JP4826814B2 - Audio signal processing device - Google Patents

Description

  The present invention relates to an audio signal processing apparatus that allows a listener to virtually perceive bass using a missing fundamental phenomenon.

  When an audio signal is reproduced with a small speaker, the frequency characteristic of the speaker is biased toward high sounds, so that almost no deep bass is output, and the reproduced sound may be low in bass and lack of power.

  In order to solve this, a virtual bass reproduction technique using missing fundamental has been put into practical use. Missing fundamentals means that if there are multiple sounds that are an integral multiple of the fundamental frequency (eg, 200 Hz) (eg, 1000 Hz, 1200 Hz, 1400 Hz, etc.), the audience can virtually hear the fundamental frequency of 200 Hz even if the fundamental tone is not sounding. This phenomenon is based on the human auditory function of perceiving

  For example, a device disclosed in Patent Document 1 has been proposed in which a peaking filter is used to emphasize a specific frequency component of an audio signal, thereby causing a missing fundamental phenomenon.

Japanese translation of PCT publication No. 2003-511881

  The device of Patent Document 1 uses a peaking filter having peaks at two fixed frequencies, and passes the audio signal through this filter and emits the sound, thereby emphasizing the above two frequency components to the listener. Let them listen, thereby making the bass of the difference between these two frequencies emphasized by the missing fundamental. However, in the configuration of Patent Document 1, the band to be emphasized is fixed, and an unnatural feeling is unavoidable.

  Accordingly, an object of the present invention is to provide a signal processing apparatus capable of causing a missing fundamental phenomenon with natural sound quality by causing a filter to perform an operation adapted to an input signal.

The present invention relates to a plurality of peaking filters, and a low frequency detection unit that extracts a low frequency signal component from an audio signal input to the peaking filter and detects the frequency of the lowest frequency component in the extracted low frequency signal component And a virtual fundamental determination unit that determines a frequency of a natural fraction of the detected frequency as a missing fundamental virtual fundamental frequency, and a frequency of each of the virtual fundamentals for each of the plurality of peaking filters. And a coefficient calculation unit that sets a filter coefficient that emphasizes a predetermined frequency component centered on the center frequency and having a harmonic frequency of a different order as a center frequency.

The present invention further includes a smoothing unit that smoothes a change in the frequency value of the lowest frequency component detected by the low frequency detection unit, wherein the virtual fundamental tone determination unit is a natural number of the smoothed frequency. The frequency of 1 / minute is determined as the frequency of the virtual fundamental tone of the missing fundamental.

  In the present invention, the frequency of the low-frequency signal component included in the input audio signal is detected, and the center frequencies of the plurality of peaking filters are determined based on the detected frequency. The plurality of peaking filters are filters for causing a missing fundamental phenomenon to the listener. As a result, a missing fundamental phenomenon corresponding to the frequency of the low-frequency signal component included in the input audio signal can be generated, so that the listener can perceive a natural virtual fundamental tone in harmony with the original audio signal. It becomes possible to make it.

  Also, by boosting some frequency components of the original audio signal using a peaking filter, the some frequency components can be gently emphasized without impairing other frequency components, resulting in poor sound quality. Never become natural.

  Also, by smoothing the change curve of the low-frequency signal component, the filter coefficient does not change abruptly, and noise associated with the change of the filter coefficient can be prevented.

  According to the present invention, the frequency corresponding to the frequency of the low-frequency component of the input audio signal is used as a virtual fundamental, and the filtering process that causes the missing fundamental phenomenon to perceive the virtual fundamental is performed.・ A fundamental phenomenon can be generated.

An audio signal processing apparatus according to an embodiment of the present invention will be described with reference to the drawings.
The audio signal processing apparatus according to this embodiment is an apparatus that performs signal processing that causes a listener to perceive bass that is not included in an input audio signal (source signal) due to an auditory effect due to a missing fundamental phenomenon. This audio signal processing apparatus detects the lowest frequency component, which is the lowest frequency component included in the source signal, and sets the frequency of the lowest frequency component (lowest frequency) to a natural fraction of the frequency of the virtual fundamental tone ( f r ). Then, a filter process for emphasizing the frequency component of n × f r (n is an integer of 2 or more) of the source signal is performed so that a missing fundamental phenomenon occurs in the virtual fundamental tone.
As a result, when a sound signal is reproduced by a speaker device that does not have a low sound, or when an audio signal having a low sound is reproduced, a natural low sound can be added to the source signal.

