JP2008245069A - Filter device and method of calculating filter coefficient - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a technique with which an output signal of high quality can be obtained at low cost when adjusting frequency characteristics of an input signal. <P>SOLUTION: A filter device 1 includes: gain setting sections 50a-50n for setting a gain for each frequency band; transmission sections 41a-41n for generating a reference waveform that becomes a reference for each frequency band; and a filter coefficient arithmetic section 40 in which reference waveforms generated by computing elements 42a-42n are integrated with the set gain for each frequency band, filter coefficients are calculated by integrating a raised cosine window function to the relevant integration results by raised cosine arithmetic sections 43a-43n, and filter coefficients for each of frequency bands are added by an adder 44. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a filter device and a filter coefficient calculation method.

2. Description of the Related Art Conventionally, there is a filter device that changes and outputs characteristics of a predetermined frequency band with respect to an input signal. Specifically, there are an equalizer that changes the frequency characteristics of the input audio signal, an IF (Intermediate Frequency) filter in the receiving unit of the wireless device, and the like. For example, Patent Document 1 discloses an audio output device including an equalizer.
JP 2005-347876 A

  The conventional filter device described above calculates a filter coefficient using a window function based on a frequency characteristic to be adjusted, and adjusts an input signal with the calculated filter coefficient, or a plurality of input signals The frequency characteristics of the input signal are adjusted by, for example, a method of dividing into a plurality of bands using a band pass filter and adjusting the gain of each band.

  Here, a conventional filter device using a window function will be described with reference to FIGS. FIG. 8A schematically shows a configuration of a wireless device 200 including a filter device 230 as an audio equalizer. FIG. 8B schematically shows the configuration of the filter device 230 that calculates the filter coefficient using the window function and adjusts the input signal. FIG. 9A illustrates filter coefficient calculation by a conventional window function method. FIG. 9B illustrates the adjustment of the frequency characteristics by the window function method.

  As shown in FIG. 8A, the wireless device 200 includes an antenna 210, a demodulator 220, a filter device 230, an amplifier 240, and a speaker 250. The radio apparatus 200 demodulates the electrical signal acquired via the antenna 210 into an audio signal to be reproduced by the demodulator 220, adjusts the frequency characteristic of the audio signal by the filter apparatus 230, and then amplifies the signal by the amplifier 240. Play from 250.

  As shown in FIG. 8B, the filter device 230 includes an ADC 231 (Analog to Digital Converter), a filter unit 232, a DAC 233 (Digital to Analog Converter), a frequency characteristic input unit 234, a filter coefficient calculation unit 235, and the like. An input signal to the filter device 230 is converted into a digital signal by the ADC 231, frequency characteristics are adjusted by the filter unit 232, converted to an analog signal by the DAC 233, and output.

  The filter unit 232 is an FIR (Finite Impulse Response) filter such as a DSP (Digital Signal Processor), and performs FIR filter calculation based on the filter coefficient calculated by the filter coefficient calculation unit 235. The filter coefficient calculation unit 235 performs inverse Fourier transform on the frequency characteristic based on an instruction from the frequency characteristic input unit 234 that is an adjustment knob or the like that receives a setting input such as a gain for each frequency band, and obtains an impulse response. (Fast Fourier Transform), a DSP having a window function calculation unit 235b that multiplies the impulse response by a window function to obtain a filter coefficient.

  As shown in FIG. 9A, when the frequency characteristic is set by the frequency characteristic input unit 234 (set as a low-pass filter in the illustrated example), the filter coefficient calculation unit 235 performs inverse processing based on the frequency characteristic. Fourier transform is performed and the filter coefficient is calculated by multiplying by the window function. As shown in FIG. 9B, the filter coefficient calculation unit 235 performs detailed gain setting for each frequency band, and calculates a filter coefficient based on the setting.

  Next, a conventional filter device using a plurality of bandpass filters will be described with reference to FIGS. FIG. 10A schematically shows a configuration of a wireless device 300 including a filter device 330 as an audio equalizer. FIG. 10B schematically shows the configuration of a filter device 230 that uses a plurality of bandpass filters. FIG. 11 illustrates the adjustment of frequency characteristics by a plurality of bandpass filters.

