JP2005160038A - Processing equipment and processing method for sound signal - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a sound signal processing means capable of performing compression and expansion of each sound pressure band in combination, and capable of making the distortion of a sound after being processed small as much as possible by a simple circuit configuration. <P>SOLUTION: This sound signal processing equipment and processing method are provided with a plurality of first filters for splitting a frequency band, a plurality of sound-signal compression/expansion processing means which receive sound signals V having passed these first filters and convert their amplitudes so that the amplitudes satisfy the relation of W = sgn(V)×C×¾V/Vc¾n, a plurality of second filters which remove high-frequency components exceeding the upper limits of respective split frequency bands from the sound signals W of their outputs, and a composition device for compounding the output signals of these second filters. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to a sound signal processing apparatus and method. The present invention is a device that generates sound (sound, sound) through a speaker, such as a playback device that reproduces a sound signal from a recording medium such as a CD, MD, or magnetic tape, a television, a radio, a telephone (including a mobile phone), etc. For receiving (receiving) devices that receive electrical signals (radio waves), process demodulation, amplification, etc., and output sound signals, devices that process sounds collected by microphones such as hearing aids, and output them from speakers Applied. The present invention can also be applied to time-varying vibration or wave phenomena such as electrocardiogram processing.

  In the above-mentioned various audio and audio devices, in the process of recording, reproducing, and outputting the sound, depending on the sound pressure band (range from a small sound to a large sound) that the sound has, compression, expansion, noise reduction, Various ideas such as amplification techniques have been made, but in principle, the final output sound should be as close as possible to the original sound (the sound of the sound source). That is, small sounds are small and loud sounds are output loudly.

  For example, the sound pressure band of sound recorded on a CD is 90 dB. If there is a premise that such a wide sound pressure band sound is reproduced in a quiet environment, it is preferable to follow the above principle, and if the sound is reproduced in a quiet environment, fine nuances can be heard. .

However, it is not always possible to ensure a quiet environment. When listening to the playback sound of the playback device using headphones or earphones in a noisy place, or when using a mobile phone under the noise of a factory or the like, small sounds are difficult to hear due to ambient noise. The relationship between the loudness L and the sound intensity I heard by people under noisy conditions is given by Lochner and Burger.
L = C (I ⁿ −I ₀ ⁿ )
(Non-Patent Document 1 below). Here, C is a constant, I ₀ is a critical value determined by noise, and n is a power exponent. Sounds smaller than I ₀ cannot be heard.

  If you increase the volume (sound pressure level) to hear small sounds, loud sounds can become excessive and can cause damage to the hearing organ. Therefore, in such a case, it is necessary to process a sound signal that compresses the sound pressure band by increasing the amplification level of a small sound and decreasing the amplification level of a large sound.

  Also, hearing aids for the hearing impaired need to process sound signals similar to the above. In the sensorineural hearing loss common to elderly people, the weak sound amplification function is lost, while the strong sound is heard like a normal hearing person. Therefore, it is necessary to compress the sound pressure band by relatively increasing the amplification factor of a small sound.

  Such compression of the sound pressure band requires non-linear amplification, which causes distortion in the sound, or the sound pressure balance between the vowels and the consonants is lost, making it difficult to hear or uncomfortable. Problems always accompany. In order to reduce this, various means have been studied conventionally. For example, various methods have been proposed in which the audible sound range is divided into a plurality of frequency bands, the amplification degree of the amplifier is controlled in each band, and the high frequency components in the amplified sound signal are removed by filters in each band. (Patent Documents 1 and 2 below).

In recent years, various methods have been proposed to digitize sound signals, change the gain and frequency characteristics using digital signal processing, and process the input sound so that it falls within the narrow audible sound pressure band of the hearing-impaired deaf. (Non-Patent Document 2 below).
Japanese Patent Laid-Open No. 56-136099 JP-A-5-64297 Lochner JPA and Burger JF, Acoust.Soc.Am., Pp.1705-1707 (1961) Asano et al .: Journal of the Acoustical Society of Japan, 47 (6), pp.373-379 (1991)

  In the conventional sound pressure band compression method, it is difficult to completely remove the distortion generated in the compressed sound, and the processed sound is often difficult to hear or unnatural. Various automatic gain control (AGC) methods for changing the amplification factor in response to changes in the input sound have been proposed, but a feedback circuit, a comparison circuit, etc. are required, and the circuit configuration is complicated. There's a problem.

