of the reverberation component with respect to the non-harmonic component. Each of the units 50 executes reverberation component suppression with different suppression intensities.

  Specifically, the difference between the average number NH1 of the average intensity QH1 (k, m) corresponding to the harmonic component and the average number NH2 of the average intensity QH2 (k, m) (for example, the difference (NH1−NH2)) is Each average so as to exceed the difference (difference (NP1-NP2)) between the average number NP1 of the average intensity QP1 (k, m) corresponding to the non-harmonic component and the average number NP2 of the average intensity QP2 (k, m) The number (NH1, NH2, NP1, NP2) is selected. For example, assuming that the average number NH2 and the average number NP2 are set to the same numerical value, the average number NH1 is set to a value higher than the average number NP1.

  As described above, in the first embodiment, the acoustic signal SX is separated into the harmonic component and the non-harmonic component, and the reverberation component is suppressed with different suppression intensities for the harmonic component and the non-harmonic component. . Specifically, the suppression strength for the harmonic component exceeds the suppression strength for the non-harmonic component. Therefore, compared with the case where the reverberation component is suppressed without separating the acoustic signal SX into the harmonic component and the non-harmonic component, an effective reverberation suppression effect is realized while suppressing deterioration in sound quality due to the reverberation suppression. Is possible.

  In the first embodiment, according to the intensity ratio RH (k, m) between the average intensity QH1 (k, m) and the average intensity QH2 (k, m) following the time change of the harmonic component of the acoustic signal SX. The adjustment value GH (k, m) is calculated and the intensity ratio RP (k, m) between the average intensity QP1 (k, m) and the average intensity QP2 (k, m) following the time change of the non-harmonic component The adjustment value GP (k, m) is calculated accordingly. Therefore, the reverberation component of the acoustic signal SX is suppressed by a simple process compared with the technique of Patent Document 1 that estimates the prediction filter coefficient of the reverberation component and the technique of Non-Patent Document 1 that estimates the transfer function and generates the inverse filter. There is an advantage that you can.

Second Embodiment
A second embodiment of the present invention will be described below. In addition, about the element which an effect | action and function are the same as that of 1st Embodiment in each form illustrated below, the reference | standard referred by description of 1st Embodiment is diverted and each detailed description is abbreviate | omitted suitably.

The intensity average unit 42 of the first dereverberation suppressing unit 40 in the second embodiment has the harmonic component power | XH (k, m) | ² as expressed by the following equations (7A) and (7B). The average strength QH1 (k, m) and average strength QH2 (k, m) are calculated by the exponential moving average.

The smoothing coefficient (forgetting coefficient) αH1 in Expression (7A) is lower than the smoothing coefficient αH2 in Expression (7B) (0 ≦ αH1 <αH2 ≦ 1). Therefore, like the first embodiment, the average intensity QH1 (k, m) is the mean intensity QH2 (k, m) and compared to low followability in the harmonic component power | XH (k, m) | 2 (ΤH1> τH2).

As in the first embodiment, the adjustment value calculation unit 44 adjusts the adjustment value GH (k, m) according to the average intensity QH1 (k, m) and the average intensity QH2 (k, m) calculated by the intensity average unit 42. And the suppression processing unit 46 uses the adjustment value GH (k, m) calculated by the adjustment value calculation unit 44 as the harmonic component (spectrum XH (k, m)) of the acoustic signal SX. To act on. Thus, equation (7A) of the power difference (harmonic component of the smoothing coefficient αH1 smoothing coefficient equation (7B) αH2 | XH (k , m) | average intensity for ² QH1 (k, m) and average There is a tendency that the suppression intensity of the reverberation component increases as the intensity QH2 (k, m) differs in the degree of follow-up.

The intensity averaging unit 52 of the second dereverberation suppression unit 50 is similar to the intensity averaging unit 42 of the first dereverberation suppression unit 40, as expressed by the following equations (8A) and (8B). The average intensity QP1 (k, m) and average intensity QP2 (k, m) are calculated by the exponential moving average of power | XP (k, m) | ² .

