JP3465638B2 - Noise removal device and audio output device - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a noise eliminating device for receiving an audio signal, and more specifically, a noise eliminating device and an audio output which are apt to be mixed with pulse noise and which are used for a car radio or the like. Regarding the device.

[0002]

2. Description of the Related Art Considering electromagnetic noise in an automobile environment, for example, a lot of pulse electromagnetic noise (sometimes referred to as pulse noise) such as ignition noise and mirror noise is generated. Since these pulse noises are mixed into the receiving antenna connected to the car radio inside the vehicle, it is common to experience pulse noise in the output audio signal. A noise eliminator for eliminating the sexual noise is used.

FIG. 9 is a block diagram of a conventional (pulse type) noise eliminator described in, for example, Japanese Patent Laid-Open No. 63-87026. In the figure, when the FM intermediate frequency signal of the FM receiver is input, the detection signal output from the FM detection circuit 1 is supplied to the delay circuit 2 formed of an LPF (low pass filter) and delayed, and the output of the delay circuit 2 is the gate circuit 3 , And is supplied to the stereo demodulation circuit 5 via the level hold circuit 4. Further, the detection signal is supplied to the HPF (high-pass filter) 6, and the noise component signal that has passed through the HPF 6 is amplified by the noise amplifier 7 and supplied to the noise detection circuit 8.

The noise detection circuit 8 is composed of a rectifying circuit for rectifying the output signal of the noise amplifier 7 and obtains a noise detection output. This noise detection output is supplied to the waveform shaping circuit 9 and the integrating circuit 10. The noise detection means 11 is configured to include the HPF 6, the noise amplifier 7, the noise detection circuit 8, the waveform shaping circuit 9, and the integration circuit 10.

The waveform shaping circuit 9 converts the noise detection output into a pulse having a pulse width of a predetermined time width and supplies it to the gate circuit 3. The gate circuit 3 is driven by the pulse supplied from the waveform shaping circuit 9 to the gate circuit 3 to be in the signal cutoff state. In the signal cutoff state, the level output circuit 4 holds the delayed output level before the signal cutoff, and the stereo demodulation circuit. 5 is supplied.

This prevents the generation of spike noise due to the sudden change in the potential of the demodulated signal due to the pulse noise. When the pulse is not supplied from the waveform shaping circuit 9, the gate circuit 3 and the level hold circuit 4 are in a signal passing state (through).

Further, the integrating circuit 10 forms an AGC loop by smoothing the noise detection output to obtain a DC signal corresponding to the noise level and giving (feeding back) the output of the integrating circuit 10 to the noise amplifier 7.

In the delay circuit 2, pulse noise is HP.
It is provided in order to compensate for the time from the supply to F6 until the gate circuit 3 is turned off. Further, in the stereo demodulation circuit 5, as shown in FIG. 10, the Lch (left channel) signal and the Rch (right channel) signal are (Lch +
Since Rch) / 2 is input in the form of balanced modulation at a frequency of 38 kHz, the Lch signal and the Rch signal can be separated and taken out by time division at 38 kHz, for example.

[0009]

FIG. 11A shows a case where pulse noise is generated in the period ta of the signal demodulated by the FM demodulator 1. The level hold circuit 4 holds the value of A in (a) for a period of ta.

An Lch waveform and a Rch waveform obtained by stereo demodulating the corrected signal are shown in FIGS.

In this case, Lch is L as shown in (b).
Lch after stereo demodulation because it is held at the ch level
Can be corrected correctly, but Rch is (c)
Since it is held in Lch as shown in, Rch and Lc
When the difference in h is large, correction noise occurs as shown in (c).

In view of this point, an object of the present invention is to provide a noise elimination device and an audio device in which the correction error is not influenced by the other channel.

[0013]

In the noise eliminator according to the present invention, audio signals of a plurality of channels are processed.
The plurality of channels included in the demodulated signal having corresponding information
Noise for each channel is set for each of the multiple channels
Part of each noise detection period is overlapped and detected from the demodulated signal.
The maximum of the signal level in each noise detection period
The value is output as the noise level for each of the multiple channels.
Noise detecting means and an audio signal included in the demodulated signal.
From the information corresponding to the number to each of the above multiple channels
Audio that demodulates and outputs the corresponding audio signal
A signal demodulating means, and the signal output from the noise detecting means
Based on the noise level of each channel, the audio
For each of the audio signals output from the audio signal demodulation means
And a correction means capable of independently correcting
To do.