  As shown in FIG. 10A, the wireless device 300 includes an antenna 310, a demodulator 320, a filter device 330, an amplifier 340, and a speaker 350. The radio apparatus 300 demodulates the electrical signal acquired via the antenna 310 into an audio signal to be reproduced by the demodulator 320, adjusts the frequency characteristic of the audio signal by the filter apparatus 330, and then amplifies the signal by the amplifier 340. Play from 350.

  As shown in FIG. 10B, the filter device 330 includes an ADC 331, a filter unit 332, a gain adjustment unit 333, an adder 334, an ADC 335, a gain input unit 336, and the like. An input signal to the filter device 330 is converted into a digital signal by the ADC 331 and divided into signals for each band by the filter unit 332, and then gain adjustment for each band is performed by the gain adjustment unit 333. Next, the signals for each band adjusted by the gain adjusting unit 333 are combined by the adder 334, converted into an analog signal by the ADC 335, and output.

  The filter unit 332 includes a plurality of bandpass filters such as filters 332a to 332e for dividing the input signal into signals for each band. The gain adjustment unit 333 includes amplifiers 333a to 333e that perform gain adjustment of a signal for each band based on an instruction from a gain input unit 336 that is an adjustment knob or the like that receives a setting input such as gain for each frequency band. Yes. The filter unit 332 and the gain adjustment unit 333 are functional units realized by, for example, one DSP.

  As illustrated in FIG. 11, the filter device 330 divides an input signal into signals for each band, performs gain adjustment, and then synthesizes the resultant to adjust frequency characteristics in an audio signal or the like.

  As described above, some conventional filter devices use the window function method and others use a plurality of bandpass filters. However, when the window function method is used, a large load is applied to the calculation of the filter coefficient because inverse Fourier transform with a large amount of calculation is performed. In particular, as shown in FIG. 9B, when gain setting is performed in accordance with the frequency resolution, the amount of calculation increases as the frequency resolution increases. For this reason, in order to update the filter coefficient in real time, a DSP having a high-speed arithmetic processing capability is required, resulting in an increase in cost.

  When a plurality of bandpass filters are realized using a DSP or the like, many operations are required when performing the band division process, which causes an increase in cost due to the use of a DPS having a high processing capability. Further, since the number of bands that can be divided is limited due to the problem of the amount of computation, it is difficult to form a steep filter like a band stop filter, and a high-quality output signal cannot be obtained.

  An object of the present invention has been made in view of the above-described problems of the prior art, and provides a technique capable of obtaining a high-quality output signal at low cost when adjusting the frequency characteristics of an input signal. It is.

In order to solve the above problem, the invention described in claim 1 is a filter device that adjusts and outputs a frequency characteristic of an input signal based on a set filter coefficient,
Gain setting means for setting a gain for each frequency band;
Generating means for generating a reference waveform serving as a reference for each frequency band;
For each frequency band, the generated reference waveform is integrated with the set gain, a raised cosine window function related to the frequency band is integrated with the integration result, and a filter coefficient is calculated. Filter coefficient calculation means for combining the filter coefficients and calculating the filter coefficient to be set;
Is provided.

According to a second aspect of the present invention, in the first aspect of the present invention, a calculation result obtained by integrating a reference waveform serving as a reference for each frequency band and a raised cosine window function related to the frequency band is calculated for each frequency band. And further comprising storage means for storing,
The filter coefficient calculation means calculates the filter coefficient for each frequency band by integrating the calculation result stored in the storage means and the set gain for each frequency band.

The invention according to claim 3 is a filter coefficient calculation method in a filter device that adjusts and outputs a frequency characteristic of an input signal based on a set filter coefficient,
A gain is set for each frequency band, a reference waveform serving as a reference is generated for each frequency band, the generated reference waveform is integrated with the set gain for each frequency band, and the integration result is A filter coefficient is calculated by integrating the raised cosine window functions related to the frequency band, and the filter coefficient for each frequency band is synthesized to calculate the filter coefficient to be set.

  According to the present invention, when adjusting the frequency characteristics of an input signal, a high-quality output signal can be obtained at low cost.