  On the other hand, if the entire sound pressure range is uniformly compressed, very little noise and noise are greatly amplified, and howling is a problem in hearing aids. Therefore, in the region where the sound pressure is low, it is necessary to extend the sound pressure band so as to suppress the amplification of a very weak sound and increase the amplification degree of a slightly loud sound. In other words, it is necessary to process a sound signal that combines sound pressure compression and sound pressure expansion, but it is difficult to say that a simple means for realizing both of them has been obtained.

  Therefore, the present invention provides a sound signal processing apparatus and method that can be combined with compression and expansion of the sound pressure band with a simple circuit configuration, and can reduce the distortion of the processed sound as much as possible. It is an issue.

  The first of the sound signal processing apparatus of the present invention for solving the above-described problems is a plurality of first filters that frequency-divide a frequency band of the sound signal into a plurality of divided frequency bands, and the first filter has passed through the first filter. A plurality of sound signal compression / expansion processing means (hereinafter simply referred to as “sound signal processing means”) for inputting each of the sound signals and converting their amplitudes so as to satisfy the relationship of the following formula, and the sound signal processing means A plurality of second filters for respectively removing at least high frequency components exceeding the upper limit of the divided frequency band corresponding to each of the sound signals obtained from each of the sound signals, and a synthesis device for synthesizing the output signals of the plurality of second filters It is characterized by this.

W = sgn (V) · C | V / Vc | ⁿ (1)
Where V: instantaneous value of the input sound signal to the sound signal processing means Vc: positive reference value of the input sound signal set according to the processing purpose W: instantaneous value of the output sound signal from the sound signal processing means
sgn (V): sign function (+1 for V> 0, -1 for V <0)
C: Proportional constant n: 指数 (power) index set according to the purpose of processing In addition, the first sound signal processing device includes a plurality of sound signal compression / expansion processing means in the formula (1) A function for performing sound signal amplitude conversion processing when n is set to a predetermined value smaller than 1 when | V | ≧ Vc, and n is set to a predetermined value larger than 1 when | V | <Vc. It is preferable that

  The first of the sound signal processing method of the present invention, the step of frequency dividing the input sound signal into sound signals of a plurality of different frequency bands, respectively, and the amplitude of each of the frequency-divided sound signals, The step of converting to a sound signal satisfying the relationship of the expression (1), the step of removing high frequency components generated by the conversion from the converted sound signals, and the sound signals from which the high frequency components have been removed are synthesized. And a step of performing.

  Further, in the first sound signal processing method, all or part of the sound signal frequency-divided into a plurality of frequency bands in the sound signal amplitude converting step is represented by | V in the equation (1). Sound signal amplitude conversion processing may be performed by setting n to a predetermined value smaller than 1 when | ≧ Vc, and setting n to a predetermined value larger than 1 when | V | <Vc.

  The second of the sound signal processing apparatus of the present invention inputs a plurality of first filters that frequency-divide a frequency band of the sound signal into a plurality of divided frequency bands, and a sound signal that has passed through the plurality of first filters. Then, from a plurality of sound signal processing means for converting the amplitude so as to satisfy the relationship of the following equation (2), and the sound signals respectively obtained from the sound signal processing means, the upper limit of the divided frequency band corresponding to them is set. The sound signal processing apparatus includes a plurality of second filters that respectively remove excess high-frequency components and a synthesis device that synthesizes output signals of the plurality of second filters.

W = sgn (V) · C · (| V | n -Vc n) ......... (2)
However, if | V | <Vc, W = 0.
Here, V, Vc, W, sgn (V), and C are defined in the same manner as in the above equation (1). N is a positive exponent that is set according to the processing purpose.

In the second sound signal processing apparatus, the sound signal processing means performs signal compression with n set to a predetermined value smaller than 1, and performs signal expansion with n set to a predetermined value larger than 1.
A second method of processing a sound signal according to the present invention includes a step of frequency-dividing an input sound signal into sound signals of a plurality of different frequency bands, and respective amplitudes of the frequency-divided sound signal. Are converted into sound signals satisfying the relationship of the above expression (2), high-frequency components generated by the conversion are removed from the converted sound signals, and the sound signals from which the high-frequency components are removed And a step of synthesizing.

