JP3143563B2 - Audio processing 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 voice processing apparatus for processing voice, and more particularly, to detecting wind noise generated when a wind hits the microphone when the voice is picked up by a microphone or the like. It is for doing.

[0002]

2. Description of the Related Art When recording sound outdoors, if the wind hits a microphone for sound collection, the microphone collects the sound of the wind, causing a problem of generating wind noise. .

Since the wind noise is mainly concentrated on low frequency components, the sound signal output from the microphone is removed by passing it through a high-pass filter (hereinafter referred to as HPF) or the like. Was. However, in this case, since the audio signal is always passed through the HPF, the low frequency component of the audio signal is not recorded at all. Therefore, a circuit for detecting the occurrence of wind noise is provided, and when the wind noise is generated, the audio signal passed through the HPF is output, and when the wind noise is not generated, the audio signal not passed through the HPF is output. A circuit that can be automatically switched to record a signal has been considered.

A circuit for detecting the occurrence of wind noise as described above includes, for example, a band-pass filter (hereinafter, referred to as BPF) mainly passing wind noise, and an audio signal other than wind noise. A passing BPF is provided, and the occurrence of wind noise is detected by comparing the output levels of the respective BPFs with a comparison circuit.

That is, when the output of the wind noise BPF is greater than the output of the audio signal BPF, it is determined that wind noise is occurring, and a signal instructing wind noise removal is output. The output from the comparison circuit, the characteristics of the HPF are switched in accordance with the output from the comparison circuit, and whether or not the audio signal is passed through the HPF is switched, thereby automatically removing wind noise. It was done.

As described above, conventionally, the presence / absence of wind has been detected by comparing the level of wind noise and the level of other audio signals.

[0007]

However, in the above-described conventional detection circuit, when an audio signal having a frequency component similar to that of wind noise is input at a predetermined level, merely comparing the levels is not sufficient. If it is not possible to determine whether it is wind noise or the original audio signal, and if it is incorrectly determined that it is wind noise, the audio signal to be recorded is removed without recording. Problem.

The present invention has been made in view of the above-mentioned conventional problems, and accurately and stably generates wind noise without erroneously detecting an audio signal having a low frequency component as wind noise. It is an object of the present invention to provide a voice processing device capable of detecting a sound.

[0009]

According to the present invention, there is provided an audio processing apparatus comprising: left and right audio signal output means for converting and outputting audio generated from a sound source to audio signals; Multiplying means for inputting and multiplying the output audio signal, and detecting a wind noise by comparing a period in which the polarity of the output signal from the multiplying unit is positive and a period in which the polarity is negative at predetermined intervals. And comparison means.

[0010]

According to the above arrangement, it is possible to accurately and stably detect the occurrence of wind noise without erroneously detecting an audio signal having a low frequency component as wind noise.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below with reference to embodiments of the present invention.

FIG. 1 is a diagram showing the configuration of a first embodiment of the present invention.

In FIG. 1, reference numerals 1 and 2 denote low-pass filters (hereinafter referred to as LPFs), which extract only low-frequency components from an input audio signal. That is, since the frequency component of the wind noise is substantially biased to the low frequency region, a signal of the frequency component in which the wind noise exists can be extracted by the LPF.

Reference numeral 3 denotes a multiplier, which multiplies the signals output from the LPFs 1 and 2 and outputs a signal indicating the result of the multiplication.

Numeral 4 denotes a polarity discriminator, which discriminates the polarity of the signal output from the multiplier 3 and, if the polarity of the signal output from the multiplier 3 is positive, outputs the signal to the positive counter 5. A signal indicating “1” is supplied to the negative counter 7 as “0”.
Is output, and when the polarity of the signal output from the multiplier 3 is negative, the positive counter 5 indicates “0”.
And outputs a signal indicating “1” to the negative polarity counter 7.

Reference numeral 5 denotes a positive counter for counting positive polarity, 7 denotes a negative counter for counting negative polarity, and both the positive counter 5 and the negative counter 7 are output from the polarity discriminator 4. "1" output
Is counted, and the count value is incremented. The count values of the counters 5 and 7 are configured to be reset every predetermined period by a reset signal output from the reset circuit 6 every predetermined period. Are positive and negative within a predetermined period,
The degree is counted.

Reference numeral 8 denotes a comparison circuit which compares the count values output from the counters 5 and 7 at predetermined intervals, and when the count value of the counter 5 is larger, wind noise is generated. The counter 7 outputs a signal indicating that wind noise is occurring when the count value in the counter 7 is larger.

Next, a description will be given of a method of discriminating between wind noise and voice having a low frequency component in the present embodiment.