  Hereinafter, embodiments of the present invention will be described with reference to FIGS. 1 to 7, but the present invention is not limited to the following embodiments. The embodiments of the present invention show the most preferable modes of the invention, and the uses and terms of the invention are not limited thereto.

  First, the configuration of the filter device will be described with reference to FIG. FIG. 1 schematically shows a functional configuration of the filter device 1.

  As illustrated in FIG. 1, the filter device 1 includes an ADC 10, a filter unit 20, a DAC 30, a filter coefficient calculation unit 40, and a frequency characteristic input unit 50. The input signal to the filter device 1 is converted to a digital signal by the ADC 10 and the frequency characteristic is adjusted by the filter unit 20, and then converted to an analog signal by the DAC 30 and output.

  The filter unit 20 is, for example, a FIR filter such as a DSP, and performs a known FIR filter calculation based on a set filter coefficient. The filter coefficient calculation unit 40 is a DSP or the like that calculates a filter coefficient to be set in the filter unit 20 based on the setting of the frequency characteristic input unit 50 that is an adjustment knob or the like that receives a gain setting for each frequency band.

  Here, calculation of filter coefficients to be set in the filter unit 20 will be described with reference to FIGS. FIG. 2 schematically shows a configuration relating to the calculation of the filter coefficient in the filter device 1. FIG. 3A illustrates the calculation of the frequency band filter coefficient by the raised cosine window function in the time domain. FIG. 3B illustrates the calculation of the frequency band filter coefficient by the raised cosine window function in the frequency domain. FIGS. 4A and 4B show examples of synthesis of filter coefficients in a plurality of frequency bands in the frequency domain. FIG. 5 illustrates the addition of filter coefficients calculated for each frequency band in the time domain. FIG. 6 schematically shows a configuration for calculating a filter coefficient using table information.

  As illustrated in FIG. 2, the frequency characteristic input unit 50 includes gain setting units 50 a to 50 n that set a gain for each frequency band. The filter coefficient calculation unit 40 includes transmission units 41a to 41n, calculation units 42a to 42n, and raised cosine calculation units 43a to 43n for each frequency band, and adds an operation result calculated for each frequency band. 44.

  The transmitters 41a to 41n generate a reference waveform for each frequency band, for example, a sine waveform having the same frequency as the center frequency of the frequency band. The computing units 42a to 42n integrate the gain of the frequency band set by the frequency characteristic input unit 50 to the reference waveform generated for each frequency band by the transmission units 41a to 41n.

  Raised cosine computing units 43a to 43n integrate the raised cosine window function related to the frequency band to the integration results for each frequency band in computing units 42a to 42n. An expanded version of the raised cosine window function for this frequency band in the time domain is as shown in the following equation. Where α is the roll-off rate and N is the window function length.

  As shown in FIG. 3A, the raised cosine computing units 43a to 43n calculate the filter coefficient in each frequency band by multiplying the sine wave after gain adjustment in each frequency band by the raised cosine window function described above. Is done. In addition, multiplication of the sine wave of the center frequency of the frequency band and the raised cosine window function in the time domain corresponds to these convolutions in the frequency domain, as shown in FIG. It is to generate a filter coefficient having a raised cosine characteristic.

  In addition, the filter coefficient having the raised cosine characteristic as the center frequency of the frequency band has an attenuation of 50% at the half frequency between the center frequencies of each frequency band as the property of the raised cosine window function in the above formula. Have. In other words, the influence coefficients of the filter coefficients of adjacent frequency bands are equal at the intermediate frequency.

  For this reason, when the filter coefficient in the filter apparatus 1 is calculated by adding the filter coefficient for each frequency band by the adder 44, the frequency characteristic becomes smooth. Specifically, as shown in FIG. 4A, when the gains of the frequency bands having the center frequencies f1, f2, and f3 are set equal, the filter characteristics of the filter device 1 that is added and synthesized by the adder 44 Is flat between f1 and f3.

  Further, as shown in FIG. 4B, even when the gain of the intermediate frequency is set to the gain G1 and the gain of the frequency b2 is set to the gain G2, the filter added and synthesized by the adder 44. The filter characteristics of the device 1 are smoothly connected between the frequencies b1 and b2.