  Here, the sound signal is a concept including an audio signal and an acoustic signal. It does not prevent the amplifier from amplifying the sound signal before the sound signal is input to the processing apparatus, or the output sound signal of the processing apparatus at the subsequent stage (before the sound signal is supplied to the speaker) and both of the above. The speaker means a device that converts a sound signal (electric signal) into sound, including headphones, earphones, and the like.

  In a preferred embodiment of the invention, the sound signal provided to the processing device is in digital form. In this case, the first filter and the second filter are digital filters, and the sound signal processing means is also a digital signal processing means. The output digital signal of the processing apparatus is converted into an analog signal and given to the speaker. However, some components of the machining apparatus can be realized by an analog signal processing circuit. For example, the first filter can be realized by an analog filter, and the output signal of the first filter can be converted into a digital signal and provided to the sound signal processing means.

  The plurality of filters and the plurality of sound signal processing means include those that perform time division processing. For example, a form in which the outputs of a plurality of first filters are sequentially given to one sound signal processing means in a time-division manner and processed in time series also means that sound signals in a plurality of frequency bands are processed separately. This corresponds to the sound signal processing means. This is the same for the plurality of first and second filters.

  As an embodiment of the first sound signal processing apparatus, there is an embodiment in which uniform sound pressure compression is performed in the entire sound pressure band. For example, in the equation (1), if Vc is set to the maximum value of V (hereinafter, the absolute value of the maximum value is referred to as Vmax) and 0 <n <1, the input sound signal normalized by Vc The instantaneous value (V / Vc) is raised to the nth power and output from the sound signal processing means. In the range where | V | is close to Vc, the amplification factor is small. However, as | V | becomes smaller, the amplification factor becomes larger and compression of the sound pressure band is performed. In practice, n is often set to 0.3 to 0.8.

  In this embodiment, sound pressure compression in the same mode (same values of n and C in equation (1)) may be performed in each frequency band divided by the first filter. Depending on the case, the value of n or C may be changed in each frequency band to process the sound signal. Alternatively, the value of Vc may be set to an intermediate sound pressure, and sound pressure compression centered on this value (amplification factor of 1 when | V | is Vc) may be performed.

  As another embodiment of the first sound signal processing apparatus, the entire sound pressure band is divided into two with the value of Vc as a boundary, and the value of n is changed on both sides thereof, so that sound pressure compression and / or different modes are performed. Or there is a form which performs sound pressure extension. For example, the value of Vc is set near the lower limit of the sound pressure range of normal conversation, the sound pressure is compressed as n <1 when | V | is larger than Vc, and n> 1 when | V | is smaller than Vc. If the sound pressure is expanded, not only a simple sound pressure compression function but also a function of blocking weak environmental noise can be provided.

  In the second sound signal processing apparatus of the present invention, sound pressure compression or expansion in the same mode (same values of n and C in equation (2)) are performed in each frequency band divided by the first filter. However, depending on the purpose of processing, the sound signal may be processed by changing the values of n and C in each frequency band.

  In both the first and second sound signal processing apparatuses described above, since compression / expansion of the sound pressure band is nonlinear amplification, the waveform is distorted and a high frequency component is generated. This high frequency component is removed by the second filter. In order to remove high frequency components by the second filter, it is necessary to limit the frequency band of the original sound signal passing through the second filter, and the first filter is provided for this purpose. Therefore, the first filter and the second filter basically correspond (overlap) in the pass frequency band.

  Since the present invention is configured as described above, first of all, it has a feature that various sound signals can be processed with a single circuit configuration. In a preferred embodiment of the present invention, since at least the sound signal processing means is digital processing, there is no difficulty in changing the parameters n, C, and Vc in the expression (1) or (2). It is also easy to change these values for each frequency band. Therefore, the present invention is not only used for the selective expansion of weak sounds in the above-mentioned hearing aids, hearing instruments, etc., but also for processing music data, pre-processing of sounds input to a speech recognition device, and various types using sound. It can be widely used for data processing of diagnostic methods.