FIG. 2 shows the sound generated from the sound source as Lch
When sound is picked up by a stereo microphone composed of a (left) microphone and an Rch (right) microphone, the audio signal output from the Lch microphone and the Rch corresponding to the collected stereo sound. And a sound signal output from the microphone.

In FIG. 2, the sound from the sound source hits the microphones almost from the front, and during the period A in the figure, the winds hit both the microphones and the wind noise is generated. The state of the audio signal output from the microphone for Lch and the microphone for Rch is shown.

As shown in FIG. 2, when a sound source is almost in front of both microphones, almost the same sound is input to the Lch microphone and the Rch microphone, so that the wind hits both microphones. During this period, audio signals having substantially the same waveform are output from both microphones.

In this state, if the sound source is shifted to the left or right with respect to both microphones, the phase of the audio signals output from both microphones is shifted due to the movement of the sound source, and the signal waveforms are different. However, the signal waveform does not extremely differ.

The phase shift between the audio signal output from the Lch microphone and the audio signal output from the Rch microphone is caused by the mounting interval between the Lch microphone and the Rch microphone. In particular, when the sound generated from the sound source has a low frequency component, the effect of the mounting interval of the microphones on the phase difference between the sound signals output from the two microphones is extremely small. A large phase difference does not occur between the audio signal output from the microphone and the audio signal output from the Rch microphone.

On the other hand, when wind is applied to both microphones and wind noise is generated from both microphones, there is no correlation between the audio signal output from the Lch microphone and the audio signal output from the Rch microphone. And the signal waveforms are both disturbed as shown in the period A in FIG.

FIG. 3 shows a waveform of a signal obtained by multiplying the signal output from the Lch microphone and the signal output from the Rch microphone shown in FIG.

As shown in FIG. 3, when wind does not hit both microphones and no wind noise is generated, an audio signal output from the Lch microphone and an audio signal output from the Rch microphone are used. When there is almost no phase difference between the two microphones, there is almost no negative polarity in the multiplication value of the audio signals output from both microphones, and when the wind hits both microphones and wind noise is generated, Since there is no correlation between the audio signal output from the Lch microphone and the audio signal output from the Rch microphone, a negative polarity is likely to occur in the multiplication value of the audio signals output from both microphones.

Therefore, in this embodiment, L is set every predetermined period.
Audio signal output from microphone for ch and Rch
By detecting the polarity of the multiplied value with the audio signal output from the microphone for use and comparing the magnitude of the period indicating the positive state with the period indicating the negative state, wind noise or low frequency components can be detected. It distinguishes whether it has a voice.

The audio signals output from the Lch microphone and the Rch microphone both have a DC offset removed.

Next, a second embodiment of the present invention will be described with reference to FIG.

FIG. 4 is a diagram showing the configuration of the second embodiment of the present invention.

In FIG. 4, LPFs 1 and 2 extract only low-frequency components from an input audio signal. That is, since the frequency component of the wind noise is substantially biased to the low frequency region, a signal of the frequency component in which the wind noise exists can be extracted by the LPF.

A multiplier 3 multiplies the signals output from the LPFs 1 and 2 and outputs a signal indicating the result of the multiplication.

A polarity discriminator 4 discriminates the polarity of the signal output from the multiplier 3 and, when the polarity of the signal output from the multiplier 3 is positive, a positive LPF 9
A signal indicating “1” is output to the negative LPF 10 and a signal indicating “0” is output to the negative LPF 10. If the polarity of the signal output from the multiplier 3 is negative, “0” is output to the positive LPF 9. A signal indicating “1” is output to the negative-polarity LPF 10.

Reference numeral 9 denotes a positive LPF for smoothing a signal indicating positive polarity, and 10 denotes a negative LPF for smoothing a signal indicating negative polarity. The period during which the polarity of the signal output from the multiplier 3 is positive is long, and the polarity discriminator 4 outputs a positive L signal.
Many signals of "1" are supplied to the PF9, and the negative LPF1 is supplied.
When a large number of signals of “0” are supplied to 0, the positive polarity L
The level of the output signal of the PF 9 approaches “1” and the negative polarity L
The level of the output signal of the PF 10 approaches “0”, the period during which the polarity of the signal output from the multiplier 3 is negative is long, and the polarity discriminator 4 outputs the positive LPF.
9 are supplied with a large number of "0" signals, and the negative LPF 10
When many signals of “1” are supplied to the positive polarity LP
The level of the output signal of F9 approaches "0" and the negative polarity LP
The level of the output signal of F10 approaches "1".

Therefore, the magnitude of the level of the signal output from both LPFs 9 and 10 indicates how much the polarity of the signal output from the multiplier 3 is positive or negative. A signal indicating whether there is any is output.