The smoothing coefficient αP1 in Expression (8A) is lower than the smoothing coefficient αP2 in Expression (8B) (0 ≦ αP1 <αP2 ≦ 1). Accordingly, as in the first embodiment, the average intensity QP1 (k, m) is lower in tracking ability than the average intensity QP2 (k, m), and the power | XP (k, m) | Follow ² (τP1> τP2).

  Similar to the first embodiment, the adjustment value calculation unit 54 calculates the adjustment value GP (k, m) corresponding to the average intensity QP1 (k, m) and the average intensity QP2 (k, m) using the formula (6). The suppression processing unit 56 causes the adjustment value GP (k, m) to act on the non-harmonic component (spectrum XP (k, m)) of the acoustic signal SX. Therefore, the reverberation component suppression strength tends to increase as the difference between the smoothing coefficient αP1 in Expression (8A) and the smoothing coefficient αP2 in Expression (8B) increases.

  Also in the second embodiment, similarly to the first embodiment, the first dereverberation suppression unit 40 and the second dereverberation suppression are performed so that the suppression strength of the reverberation component with respect to the harmonic component exceeds the suppression strength of the reverberation component with respect to the non-harmonic component. Each of the units 50 executes reverberation component suppression with different suppression intensities. Specifically, the difference between the smoothing coefficient αH1 of the average intensity QH1 (k, m) corresponding to the harmonic component and the smoothing coefficient αH2 of the average intensity QH2 (k, m) (eg, difference (αH2−αH1)) Exceeds the difference (for example, difference (αP2−αP1)) between the smoothing coefficient αP1 of the average intensity QP1 (k, m) corresponding to the non-harmonic component and the smoothing coefficient αP2 of the average intensity QP2 (k, m) Thus, each smoothing coefficient (αH1, αH2, αP1, αP2) is selected. For example, assuming that the smoothing coefficient αH2 and the smoothing coefficient αP2 are set to the same numerical value, the smoothing coefficient αH1 is set to a numerical value lower than the smoothing coefficient αP1. Therefore, the same effects as those of the first embodiment are realized in the second embodiment.

As understood from the above description, the average number (NH1, NH2, NP1, NP2) of the first embodiment and the smoothing coefficients (αH1, αH2, αP1, αP2) of the second embodiment are the values of the acoustic signal SX. It is included as a moving average coefficient applied to a moving average of power (| XH (k, m) | ² , | XP (k, m) | ² ). Therefore, the intensity average unit 42 of the first dereverberation unit 40 is a moving average of the harmonic component power | XH (k, m) | ^{2 to which} the first moving average coefficients (NH1, NH2, αH1, and αH2) are applied. It is included as an element for calculating the average intensity (QH1 (k, m), QH2 (k, m)), and the intensity average unit 52 of the second dereverberation unit 50 includes second moving average coefficients (NP1, NP2, αP1,. This is included as an element for calculating the average intensity (QP1 (k, m), QP2 (k, m)) by the moving average of the power | XP (k, m) | ² of the inharmonic component to which αP2) is applied.

<Third Embodiment>
When the reverberation time of the acoustic signal SX is long, the average intensity Q1 (k, m) (QH1 (k, m), QP1 (k, m)) is the average intensity Q2 (k, m) (QH2 (QH ( k, m), QP2 (k, m)), the intensity ratio R (k, m) (RH (k, m), RP (k, m)) becomes unstable, and as a result There is a possibility that the sound quality of the acoustic signal SY deteriorates (for example, the sound volume fluctuates in a short cycle). The above tendency is easily manifested in the non-harmonic component of the acoustic signal SX. Therefore, in the third embodiment, fluctuations in volume are suppressed in the later reverberation section of the non-harmonic component.