  Secondly, according to the present invention, it is easy to remove distortion after processing such as compression / expansion of a sound signal. When analyzing the waveform after compressing the sound pressure of a pure tone of a sine waveform based on the formula (1) or (2), it is known that only odd-order (third-order or higher) harmonics are superimposed. This harmonic can be easily removed with a low-pass filter. Although the actual sound signal is a composite sound, in the present invention, the frequency component is processed by being divided into about an octave bandwidth by the first filter, so that the distortion of the processed sound signal is removed in the same manner as a pure sound. be able to.

  The constituent elements of the sound signal processing apparatus of the present invention are only a filter, a synthesizing apparatus, and a sound signal processing means. The former two are commonly used, and in the case of digital processing, the latter is performed using a general-purpose computer such as a plurality of dedicated arithmetic units having a simple arithmetic function, a digital signal processor (DSP), or a personal computer. be able to. Therefore, the present invention has a feature that it can be realized at a low cost with a relatively simple device as compared with a conventional AGC method sound signal processing device.

  FIG. 1 is a block diagram showing an electrical configuration of a sound signal processing apparatus according to an embodiment of the present invention. The sound signal input to the input terminal of this apparatus is, for example, a sound signal reproduced from a CD, MD, etc., a sound signal obtained by demodulating a signal received by a television, radio, telephone, etc. (or the received signal itself) ), A sound signal collected by a microphone in a hearing aid. In this embodiment, the sound signal is converted into a digital signal by an A / D converter (not shown) and then input to the plurality of first filters 1.

  Each of the plurality of first filters 1 is a digital filter, and divides the frequency band of the input sound signal into a plurality of different divided frequency bands. The filter 1 is basically a band-pass filter, but the filter that passes the lowest frequency band may be a low-pass filter, and the filter that passes the highest frequency band may be a high-pass filter. Each frequency component of the input sound signal passes through the filter 1 of the corresponding frequency band.

  The plurality of signal (compression / decompression) processing means 2 are input with sound signals that have passed through the corresponding first filters 1. The signal processing means 2 converts the instantaneous value V of the voltage corresponding to the sound pressure into the output voltage W based on the equation (1). Vc, n, and C in equation (1) may be set in advance as parameters. This signal processing means 2 can be processed by a DSP, but may be processed in parallel in each frequency band using a plurality of dedicated arithmetic units having a simple arithmetic function. However, when it is not always necessary to process in real time, data of each frequency band may be temporarily stored in a memory using a general-purpose computer, and these may be sequentially processed in time series.

  When the sound signal is processed according to equation (1), Vc may be set to a sound pressure level for switching the value of n. Further, when processing the sound signal by the expression (2), Vc may be set to the upper limit sound pressure level at which the output is zero. In any case, the power index n is a parameter corresponding to the degree of sound pressure compression or expansion (sound pressure compression when n <1 and sound pressure expansion when n> 1), and sound pressure band compression as n becomes smaller (larger). The degree of (extension) is large. C is a proportionality constant corresponding to the amplification factor. It is a feature of the present invention that sound pressure compression / decompression processing in any mode can be easily performed by appropriately selecting a combination of these parameters Vc, n, and C.

  As an example of the processing of the sound signal by the expression (1), when a 0 to 100 dB signal is handled, the calculation result of the output sound pressure when the sound pressure is compressed with Vc = | Vmax |, n = 0.5, and C = 1. Is shown in FIG. Here, Vmax is the maximum value of the input signal and corresponds to 100 dB. Such a sound pressure compression characteristic is suitable for a hearing aid.

  As an example of the processing of the sound signal by the expression (2), when a 0 to 100 dB signal is handled, the calculation result of the output sound pressure when the sound pressure is compressed with Vc = 0 and n = 0.5 is The result is similar to the result of FIG. Here, C is selected so that the maximum value of W is 100 dB. Such a sound pressure characteristic is suitable as a basic amplification characteristic of a hearing aid.

  The output W of the sound signal processing means 2 whose sound pressure level is band-compressed (expanded) for each divided frequency band is input to the corresponding second filter 3. The second filter 3 removes high-frequency components generated by the compression (expansion) processing in the sound signal processing means 2, and is specifically realized by a digital filter. Since the second filter 3 is intended to remove high-frequency components, a low-pass filter can be used, and the cut-off frequency corresponds (matches) to the upper limit frequency of the corresponding first filter 1.