Numeral 8 denotes a comparison circuit, which is provided at every predetermined period.
The levels of the signals output from the PFs 9 and 10 are compared, and if the level of the signal output from the LPF 9 is higher, a signal indicating that wind noise is not generated is output. If the level of the output signal is higher, a signal indicating that wind noise is occurring is output.

In the second embodiment, the period for detecting the occurrence of wind noise is determined by the time constant of the LPFs 9 and 10.

Next, a third embodiment of the present invention will be described with reference to FIG.

FIG. 5 is a diagram showing the configuration of the third embodiment of the present invention.

In FIG. 5, LPFs 1 and 2 extract only low-frequency components from an input audio signal. That is, since the frequency component of the wind noise is substantially biased to the low frequency region, a signal of the frequency component in which the wind noise exists can be extracted by the LPF.

A multiplier 3 multiplies the signals output from the LPFs 1 and 2 with each other and outputs a signal indicating the result of the multiplication.

Numeral 4 denotes a polarity discriminator for discriminating the polarity of the signal output from the multiplier 3 and a signal indicating "0" when the polarity of the signal output from the multiplier 3 is positive. Is output to the counter 11, and when the polarity of the signal output from the multiplier 3 is negative, a signal indicating “1” is output to the counter 11.

Numeral 11 denotes a counter for counting the negative polarity, counts a signal indicating "1" output from the polarity discriminator 4, and increments the count value. The count value of the counter 11 is configured to be reset every predetermined period by a reset signal output every predetermined period from the reset circuit 12, and the polarity of the signal output from the multiplier 3 is changed. The number of cases of negative polarity within a predetermined period is counted.

Numeral 8 denotes a comparison circuit which compares the count value output from the counter 11 every predetermined period with a preset reference count value output from the reference count value generator 13, and When the reference count value is larger, a signal indicating that wind noise is not generated is output. When the count value in the counter 11 is larger, it is determined that wind noise is generated. Output the signal shown.

Note that in the third embodiment, the first
In the embodiment of the present invention, there is provided one counter provided to count how long the polarity of the signal output from the multiplier 3 is positive and negative during a predetermined period. This is suitable for simplifying the circuit configuration and reducing costs.

In this embodiment, the polarity of the signal output from the multiplier 3 is determined, and when the polarity of the signal output from the multiplier 3 is positive, "1" is indicated. The same applies to the case where the polarity discriminator 4 outputs a signal to the counter 11 and outputs a signal indicating “0” to the counter 11 when the polarity of the signal output from the multiplier 3 is negative. The effect can be obtained.

Next, a fourth embodiment of the present invention will be described with reference to FIG.

FIG. 6 is a diagram showing the configuration of the fourth embodiment of the present invention.

In FIG. 6, LPFs 1 and 2 extract only low frequency components from an input audio signal. That is, since the frequency component of the wind noise is substantially biased to the low frequency region, a signal of the frequency component in which the wind noise exists can be extracted by the LPF.

A multiplier 3 multiplies the signals output from the LPFs 1 and 2 with each other and outputs a signal indicating the result of the multiplication.

Numeral 4 denotes a polarity discriminator for discriminating the polarity of the signal output from the multiplier 3 and a signal indicating "0" when the polarity of the signal output from the multiplier 3 is positive. Is output to the LPF 14, and when the polarity of the signal output from the multiplier 3 is negative, a signal indicating “1” is output to the LPF 14.
Output to

Reference numeral 14 denotes an LPF for smoothing a signal indicating negative polarity. For example, in the polarity discriminator 4, the period during which the polarity of the signal output from the multiplier 3 is positive is long. "0" from polarity discriminator 4 to LPF 14
Is supplied, the level of the output signal of the LPF 14 approaches "0", and the period during which the polarity of the signal output from the multiplier 3 is negative is long. When many signals of “1” are supplied from 4,
The level of the output signal of the LPF 14 approaches "1".

Therefore, based on the level of the signal output from the LPF 14, a signal indicating how much the polarity of the signal output from the multiplier 3 is negative when the signal is output is output. become.

Numeral 8 denotes a comparison circuit, which is provided at every predetermined period.
The level of the signal output from the reference level signal generator 15 is compared with the level of a signal having a preset reference level output from the reference level signal generator 15 to compare the level of the signal output from the PF 14. If the level is higher, a signal indicating that wind noise is not generated is output. If the level of the signal output from the LPF 14 is higher, wind noise is generated. Is output.

In the fourth embodiment, the period for detecting the occurrence of wind noise is determined by the time constant of the LPF 14.