  The addition of the filter coefficients for each frequency band in the adder 44 is as shown in FIG. 5 in the time domain, and is set in the filter device 1 by adding the filter coefficients (k1... Kn) of each frequency band. The power filter coefficient k is calculated.

  As shown in FIG. 6, the filter coefficient calculation unit 40 calculates the filter coefficient of each frequency band, for example, in the frequency band of n, the table information stored in the storage unit 45n, which is a nonvolatile memory, and the frequency characteristic input unit 50 The gain of the frequency band set in step 1 may be integrated.

  The table information recorded in the storage unit 45n is waveform information obtained as a result of integrating the sine waveform that is the reference waveform of the frequency band described above and the raised cosine window function of the frequency band. More than a few hours are recorded. This table information is a value that can be calculated in advance at the design stage of the filter device 1, in which the frequency band for setting the gain is determined by the frequency characteristic input unit 50.

  Therefore, by calculating the filter coefficient of each frequency band using the table information stored in the storage unit for each frequency band, the amount of calculation can be reduced, and the filter coefficient can be calculated at a higher speed.

  As described above, the filter device 1 is configured to calculate the filter coefficient for adjusting the frequency characteristics of the input signal, and the gain setting units 50a to 50n for setting the gain for each frequency band and the reference for each frequency band. The reference waveforms generated by the calculators 42a to 42n are integrated with a set gain for each of the frequency bands and the transmitters 41a to 41n that generate the reference waveform to be obtained, and the raised cosine calculators 43a to 43n perform the integration. A filter coefficient calculation unit 40 that adds a raised cosine window function to the result to calculate a filter coefficient and adds a filter coefficient for each frequency band by an adder 44 is provided.

  For this reason, the filter device 1 can calculate the filter coefficient to be set only by performing the integration of the gain and the window function for each frequency band and the addition of the integration result for each frequency band. There is no need to perform high-load operation processing. Therefore, even a low-cost configuration using a DSP with low processing capability can be implemented.

  Further, since the filter device 1 can increase the number of frequency bands to be divided by reducing the calculation load, it is possible to configure a steep filter such as a band stop filter, and to improve the quality according to the user's request. An output signal can be acquired.

  Further, the filter device 1 records, for each frequency band, waveform information, which is an integration result obtained by integrating the reference waveform as a reference and the raised cosine window function related to the frequency band, for a time length equal to or greater than the number of taps of the filter. Table information is stored in advance in the storage unit, and the filter coefficient calculation unit 40 is configured to calculate the filter coefficient for each frequency band by integrating the table information and the set gain.

  For this reason, the filter device 1 can omit the calculation for integrating the reference waveform as a reference and the raised cosine window function related to the frequency band, and can calculate the filter coefficient at a higher speed.

  Note that the filter device 1 described above can be used for an IF filter in the receiving unit of the wireless device, an equalizer that is a sound signal processing device that adjusts the frequency characteristics of the sound signal, and the like. Here, a form in which the filter device 1 is used as an IF filter or an equalizer will be described with reference to FIG.

  FIG. 7A schematically illustrates a functional configuration of the wireless device 100. FIG. 7B schematically shows a functional configuration of the audio processing unit 174.

  As shown in FIG. 7A, the radio apparatus 100 includes an antenna 110, an amplifier 120, an RF filter 130 (Radio Frequency), an oscillator 140, a mixer 150, an ADC 160, a DSP 170, a DAC 180, and a speaker 190.

  In the wireless device 100, an electrical signal acquired via the antenna 110 is amplified in a desired frequency band by an amplifier 120 and an RF filter 130, frequency-converted by an oscillator 140 and a mixer 150, and then digitally converted by an ADC 160. It is converted into a signal and input to the DSP 170. The signal digitally processed by the DSP 170 is converted into an analog signal by the DAC 180 and then played back by the speaker 190.

  The DSP 170 performs an FIR calculation process using a filter coefficient to remove an unnecessary frequency band component of a signal input from the ADC 160, an oscillator 172 that converts the frequency into an audio signal, an arithmetic unit 173, and the conversion This is a configuration having an audio processing unit 174 that adjusts the frequency characteristics of a later audio signal.