  As an example, FIG. 3 shows a waveform after compression of a 1 kHz sine wave (pure tone) with n = 0.4 and C = 1 in equation (1). In the figure, the solid line is the original sine wave, and the broken line is the waveform after compression. FIG. 4 shows the result of FFT analysis of the waveform having such distortion. As can be seen from the figure, the distortion component is limited to odd harmonics, and it is known that the distortion component can be removed by an octave bandwidth filter and returned to the original sine wave. Although only an example of a 1 kHz sine wave is shown, the same result is obtained regardless of the frequency.

  The sound signal that has passed through the second filter 3 for each divided frequency band is synthesized by the adder 4 and becomes an output sound signal. The adder 4 can be realized by a computer or a processor. Therefore, the entire sound signal processing apparatus can be realized by a computer or a processor. The output signal of the adder 4 is converted into an analog signal by a D / A converter, and given to a speaker through a drive amplifier as necessary.

Next, a case where the present invention is applied to a hearing aid will be described. FIG. 5 shows an example of a result of investigating the hearing ability of a deaf person using a PHONAK LPP device that measures the characteristics of a deaf person and investigating the output sound pressure W required to hear the same level as a normal hearing person. (The measurement results are marked with ● in the figure). From this result, it is known that when the input sound pressure V is 50 dB, it is amplified by about 40 dB, and when V is 80 dB, it is amplified by about 20 dB, so that it can be heard like a normal hearing person. When a desirable value of n is obtained based on this data, n = 0.4. That is, it is understood that sound pressure compression may be performed so that W∽V ^0.4 .

However, if the sound pressure is compressed as it is to a region where the sound pressure is low (for example, 20 dB or less), a slight environmental noise may be excessively amplified and may be felt noisy. For this reason, when processing the sound signal according to the equation (1), it is preferable to extend the sound pressure in the sound pressure range of about 40 dB or less. In this sound pressure extension region, n = 2 to 3 is sufficient. As a combination of sound pressure compression / expansion suitable for the above-mentioned hearing-impaired person, the exponent should be switched at 40 dB (corresponding to ^1/3 of 100 dB in amplitude) of the input signal having a sound pressure level of 0 to 100 dB, and Vc = | Vmax | / 10 ³ , n _L = 0.4, n _H = 3, and C = 200 Vc are appropriate. If you write this in an expression,
| V | ≧ Vc, W = sgn (V) · 200Vc · | V / Vc | ^0.4
| V | <Vc, W = sgn (V) · 200Vc · | V / Vc | ³
It becomes. This relationship is shown by the solid line in FIG. Thus, it is a feature of the present invention that the sound pressure compression / expansion can be arbitrarily changed by setting parameters.

  On the other hand, also in the case of processing the sound signal by the equation (2), it is preferable to reduce the sound pressure in the sound pressure range of about 40 dB or less. Therefore, it is appropriate to set Vc to a value corresponding to about 30 dB of the input signal. This relationship is indicated by a broken line in FIG. As described above, the sound signal compression / expansion can be arbitrarily changed by the parameter setting even when the sound signal is processed by the equation (2).

  Next, a case where the present invention is applied to a sound quality adjusting device will be described. Using a piano as a sound source, the circuit structure shown in FIG. 1 was used to process sounds in a fairly high frequency band, and a sensuous comparison was made between the processed sound and the original sound. Six test subjects evaluated the sensory difference of “sound softness / hardness” in seven levels. 0 is no difference, 1-3 is soft after processing in numerical order (3 is large, 2 is different, 1 is small), -1 to -3 is hard after processing in numerical order It is an evaluation. Note that compression / decompression of sound was performed in four stages of n = 0.8, 0.9, 1.1, and 1.2, and Vc = | Vmax | and C = Vc. The pass frequency band was tested at 5.3 to 6.4 kHz.