In the noise eliminator according to the present invention
Displays information corresponding to multi-channel audio signals.
The noise included in the demodulated signal having
The detection sensitivity of the noise based on the pulse density of the noise
And noise detection means for detecting the noise by adjusting
From the information corresponding to the audio signal that the key signal has,
The audio signal corresponding to each of the multiple channels
Audio signal demodulating means for demodulating and outputting, and the noise
The audio signal demodulating means based on the output of the detecting means
Can be independently corrected for each of the audio signals output from
It is characterized by comprising an effective correction means.

[0015]

[0016]

[0017]

BEST MODE FOR CARRYING OUT THE INVENTION The present invention can be applied to an audio output device such as a car audio device such as a car radio, a car video device of a car-mounted television, or a video / audio device including the audio output device. An embodiment of a configuration capable of exerting a great effect on noise removal will be described.

The present invention will be specifically described with reference to the drawings showing the embodiments thereof. Embodiment 1. 1 is a block diagram of a noise elimination device according to a first embodiment of the present invention. In the figure, 1
00 is an FM demodulation means for demodulating an FM signal from a received broadcast wave or the like, 5 is a stereo demodulation means, 13 is the timing of the output signal from the noise detection means 11 and the output signal from the stereo demodulation means 5. For delaying, 14 for Lch correcting means, 15 for Rch correcting means, 16 for A / D converting means, 17 for A / D converting means 16
Is a thinning-out means for thinning out the data.

Here, for simplification, the case of demodulating audio signals of two channels of Lch and Rch will be described, but the number of channels is not limited to these two channels, and a plurality of channels may be provided. The same applies.

Next, the operation will be described. For example, in the car radio as an example of the audio output device described above, the received broadcast signal is FM-demodulated by the FM demodulator means 100 by the attached antenna or the like, and this demodulated signal (demodulated signal) is output. The A / D conversion means 16 converts the digital data. Here, the signal demodulated by the FM demodulation means 100 (F as a demodulation signal
The M demodulated signal f (t)) has the following formula: f (t) = {L (t) + R (t)} / 2 + p + {L (t) −R (t)} SIN (ω · t) / 2 (1) where L (t): left channel signal R (t): right channel signal p: 19 kHz pilot signal ω: two times the angular frequency of the pilot signal p (that is, 2π * 38 KHz rad / sec) The FM demodulated signal f (t) can be expressed by taking the frequency on the horizontal axis as shown in FIG.

As shown in FIG. 2, the FM demodulated signal f (t)
Contains signal components up to 53 kHz.

The thinning means 17 determines the sampling frequency of the output signal from the A / D converter 16 as F as described above.
The Nyquist frequency (106 KHz) of 53 kz, which is the highest frequency of the M demodulated signal f (t), is reduced to at least an extent that the Nyquist frequency is satisfied and output to the stereo demodulation means 5.

Next, the stereo demodulation means 5 will be described.
As can be understood from the equation (1), SIN (ω · t) = 1
F (t) = L (t) + p, SIN (ω · t) = −
It can be seen that when 1, f (t) = R (t) + p. In this case, p is a pilot signal component and is canceled in the latter stage of the thinning means 17, for example,
F (t) = L (t), SI when SIN (ω · t) = 1
When N (ω · t) = − 1, it is treated as f (t) = R (t) without any substantial problem.

That is, from the FM demodulated signal f (t), L
In order to extract the ch and Rch signals, the SIN (ω ·
Sampling may be performed at the timing when t) = 1 or SIN (ω · t) = − 1.

In this case, for example, when synchronizing with a pilot signal having a sampling frequency of 4n times 38 kHz (where n is an integer) as an input signal to the stereo demodulation means 5, sampling is performed at every 4n timing. When the Lch is sampled at a timing of every 4n, which is shifted by 2n from the timing of every 4n, each signal of the Rch can be taken out.

Next, Lc when pulse noise is generated
The correction period of h and Rch will be described. Lch in Figure 3
2 shows a waveform in which pulse noise as noise is mixed in the output signal of the FM demodulation means 1 when Rch and Rch are originally the same waveform. (A) of the figure shows the output from the FM demodulation means 1, and ● indicates Lch, ◯ indicates Rch, and ◆ indicates sample values at other timings.

The sample value at the timing of the above-mentioned ♦ is basically sampled to detect the position where noise is generated and the time when it is generated. For example, the sampling period is 53 KHz such as 60 KHz to 100 KHz. Higher sampling frequencies.