  The adjustment value calculation unit 54 of the second reverberation suppression unit 50 in the third embodiment uses the adjustment value GP (k, m) of each unit period as a unit period in the later reverberation section of the non-harmonic component and outside the latter reverberation section. Thus, the fluctuation in the volume of the acoustic signal SY in the later reverberation section is suppressed by calculating separately from the unit section. Specifically, the adjustment value calculation unit 54 determines that the adjustment value GP (k, m) of each unit period in the later reverberation section is lower than the adjustment value GP (k, m) of each unit period outside the later reverberation section. The adjustment value GP (k, m) is calculated so that the suppression strength of the reverberation component in the later reverberation section is higher than the suppression strength of the reverberation component outside the later reverberation section. FIG. 4 is a flowchart of processing executed by the adjustment value calculation unit 54 of the third embodiment.

  As shown in FIG. 4, the adjustment value calculation unit 54 calculates the adjustment value GP (k, m) for each unit period by the calculation of Equation (6) (ST1), and each unit period corresponds to the late reverberation section. Whether or not (ST2). For example, in consideration of the tendency that the average intensity QP2 (k, m) decreases to a smaller value in the late reverberation interval, the adjustment value calculation unit 54 calculates the average intensity QP2 (k, m) for each unit period and a predetermined value. It is determined whether each unit period corresponds to a late reverberation section by comparing with the threshold value QTH. That is, a unit period in which the average intensity QP2 (k, m) exceeds the threshold value QTH is determined as a unit period in the later reverberation section, and a unit period in which the average intensity QP2 (k, m) is lower than the threshold value QTH is outside the later reverberation section. It is determined as a unit period. In addition, the method of determining the inside / outside of a late reverberation section is not limited to the above illustration.

  The adjustment value calculation unit 54 corrects the adjustment value GP (k, m) calculated in step ST1 according to the determination result in step ST2 (ST3). Specifically, the adjustment value calculation unit 54 determines the adjustment value GP (k, m) for each unit period outside the late reverberation section as the calculated value in Expression (6). On the other hand, the adjustment value calculation unit 54 lowers the adjustment value GP (k, m) of the unit period in the late reverberation section from the calculation value in Expression (6). For example, the adjustment value calculation unit 54 determines the adjustment value GP (k, m) by multiplying a predetermined coefficient γ (0 <γ <1) by the calculated value in Expression (6). As can be understood from the above description, the volume of the acoustic signal SY decreases in the section corresponding to the late reverberation section of the non-harmonic component. Therefore, it is possible to suppress the deterioration of the sound quality of the acoustic signal SY (for example, the fluctuation of the volume).

  In the above description, the adjustment value GP (k, m) calculated by the adjustment value calculation unit 54 is illustrated as being distinguished inside and outside the latter reverberation section of the non-harmonic component. A similar configuration can be adopted for the adjustment value GH (k, m) calculated for the wave component. That is, the adjustment value calculation unit 44 adjusts the adjustment value GH (k, m) for each unit period in the later reverberation section to be lower than the adjustment value GH (k, m) for each unit period outside the later reverberation section. The value GH (k, m) is calculated.

<Modification>
Each of the above-described embodiments can be variously modified. Specific modifications are exemplified below. Two or more aspects arbitrarily selected from the following examples can be appropriately combined.

(1) A configuration in which the average intensity QH2 (k, m) in each of the above-described embodiments is replaced with the harmonic component power | XH (k, m) | ² (that is, the average number NH2 of Equation (3B) or Equation (7B ) In which the smoothing coefficient αH2 is set to 1). Similarly, the configuration in which the average intensity QP2 (k, m) is replaced with the power of the inharmonic component | XP (k, m) | ² (the average number NP2 of the equation (5B) and the smoothing coefficient αP2 of the equation (8B)) Can be adopted. In addition, the method of calculating the adjustment value GH (k, m) according to the average intensity QH1 (k, m) (and the average intensity QH2 (k, m)), the average intensity QP1 (k, m) (and the average intensity The method of calculating the adjustment value GP (k, m) according to QP2 (k, m)) is arbitrary. For example, the adjustment value GH (k, m) is calculated by a predetermined calculation using the average intensity QH1 (k, m) as a variable, or the adjustment value is calculated by a predetermined calculation using the average intensity QP1 (k, m) as a variable. A configuration for calculating GP (k, m) may also be employed.