  The evaluation results are shown in FIG. Reference signs A to F in the figure represent evaluator distinction. As can be seen from the figure, as the power index n in equation (1) increases from 0.8 to 1.2, the evaluation that “sound is soft” clearly increases. Of course, such sound processing can be applied as a sound quality adjusting device, but according to the knowledge of the present inventors, the identification of the sound can also be performed by applying the processing of the present invention. It has been confirmed that the property is remarkably improved. In addition, it has been confirmed that in the case of acoustic defect inspection (consultation inspection), the midpoint ratio is improved by applying the processing of the present invention.

It is a block diagram which shows the electric constitution of the processing apparatus of the sound signal which is an Example of this invention. It is a figure which shows the example of the calculation result of the output sound pressure at the time of compressing a sound pressure zone | band by the method of this invention. It is a figure which shows the example of the change of the waveform after a process when the sound pressure compression of the sine wave is carried out by the method of this invention. It is a figure which shows the result of having carried out the FFT analysis of the waveform after the process of FIG. It is a figure which shows the example of the measurement result of a hearing loss characteristic of a hard-of-hearing person. It is a figure which shows the example of the evaluation result of the sound quality before and behind a process at the time of using the method of this invention for the sound quality adjustment of a piano sound source.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 1st filter 2 Sound signal compression / decompression processing means 3 2nd filter 4 Synthesizer

Claims (5)

Input each of a plurality of first filters that divide the frequency band of the sound signal into a plurality of divided frequency bands and a sound signal that has passed through the plurality of first filters, so that the amplitude satisfies the following relationship: A plurality of sound signal compression / decompression processing means for converting to a plurality of sound signals respectively obtained from the plurality of sound signal compression / decompression processing means, and a plurality of high frequency components exceeding the upper limit of the division frequency band corresponding thereto A sound signal processing apparatus including a second filter and a synthesizing apparatus that synthesizes output signals of the plurality of second filters.
W = sgn (V) · C · | V / Vc | ⁿ (1)
Where V: instantaneous value of the input sound signal to the sound signal compression / expansion processing means Vc: positive reference value of the input sound signal set according to the processing purpose W: output from the sound signal compression / expansion processing means Instantaneous value of sound signal
sgn (V): sign function (+1 for V> 0, -1 for V <0)
C: Proportional constant n: Index to be set according to the processing purpose   The step of frequency-dividing the input sound signal into sound signals of a plurality of different frequency bands, and converting the amplitude of each of the frequency-divided sound signals into a sound signal satisfying the relationship of the expression (1). A sound signal comprising the steps of: removing each of the converted high-frequency components from the converted sound signals; and synthesizing the sound signals from which the high-frequency components have been removed. Processing method.   In the sound signal amplitude converting step, for all or part of the sound signal frequency-divided into a plurality of frequency bands, n is a predetermined value smaller than 1 when | V | ≧ Vc in the equation (1). 3. The sound signal processing method according to claim 2, wherein when | V | <Vc, n is set to a predetermined value larger than 1 to perform sound signal amplitude conversion processing. Each of the plurality of first filters that divide the frequency band of the sound signal into a plurality of divided frequency bands and the sound signal that has passed through the plurality of first filters are input, and the amplitude thereof is expressed by the following equation (2). The high-frequency components that exceed the upper limit of the corresponding divided frequency band are removed from the sound signals obtained from the plurality of sound signal compression / decompression processing means and the sound signal compression / decompression processing means for conversion to satisfy each of them. A sound signal processing apparatus comprising: a plurality of second filters that perform synthesis; and a synthesis device that synthesizes output signals of the plurality of second filters.
W = sgn (V) · C · (| V | n -Vc n) ......... (2)
However, if | V | <Vc, W = 0.
Where V: instantaneous value of the input sound signal to the sound signal compression / expansion processing means Vc: positive reference value of the input sound signal set according to the processing purpose W: output from the sound signal compression / expansion processing means Instantaneous value of sound signal
sgn (V): sign function (+1 for V> 0, -1 for V <0)
C: Proportional constant n: Positive exponent set according to processing purpose   The step of frequency-dividing the input sound signal into sound signals of a plurality of different frequency bands, and converting the amplitude of each of the frequency-divided sound signals into a sound signal satisfying the relationship of the expression (2). A sound signal comprising the steps of: removing each of the converted high-frequency components from the converted sound signals; and synthesizing the sound signals from which the high-frequency components have been removed. Processing method.