Further, FIG. 6B shows Rc after stereo demodulation.
The h signal, and FIG. 7C shows the Lch signal after stereo demodulation. The examples of (b) and (c) in the figure show a situation in which pulse noise in which two samples are mixed in the Lch signal and two samples are mixed in the Rch signal is generated.

Looking at the noise start position in this example, the Lch signal is t5 and the Rch signal is t4. In this case, Lch signal and Rch
Independent timing for each signal, that is, the correction unit 15 corrects the Rch signal at t2 and t8, and the correction unit 14 corrects the Lch signal at t3 and t9.

The correction in this case is performed by changing the value of t2 to t4, for example.
And the value of t6 (Rch) and the value of t3 are the values of t5 and t7 (Lch), so-called pre-hold type interpolation, or t4 and t6 when linear interpolation is performed using the values of t2 and t8. A so-called linear interpolation type interpolation or the like is adopted, which is a value (Rch) corresponding to each time of, and a value (Lch) corresponding to each time of t5 and t7 when linear interpolation is performed using the value of t3 and the value of t9. By doing so, the signal is corrected.

Next, the noise detecting means 11 as the noise detecting means will be described. In order to detect pulse noise, it is easier to detect noise by using a band that does not overlap with the FM demodulated signal. That is, as described above, in the noise detecting means 11, the sampling frequency is set to Lc.
Pulse noise is detected using signals having frequencies higher than the sampling frequencies of h and Rch (this determines the noise detection resolution).

The output signal (noise detection signal) from the noise detecting means 11 outputs an H level gate while noise is detected and an L level gate when noise is not detected.

In this example, the period during which noise is generated is from t4 to t7, and the detecting means 11 outputs the H level during this period, and the sampling frequency of the output data in the noise detecting means 11 at this time is the Lch signal. Or Rc
The sampling frequency of the h signal should be twice or more.

On the other hand, in FM stereo demodulation, FM
5 of the highest frequency in the signal output from the demodulation means 1
If the demodulation at 3 kHz can be performed reliably, the thinning means 17 is used to reduce the amount of processing in this case, so that the signal input to the stereo demodulating means 5 in the thinning means 17 is delayed. The correction means 14 and 1
In the case of 5, a delay occurs.

Therefore, the output signal from the noise detection means 11 is delayed by using the delay means 13, and the time (position) at which pulse noise occurs in the output signal of the stereo demodulation means 5 in which the delay has occurred and the noise detection means 11 are detected. Match the timing of the noise detection signal output from.

As described above, the noise is detected from the FM stereo demodulated signal, and the signal portion in which the pulse noise of the FM stereo demodulated signal is mixed (generated) is corrected left and right independently. Since this is done, the magnitude of the signal of the other channel (as viewed from each channel) does not affect the correction result, and the noise suppression effect is improved.

That is, since the correction of the signal of one channel is not affected by the other channels, and the noise is detected for the demodulated signal, the noise can be detected without losing the noise component. It can be carried out.

FIG. 4 shows the noise detecting means 11 by digital processing. The HPF 18 extracts high-frequency components from the output signal of the A / D converter 16. When pulse noise occurs, the output level of the HPF 18 increases.
Next, the absolute value circuit 19 calculates the absolute value of the output signal of the HPF 18.

Next, the thinning means 20 will be described.
Sampling frequency of A / D converter 16 fs = 304k
The output of the absolute value circuit 19 in the case of Hz is shown in FIG. Figure 5
The ⊚ mark indicates Lch, the mark indicates Rch, and the ● mark indicates other sample values. The maximum values st1, st2, and st3 during the t1, t2, and t3 periods in FIG. 5A are output as the noise levels of L1, R1, and L2.

Next, when the output of the thinning means 20 is larger than the threshold value input by the comparing means 21, it is determined that noise is generated in the Lch or Rch signal in the period at that time. In this case, since st2 is larger than the threshold value, it is determined that noise is generated in R1.

Further, the pulse noise included in the output signal of the A / D converter 16 has a widened pulse width due to the filtering process by the thinning means 17 and the stereo demodulator 5 such as an FIR filter. For this reason,
Even if almost no noise is generated in L2, noise may appear in the signal after stereo demodulation.

In such a case, the thinning means 20 is shown in FIG.
L1 and R such as t1 ′, t2 ′ and t3 ′ shown in (b)
The maximum values st1 ′, st2 ′, st of the respective periods by overlapping the periods for detecting the noises of 1 and L2
3'is output as the noise level of L1, R1, and L2.