As understood from the above description, the intensity averaging unit 42 of the first dereverberation suppressing unit 40 uses the power of the harmonic component | XH (k to which the first moving average coefficient (average number NH1 or smoothing coefficient αH1) is applied. m) | ² is included as an element for calculating the average intensity QH1 (k, m) with a moving average of ² , and the adjustment value calculation unit 44 averages the adjustment value GH (k, m) for suppressing the reverberation of the harmonic component. It is included as an element to be calculated according to the strength QH1 (k, m). Similarly, the intensity average unit 52 of the second dereverberation suppression unit 50 has the power | XP (k, m) | ² of the non-harmonic component to which the second moving average coefficient (average number NP1 and smoothing coefficient αP1) is applied. It is included as an element for calculating the average intensity QP1 (k, m) by the moving average, and the adjustment value calculation unit 54 uses the adjustment value GP (k, m) for reverberation suppression of the non-harmonic component as the average intensity QP1 (k, m, It is included as an element to be calculated according to m).

(2) The object of the moving average by the intensity averaging unit 42 of the first dereverberation unit 40 is not limited to the harmonic component power | XH (k, m) | ² . For example, the amplitude of the harmonic component | XH (k, m) | 4 squared or amplitude | XH (k, m) | 4 mean intensity in the moving average of QH1 (k, m) and average intensity QH2 (k, m) Can also be calculated. That is, the intensity averaging unit 42 calculates the moving average of the harmonic component intensities (| XH (k, m) |, | XH (k, m) | ² , | XH (k, m) | ⁴ ) of the acoustic signal SX. It is included as an element. Similarly, the intensity averaging unit 52 of the second dereverberation suppressing unit 50 uses the intensities (| XP (k, m) |, | XP (k, m) | ² , | XP (k m) | ⁴ ) is included as a moving average element. In each of the above-described embodiments, the adjustment value GH (k, m) is applied to the harmonic component spectrum XH (k, m), but the suppression processing unit 46 uses the harmonic component power | XH (k, m). ) | ² may be adopted in which the adjustment value GH (k, m) is applied. Similarly, the adjustment value GP (k, m) can be applied to the power | XP (k, m) | ² of the non-harmonic component.

(3) The configuration and method in which the first dereverberation suppression unit 40 and the second dereverberation suppression unit 50 suppress the reverberation component are arbitrary, and are not limited to the above-described examples. Also, a configuration in which the reverberation component suppression method is different between the first dereverberation suppressing unit 40 and the second dereverberation suppressing unit 50 may be employed. For example, the first dereverberation suppression unit 40 suppresses the reverberation component of the harmonic component by the method exemplified in one of the first and second embodiments, and the second dereverberation suppression unit 50 includes the first and second embodiments. It is possible to suppress the reverberation component of the non-harmonic component by the method exemplified in the other of the two embodiments. Further, the intensity average unit 42 of the first reverberation suppression unit 40 calculates one moving average of the amplitude and power of the harmonic component, and the intensity average unit 52 of the second reverberation suppression unit 50 calculates the amplitude and power of the non-harmonic component. A configuration for calculating the other moving average of the above may also be adopted.

(4) Calculation of the adjustment value GH (k, m) of the harmonic component (formula (4)) using the predetermined value Gmax or the predetermined value Gmin applied to the calculations of the mathematical formulas (4) and (6). It is also possible to individually set the adjustment value GP (k, m) (formula (6)) of the inharmonic component. That is, the predetermined value Gmax and the predetermined value Gmin in the equation (4) and the predetermined value Gmax and the predetermined value Gmin in the equation (6) can be set to different numerical values. In addition, a configuration in which the predetermined value Gmax and the predetermined value Gmin are variably set individually for each of the harmonic component and the non-harmonic component in accordance with an instruction from the user is also preferable.