Note that in the fourth embodiment, the second
In the embodiment, since only one LPF is required to smooth the signal indicating the period during which the polarity of the signal output from the multiplier 3 is positive and the period during which the polarity is negative, only one LPF is required. It is suitable for simplification and cost reduction.

In this embodiment, the polarity of the signal output from the multiplier 3 is determined, and when the polarity of the signal output from the multiplier 3 is positive, "1" is indicated. The same effect can be obtained by outputting a signal to the LPF 14 and using a polarity discriminator 4 that outputs a signal indicating “0” to the LPF 14 when the polarity of the signal output from the multiplier 3 is negative. What you can get.

[0058]

As described above, according to the present invention,
Since the occurrence of wind noise is detected based on the polarity of a signal obtained by multiplying the left audio signal and the right audio signal in the stereo audio signal, an audio signal having a low frequency component is erroneously detected as wind noise. Without this, it is possible to accurately and stably detect occurrence of wind noise.

[Brief description of the drawings]

FIG. 1 is a diagram showing a configuration of a first exemplary embodiment of the present invention.

FIG. 2 is a diagram illustrating a left audio signal and a right audio signal of a stereo audio signal when audio generated from a sound source is collected by a stereo microphone.

FIG. 3 is a diagram showing a waveform of a signal obtained by multiplying a signal output from the Lch microphone and a signal output from the Rch microphone shown in FIG. 2;

FIG. 4 is a diagram showing a configuration of a second exemplary embodiment of the present invention.

FIG. 5 is a diagram showing a configuration of a third exemplary embodiment of the present invention.

FIG. 6 is a diagram showing a configuration of a fourth embodiment of the present invention.

[Explanation of symbols]

 1, 2, 9, 10, 14 LPF 3 Multiplier 4 Polarity discriminator 5, 7, 11 Counter 6, 12 Reset circuit 8, 22 Comparison circuit 13 Reference count value generator 15 Reference level signal generator

────────────────────────────────────────────────── ─── Continued on the front page (58) Fields investigated (Int. Cl. ⁷ , DB name) G10L 15/00-17/00 G10L 21/02 H04R 3/00 320 INSPEC (DIALOG) JICST file (JOIS) WPI (DIALOG)

Claims (6)

(57) [Claims] 1. Left and right audio signal output means for converting and outputting audio generated from a sound source into audio signals, respectively, and multiplication for inputting and multiplying audio signals output from the left and right audio signal output means. Means for detecting wind noise by comparing a period during which the polarity of the output signal from the multiplying unit is positive and a period during which the polarity is negative at predetermined intervals. Processing equipment. 2. Left and right audio signal output means for converting and outputting audio generated from a sound source into audio signals, respectively, and multiplying by inputting and multiplying audio signals output from the left and right audio signal output means. Means for detecting, during each predetermined period, whether the polarity of the output signal from the multiplying means is positive or negative, and comparing the length of the positive period with a reference value to reduce wind noise. An audio processing device comprising: a comparing means for detecting. 3. Left and right audio signal output means for converting and outputting audio generated from a sound source into audio signals, respectively, and multiplying by inputting and multiplying audio signals output from the left and right audio signal output means. Means, detecting whether the polarity of the output signal from the multiplying means is positive or negative for each predetermined period, and comparing the length of the negative period with a reference value to reduce wind noise. An audio processing device comprising: a comparing means for detecting. 4. A left and right voice signal output means for converting voices generated from a sound source into voice signals, respectively, and multiplying by inputting and multiplying voice signals output from the right and left voice signal output means. Means, a smoothing means for smoothing and outputting the output signal from the multiplying means, and a period in which the polarity of the output signal smoothed by the smoothing means is positive and a period in which the polarity is negative are set at predetermined intervals. An audio processing apparatus comprising: a comparing unit that detects a wind noise by comparing. 5. Left and right voice signal output means for converting voices generated from a sound source into voice signals and outputting the voice signals, and multiplying by inputting and multiplying voice signals output from the right and left voice signal output means. Means, a smoothing means for smoothing and outputting the output signal from the multiplying means, and a length of a period in which the polarity of the output signal smoothed by the smoothing means is positive is defined as a reference value every predetermined period. An audio processing apparatus comprising: a comparing unit that detects a wind noise by comparing. 6. Left and right voice signal output means for converting and outputting voice generated from a sound source to voice signals, respectively, and multiplying by inputting and multiplying voice signals output from the right and left voice signal output means. Means, a smoothing means for smoothing and outputting the output signal from the multiplying means, and a length of a period in which the polarity of the output signal smoothed by the smoothing means is negative is defined as a reference value every predetermined period. An audio processing apparatus comprising: a comparing unit that detects a wind noise by comparing.