Next, since it is st2 'and st3' that the output of the thinning means 20 is larger than the threshold value by the comparison means 21, it is judged that pulse noise is generated in R1 and L2.

As described above, it is possible to detect that there is noise by overlapping the period for detecting sample noise such as L2 which is affected by noise after stereo demodulation due to large noise in nearby samples.

By doing so, the noise detecting means is
Since the noise detection is performed such that a part of the periods are overlapped (overlapped) with each other, the noise detection can be performed more accurately.

In FIG. 5B, the detection of noise that overlaps one sample and affects a subsequent signal has been described, but sample overlap may be performed at a plurality of points. It may be determined in consideration of the pass band characteristic, the amplitude level, etc. of the filtering process by the thinning means 17 and the stereo demodulator 5 such as an FIR filter (especially the LPF included in these configurations).

When the density of pulse noise is high, correcting all pulse noise tends to destroy the original sound, resulting in an artificial and offensive sound.

In such a case, it may be possible to perform audio reproduction with less harshness by correcting large pulse noise and not correcting small pulse noise. Therefore, in order to cope with such a case, the detection sensitivity of the pulse noise may be controlled so that the pulse width of the detected pulse noise does not exceed a predetermined value (predetermined width).

FIG. 6 shows a structure capable of controlling the density of the pulse noise to be detected (hereinafter sometimes simply referred to as pulse density) (the structure shown in FIG. 6 is shown in FIG. 4). Thinning means 20 shown in FIG.
The output is the input. In this case, the noise detection means 11 is configured to include the configuration shown in FIG. 4 and the configuration shown in FIG.

In the configuration shown in FIG. 6, the LPF 23
Is smoothed by applying the output of the comparison means 21 shown in FIG.

Here, the output signal of the LPF 23 corresponds to the level of continuously generated noise such as system background noise (caused by equipment noise or thermal noise).

The adder 24 adds the output signal of the LPF 23 and the output signal of the pulse density detecting means 25. When the output of the thinning means 20 is larger than the output signal of the adder 24, the comparing means 21 produces a pulse. It is determined that there is sex noise.

The output signal of the comparing means 21 is inputted to the pulse density detecting means 25, and when the pulse density of the inputted signal is higher than a predetermined value, the output signal is increased and inputted to the adder 24. This increases the output signal of the adder 24. That is, the pulse density detecting means 25 constitutes a feedback loop between the output of the comparing means 21 and the adder 24.

FIG. 7 is a diagram showing an example of the operation of the configuration shown in FIG. The output of the adder 24 when the pulse density is small is shown by the waveform b. Further, the output of the adder 24 when the pulse density becomes high and the output of the pulse density detecting means 25 is input to the adder 24 is shown by a waveform a (note that in the figure, the waveforms a and b are straight lines for simplicity. I draw it, but
Generally becomes a waveform).

An example of the output from the thinning means 20 is a waveform c
Indicate. When the pulse density is low, the waveform b is input to the comparison means 21 as a threshold value. In the case shown in the figure, the comparison means 21 outputs an H level because pulse noise is generated in the portion of the waveform c having a value larger than the waveform b input as the threshold value. Therefore, in the example shown in the drawing, it is shown that the pulse noise is detected at two places as shown by the waveform d.

On the other hand, when the pulse density is high, the waveform a is input to the comparison means 21 as a threshold value. In the case shown in the figure, the comparison means 21 uses the waveform a input as the threshold value.
The H level is output because pulse noise is generated in the portion of the waveform c having a larger value. Therefore, in the example shown in the figure, small pulse noise as shown by the waveform e is not detected.

Next, FIG. 8 shows a block diagram of the pulse density detecting means 25. The peak value of the output signal of the comparison means 21 shown in FIG. 6 is constant, and is input to the LPF 26 to extract the DC component D0. The DC component D0 in this case is proportional to the pulse density.

The calculating means 27 sets the target value of the density of the pulse noise to be detected as D1, and the output signal as the feedback amount N.
Then, the following process is performed (A is a feedback gain, which is a fixed value here). Nc = Nc + A. (D0-D1) As a result, when Nc <0, Nc = 0. By this processing, when the pulse density is larger than the target value, the feedback amount Nc is increased. The density of the pulse noise detected by the above processing is adjusted.

As described above, when the detection sensitivity of the noise detecting means is controlled based on the density of pulse noise (noise) as the noise generation state, the audio quality due to excessive signal correction is reduced. You can prevent the decline.