  In FIG. 1, the source signal input from the signal input unit 16 is input to the delay unit 15 and the low frequency detector 11. The source signal delayed by the delay unit 15 is filtered by the filter unit 10 and output from the signal output unit 17 to the next stage. The filter unit 10 has a plurality of peaking filters connected in series as shown in FIG. In addition, a smoothing unit 12, an emphasis frequency determination unit 13, and a coefficient calculation unit 14 are connected to the low frequency detection unit 11 in series. The coefficient calculated by the coefficient calculation unit 14 is set in the filter unit 10 as a filter coefficient.

  FIG. 2A is a block diagram of the low frequency detector 11. The low frequency detection unit 11 is a processing unit that detects the lowest frequency component that is the lowest frequency component of the input audio signal. The input source signal is filtered by a high-pass filter (HPF) 21 and a low-pass filter (LPF) 22 connected in series, so that only a necessary frequency band component is extracted. The high-pass filter 21 is a subsonic filter that deletes unnecessary ultra-low frequency components (for example, 40 Hz or less). The low-pass filter 22 cuts out a frequency component (for example, 200 Hz or more) in the middle sound range or more from the input audio signal, and extracts a frequency component in the low sound range. The high-pass filter 21 and the low-pass filter 22 extract a low frequency component signal of 40 to 200 Hz from the source signal.

The low frequency component signals extracted by the filters 21 and 22 are input to the zero cross detection unit 23. Zero-crossing detector 23, as shown in the upper part of FIG. 2 (B), the sign of the low frequency component signal is an AC to detect the timing of inversion. The detection is performed for any timing from negative to positive or from positive to negative. In the example shown in the figure, inversion from negative to positive is detected. The zero-cross detector 23 outputs a detection pulse when detecting a zero-cross of the input low frequency component signal. This detection pulse is input to the counter 24 as a reset trigger.

  The counter 24 is a free-run counter that operates using a sampling frequency as a clock, and operates as shown in the lower part of FIG. That is, every time a reset trigger is input, the count value up to that time is output and the count value is reset to 0 (and the count is restarted from 0). The count value immediately before reset output from the counter 24 corresponds to the period of the low frequency component signal, that is, the lowest frequency signal component (lowest frequency component) of the source signal.

For example, when the count value of the zero-crossing interval is 480 in a system with a sampling frequency of 48 kHz, the lowest frequency that is the frequency of the lowest frequency component is calculated as follows.
Minimum frequency fo (Hz)
= Sampling frequency / Zero cross cycle = 48000/480 = 100 (Hz)

  In this embodiment, the low-frequency detector 11 is composed of the zero-cross detector 23, the counter 24, and the like. However, if there is a margin in the processing function, an amplitude spectrum is obtained by frequency analysis such as FFT and the like. The lowest frequency may be detected.

  The lowest frequency fo detected by the low frequency detector 11 shown in FIG. 2 does not necessarily change continuously and smoothly, but changes in a stepwise manner as indicated by the dotted line in FIG. There are many. When the filter coefficient is determined based on the lowest frequency that changes discontinuously in this way, the filter coefficient changes abruptly at the point where the lowest frequency changes discontinuously. May occur. Note that the filter coefficients are recalculated for each sample. Therefore, the smoothing unit 12 shapes the change waveform of the lowest frequency fo into a smooth change waveform as shown by the solid line in FIG. The change waveform is shaped using a low-pass filter or an interpolation filter.