(5) In addition to a reverberant component in a narrow sense caused by reflection or scattering in an acoustic space, a reverberant component (resonance component) such as a musical performance of a musical instrument can be implied by the reverberant component. Specifically, the present invention can also be applied to the adjustment of the resonance component (bottle sound, box sound) of a stringed instrument such as a violin or the like by the sound board of a keyboard instrument such as a piano as in the above-described embodiments. Is possible. That is, the reverberation component of the present invention means a component that attenuates with time (attenuation component).

(6) The sound processing apparatus 100 can be realized by a server device that communicates with a terminal device such as a mobile phone. For example, the acoustic processing device 100 generates an acoustic signal SY from the acoustic signal SX received from the terminal device and transmits the acoustic signal SY to the terminal device. In the configuration in which the sound processing device 100 receives the spectrum X (k, m) of the acoustic signal SX from the terminal device, the frequency analysis unit 32 is omitted, and the spectrum Y (k, m) of the acoustic signal SY after reverberation suppression is obtained. In the configuration in which the sound processing device 100 transmits to the terminal device, the waveform generation unit 64 is omitted. Also, the spectrum YH (k, m) of the harmonic component after reverberation suppression and the spectrum YP (k, m) of the non-harmonic component are transmitted to the terminal device, and the terminal device transmits the spectrum YH (k, m) and the spectrum. A configuration that synthesizes YP (k, m) (a configuration in which the synthesis processing unit 62 and the waveform generation unit 64 are omitted from the sound processing device 100 and mounted on the terminal device) may be employed.

DESCRIPTION OF SYMBOLS 100 ... Acoustic processing device, 12 ... Signal supply device, 14 ... Sound emission device, 22 ... Arithmetic processing device, 24 ... Memory | storage device, 32 ... Frequency analysis part, 34 ... Harmonic separation part, 40 …… First dereverberation unit, 42 …… Intensity average unit, 44 …… Adjustment value calculation unit, 46 …… Suppression processing unit, 50 …… Second dereverberation unit, 52 …… Intensity average unit, 54 …… Adjustment Value calculation unit 56... Suppression processing unit 62... Synthesis processing unit 64.

Claims (4)

Harmonic separation means for separating the acoustic signal into harmonic components and non-harmonic components;
First reverberation suppressing means for suppressing a reverberation component of the harmonic component;
And a second reverberation suppression unit that suppresses the reverberation component of the non-harmonic component with a suppression strength different from the suppression strength of the reverberation component by the first reverberation suppression unit. The acoustic processing apparatus according to claim 1, wherein the suppression strength of the reverberation component by the first reverberation suppression unit exceeds the suppression strength of the reverberation component by the second reverberation suppression unit. The acoustic processing device according to claim 1, further comprising: a synthesis processing unit configured to synthesize a harmonic component after processing by the first reverberation suppression unit and a non-harmonic component after processing by the second reverberation suppression unit. The first dereverberation suppression means includes
First intensity averaging means for calculating a first average intensity by a moving average of the intensities of the harmonic components to which a first moving average coefficient is applied;
First adjustment value calculation means for calculating a first adjustment value for suppressing a reverberation component of the harmonic component according to the first average intensity;
First suppression processing means for causing the first adjustment value calculated by the first adjustment value calculation means to act on the harmonic component;
The second dereverberation suppression means includes
A second intensity averaging means for calculating a second average intensity by a moving average of the intensity of the non-harmonic component to which a second moving average coefficient different from the first moving average coefficient is applied;
Second adjustment value calculation means for calculating a second adjustment value for suppressing the reverberation component of the non-harmonic component according to the second average intensity;
The acoustic processing apparatus according to claim 1, further comprising: second suppression processing means that causes the second adjustment value calculated by the second adjustment value calculation means to act on the non-harmonic component.

Images