In the above embodiment, the DSP (Digital) using the digital signal processing technique is used.
It is needless to say again that it can be realized by using the (1 Signal Processor), and by using such a technique, it is possible to reduce the circuit configuration and improve the adaptability of the noise removal processing that matches the nature of noise. Is.

[0061]

As described above, according to the present invention, information corresponding to audio signals of a plurality of channels can be obtained.
The noise for each of the multiple channels included in the demodulated signal
Part of each noise detection period set for each channel
Overlapping and detecting from demodulated signal, in each noise detection period
The maximum value of the signal level
Output as a bell and an audio signal based on this output
Independently complement each audio signal output from the demodulation means
Since it is possible to correct it, there is no influence of the other channel in the correction of the signal of one channel, and the noise is detected for the demodulated signal, so that the noise component is not lost. Noise detection can be performed with the pulse that is expanded during stereo demodulation.
Noise can be suppressed, enabling more accurate noise detection.
I can.

Further , according to the present invention, the noise generation state
Noise detection means based on the density of pulsed noise as
Since the detection sensitivity of the
To prevent the loss of audio quality due to
You can

[0063]

[0064]

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a noise removal device.

FIG. 2 is an explanatory diagram for explaining a spectrum of an FM demodulated signal.

FIG. 3 is an explanatory diagram for explaining a stereo demodulated signal of a signal in which pulse noise is generated.

FIG. 4 is a block diagram showing a configuration of noise detection means.

FIG. 5 is an explanatory diagram for explaining the output of the absolute value circuit.

FIG. 6 is a block diagram for explaining a configuration for controlling pulse density.

FIG. 7 is an explanatory diagram for explaining detection of pulse noise.

FIG. 8 is a block diagram showing a configuration of pulse density detection means.

FIG. 9 is a block diagram showing a configuration of a conventional noise eliminator.

FIG. 10 is an explanatory diagram for explaining stereo demodulation.

FIG. 11 is an explanatory diagram for explaining a waveform when a signal in which pulse noise is generated is stereo-demodulated by a conventional device.

[Explanation of symbols]

100 FM detection circuit, 11 noise detection means, 13
Delay means, 14 correction means, 15 correction means, 16 A
/ D converter, 17 thinning means, 18 HPF, 19
Absolute value circuit, 20 thinning-out means, 21 comparing means, 23
LPF, 24 adder, 25 pulse density detecting means, 2
6 LPF, 27 computing means.

─────────────────────────────────────────────────── ─── Continuation of front page (72) Inventor Masayuki Ishida 2-3-3 Marunouchi, Chiyoda-ku, Tokyo Mitsubishi Electric Co., Ltd. (56) References JP-A-59-182641 (JP, A) JP-A-1 −227529 (JP, A) (58) Fields investigated (Int.Cl. ⁷ , DB name) H04B 1/10

Claims (6)

(57) [Claims] 1. A plurality channels contained in the demodulated signal having information corresponding to a plurality of channels of audio signals
The noise of each
Detection from the demodulated signal by overlapping a part of the sound detection period
The maximum value of the signal level in each noise detection period
A noise detecting means for outputting as a noise level for each of the plurality of channels, and an audio signal demodulating means for demodulating and outputting an audio signal corresponding to each of the plurality of channels from information corresponding to the audio signal included in the demodulated signal. , Each of the plurality of channels output from the noise detection means
A noise removing device, comprising: a correcting unit capable of independently correcting each of the audio signals output from the audio signal demodulating unit based on the noise level of. 2. The noise generation state is detected based on the output of the noise detection means, and the detection sensitivity of the noise detection means is controlled according to the detected result. Noise eliminator. 3. An audio output device including the noise elimination device according to claim 1. 4. Supports audio signals of a plurality of channels
Noise included in the demodulated signal having information
Noise based on the pulse density of the noise contained in the signal
Detecting means for detecting the noise by adjusting the detection sensitivity of
And information corresponding to the audio signal included in the demodulated signal
Audio corresponding to each of the above multiple channels
Audio signal demodulating means for demodulating and outputting a signal, and the audio signal based on the output of the noise detecting means
Independently for each audio signal output from the demodulation means
And a correction means capable of correcting the noise.
Removal device. 5. Noise generation based on the output of the noise detecting means.
The raw state is detected, and the noise is detected according to the detected result.
The detection sensitivity of the detection means is controlled.
4. The noise eliminator according to item 4. 6. The noise elimination according to claim 4 or 5.
Audio output device including device.