  Further, the smoothing unit 12 may incorporate a limiter that saturates the lowest frequency fo in a predetermined frequency range (for example, 40 Hz to 200 Hz). For example, when the low frequency component signal includes ripples and the zero cross detector 23 outputs zero cross pulses at an incorrect (short) interval, the lowest frequency may be erroneously detected. By incorporating the limiter, even if there is a false detection, the emphasized frequency falls within the low frequency range, and it is possible to prevent the unexpected high frequency from being emphasized.

  The enhancement frequency determination unit 13 is a processing unit that determines a frequency to be enhanced based on the lowest frequency fo input from the smoothing unit 12. That is, the enhancement frequency determination unit 13 determines the frequency (virtual fundamental frequency) that is virtually perceived by the listener based on the missing fundamental phenomenon, and causes the missing fundamental phenomenon to occur in the virtual fundamental frequency. Determine multiple harmonics of the virtual fundamental frequency.

  The emphasis frequency determination unit 13 sets the half frequency of the lowest frequency fo input from the smoothing unit 12 to the virtual fundamental frequency fr, doubles the emphasis frequency (fo (= 2fr)), and triples (fo + fr). Four times (fo + 2fr), five times (fo + 3fr), and six times (fo + 4fr) are calculated, and the value of the enhancement frequency is calculated based on the input lowest frequency fo. The calculated enhancement frequency value is input to the coefficient calculation unit 14.

  The coefficient calculation unit 14 calculates a filter coefficient of a peaking filter whose center frequency is the input enhancement frequency and sets it in the filter unit 10.

  The peaking filter constituting the filter unit 10 can be constituted by, for example, a biquadratic IIR filter as shown in FIG. Since the biquadratic IIR filter can apply the established analog filter design method, a stable digital filter can be easily created. Hereinafter, the calculation procedure of the filter coefficient of this biquadratic IIR filter will be described.

  The transfer function of the biquadratic IIR filter shown in FIG.

  It is represented by Based on the characteristic specification of the peaking filter, each coefficient is calculated by the following formula. Where Fs: sampling frequency, fo: center frequency, gain: gain, Q: quality factor,

  Based on this, each coefficient is

  Is calculated by

  The gain gain and the quality factor Q may be appropriate fixed values according to the user's preference, but may be adaptively changed according to the detected lowest frequency value.

  FIG. 5 is a diagram illustrating a configuration of the filter unit 10. The filter unit 10 includes five peaking filters 31 to 35 connected in series. Filter coefficients of center frequencies fo, fo + fr, fo + 2fr, fo + 3fr, and fo + 4fr calculated by the coefficient calculation unit 14 are set in the peaking filters 31 to 35, respectively. This filter coefficient is set for each sampling interval of the sampling clock.

  The delay unit 15 provided in the preceding stage of the filter unit 10 has a delay time until the filter coefficient of the source signal input from the signal input unit 16 is set in the filter unit, that is, the source signal has a low frequency. This is a processing unit that delays a signal by a time until the coefficient calculation unit 14 outputs a filter coefficient for the signal after being input to the detection unit 11.

  As shown in FIG. 6, the peaking filters 31 to 35 in which the filter coefficients are set in this way, among the frequency components included in the source signal, the second harmonic, the third harmonic, the fourth harmonic, the fifth harmonic of the virtual fundamental frequency fr, The frequency component of the sixth harmonic is boosted and emphasized. The audio signal (source signal) in which the harmonics of the virtual fundamental frequency are emphasized in this way is output from the signal output unit 17 to the subsequent stage.

  In the subsequent processing circuit, the audio signal is amplified and emitted from the speaker. When the audio signal processing apparatus according to this embodiment is incorporated into an audio device, the insertion position is not limited, but it is preferable to insert the audio signal processing apparatus after the equalizer that performs normal equalization.

  As described above, when the sound signal in which the overtone is emphasized is emitted from the speaker, the listener virtually perceives the sound of the virtual fundamental sound (frequency fr) by the missing fundamental phenomenon.

  In this embodiment, the frequency fr of the virtual fundamental tone is ½ of the lowest frequency component frequency fo, but is not limited to ½ and may be 1 / (natural number). For example, it is possible to set to 1/3 or 1/4. Further, the center frequency of the filter coefficient set in the plurality of peaking filters 31 to 35 is not limited to 2 to 6 times for continuous harmonics.