  As shown in FIG. 7B, the audio processing unit 174 amplifies the adjusted signal by an audio equalizer 174a that adjusts frequency characteristics by performing FIR calculation processing using filter coefficients on the audio input signal. An amplifier 174b is included.

  The DSP 170 performs the filter coefficient calculation of the IF filter 171 and the audio processing unit 174 in the same manner as the filter device 1 described above, so that it is not necessary to perform high-load calculation processing such as FFT, and a low-performance DSP using a low processing capability is used. Even a costly configuration can be implemented. In addition, since the calculation time of the filter coefficient can be shortened by reducing the calculation load, when changing the frequency band to be tuned by the IF filter 171 or when the audio processing unit 174 is changed from the setting for conversation playback to the setting for music playback. It is possible to reduce the time lag that occurs when switching to and to achieve better audio reproduction.

  Note that the description in the above-described embodiment shows an example, and the present invention is not limited to this. The configuration and operation in the embodiment described above can be changed as appropriate.

It is a block diagram which shows typically the functional structure of a filter apparatus. It is a block diagram which shows typically the structure which concerns on the calculation of the filter coefficient in a filter apparatus. (A) is a conceptual diagram illustrating the calculation of frequency band filter coefficients by a raised cosine window function in the time domain, and (b) illustrates the calculation of frequency band filter coefficients by a raised cosine window function in the frequency domain. FIG. (A) is a graph which shows the synthesis example of the filter coefficient of a some frequency band in a frequency domain, (b) is a graph which shows the synthesis example of the filter coefficient of a some frequency band in a frequency domain. It is a conceptual diagram which illustrates addition of the filter coefficient calculated for every frequency band in a time domain. It is a block diagram which shows typically the structure which calculates a filter coefficient using table information. (A) is a block diagram schematically showing a functional configuration of a wireless device, and (b) is a block diagram schematically showing a functional configuration of an audio processing unit. (A) is a conceptual diagram which shows typically the structure of a radio | wireless apparatus, (b) is a conceptual diagram which shows typically the structure of the conventional filter apparatus. (A) is a conceptual diagram which illustrates the filter coefficient calculation by the conventional window function method, (b) is a graph which illustrates adjustment of the frequency characteristic by a window function method. (A) is a conceptual diagram which shows typically the structure of a radio | wireless apparatus, (b) is a conceptual diagram which shows typically the structure of the conventional filter apparatus. It is a conceptual diagram which illustrates adjustment of the frequency characteristic by a some band pass filter.

Explanation of symbols

1 Filter device 10 ADC
20 Filter unit 30 DAC
40 Filter coefficient computing units 41a to 41n Transmitting units 42a to 42n Computing units 43a to 43n Raised cosine computing unit 44 Adder 45n Storage unit 50 Frequency characteristic input units 50a to 50n Gain setting unit

Claims (3)

A filter device that adjusts and outputs a frequency characteristic of an input signal based on a set filter coefficient,
Gain setting means for setting a gain for each frequency band;
Generating means for generating a reference waveform serving as a reference for each frequency band;
For each frequency band, the generated reference waveform is integrated with the set gain, a raised cosine window function related to the frequency band is integrated with the integration result, and a filter coefficient is calculated. Filter coefficient calculation means for combining the filter coefficients and calculating the filter coefficient to be set;
A filter device comprising: A storage means for storing, for each frequency band, a calculation result obtained by integrating a reference waveform serving as a reference for each frequency band and a raised cosine window function related to the frequency band;
2. The filter coefficient calculation unit according to claim 1, wherein the filter coefficient calculation unit calculates a filter coefficient for each frequency band by integrating the calculation result stored in the storage unit and the set gain for each frequency band. Filter device. A method for calculating a filter coefficient in a filter device that adjusts and outputs a frequency characteristic of an input signal based on a set filter coefficient,
A gain is set for each frequency band, a reference waveform serving as a reference is generated for each frequency band, the generated reference waveform is integrated with the set gain for each frequency band, and the integration result is A filter coefficient calculation method for calculating a filter coefficient by integrating a raised cosine window function related to a frequency band, calculating a filter coefficient for each frequency band, and calculating the filter coefficient to be set.

Images