  Further, the number of peaking filters provided in the filter unit 10 is not limited to five. It may be a plurality (two or more).

  In the above-described embodiment, one system of audio signal processing apparatus is shown. However, when incorporating into a multi-channel (for example, 2-channel stereo) audio apparatus, the audio signal processing apparatus may be inserted for each channel.

  Further, in a multi-channel audio apparatus, an input multi-channel signal may be converted to monaural and a filter coefficient may be calculated based on the monaural signal. That is, the processing unit system for calculating the filter coefficient may be shared by all channels. Since the low frequency signal is often localized at the center, even if the detection signal is made monaural, there is no significant adverse effect on the reproduction of the audio signal.

An audio signal processing apparatus having this configuration is shown in FIG.
FIG. 7 is a diagram illustrating an example of an audio signal processing device incorporated in an audio device of L / R2 channel. In this figure, parts common to the audio signal processing apparatus shown in FIG. This audio signal processing device is different from the audio signal processing device of FIG. 1 in that the L channel and R channel source signals are input from the left and right signal input units 16L and R, and the delay units 15L and R of different systems are respectively input. The points output from the L channel and R channel signal processing units 17L and R through the filter units 10L and R, and the L channel and R channel source signals input from the left and right signal input units 16L and R A monauralization (stereo to mono) unit 18 for inputting to monaural data is provided.

  The monauralization unit 18 adds and synthesizes the L channel source signal and the R channel source signal into a monaural signal. The monaural audio signal is input to the low frequency detector 11. The configurations and functions of the low frequency detection unit 11, the smoothing unit 12, the enhancement frequency determination unit 13, and the coefficient calculation unit 14 are the same as those of the audio signal processing apparatus shown in FIG. The filter coefficients calculated by the coefficient calculation unit 14 are set in the left and right filter units 10L and 10R.

  With the above configuration, the L channel and R channel source signals are filtered by the peaking filter having a common center frequency. As a result, the L channel signal and the R channel signal cause the same missing fundamental phenomenon to the listener, so that the virtual fundamental tone virtually perceived by the listener is the same. Therefore, the virtual fundamental tone is localized in the middle of the L channel and the R channel, that is, the center.

Block diagram of an audio signal processing apparatus according to an embodiment of the present invention Block diagram of low frequency detection unit of audio signal processing device, and diagram explaining functions of zero cross detection unit and counter Diagram showing the function of the smoothing unit The figure which shows the structural example of a peaking filter The figure which shows the composition of the filter section The figure which shows the characteristic of the filter section Block diagram of an audio signal processing apparatus that calculates a filter coefficient by monauralizing a stereo signal

Explanation of symbols

10 (L, R) filter unit 11 low frequency detection unit 12 smoothing unit 13 enhanced frequency determination unit 14 coefficient calculation unit 15 (L, R) delay unit 16 (L, R) signal input unit 17 (L, R) Signal output unit 18 Monaural unit

Claims (2)

Multiple peaking filters,
A low frequency signal detection unit that extracts a low frequency signal component from the audio signal input to the peaking filter and detects the frequency of the lowest frequency component in the extracted low frequency signal component ;
A virtual fundamental determination unit that determines a frequency of a natural fraction of the detected frequency as the frequency of the missing fundamental virtual fundamental;
A coefficient calculation unit that sets, for each of the plurality of peaking filters, a filter coefficient that uses a harmonic frequency of a different order of the frequency of the virtual fundamental tone as a center frequency and emphasizes a predetermined frequency component centered on the center frequency When,
An audio signal processing apparatus. A smoothing unit that gently smoothes a change in the value of the frequency of the lowest frequency component detected by the low frequency detection unit;
The audio signal processing device according to claim 1, wherein the virtual fundamental tone determination unit determines a frequency of a natural number of the smoothed frequency as a frequency of a missing fundamental virtual fundamental tone.

Images