Speech signal processing device
18 claims: 5 independent, 13 dependent
- 1入力された音声信号を受け付けるとともに、当該受け付けた音声信号に基づいて、その音声信号が表す音声の大きさを表すパワーを取得するパワー取得手段と、 前記取得されたパワーの大きさを確率変数とする確率分布を取得する確率分布取得手段と、 所定の基準音声信号が前記パワー取得手段に入力された場合に当該パワー取得手段により取得されるパワーと、所定の基準パワーと、が一致している程度を表す一致度が所定の基準一致度よりも高いか否かを前記取得された確率分布に基づいて判定する一致度判定手段と、 を備える音声信号処理装置。
- 2請求項1に記載の音声信号処理装置であって、 前記パワー取得手段は、前記受け付けた音声信号を所定のフレーム間隔毎に分割し、当該分割された各部分に対して前記パワーを取得するように構成され、 前記確率分布取得手段は、前記分割された複数の部分のそれぞれに対して取得されたパワーに基づいて前記確率分布を取得するように構成された音声信号処理装置。
- 3請求項1又は請求項2に記載の音声信号処理装置であって、 前記一致度判定手段は、前記取得された確率分布と、所定の基準確率分布と、が一致している程度が高くなるほど小さくなる分布間距離値を取得し、当該取得した分布間距離値が予め設定された基準距離値よりも小さい場合に前記一致度が前記基準一致度よりも高いと判定するように構成された音声信号処理装置。
- 4請求項1乃至請求項3のいずれか一項に記載の音声信号処理装置であって、 前記パワー取得手段は、前記パワーを周波数毎に取得するように構成され、 前記確率分布取得手段は、所定の周波数の範囲毎に前記確率分布を取得するように構成された音声信号処理装置。
- 5請求項1乃至請求項4のいずれか一項に記載の音声信号処理装置であって、 前記パワー取得手段は、前記取得されたパワーを前記基準パワーに近づけるように補正するように構成され、 前記確率分布取得手段は、前記補正されたパワーに基づいて前記確率分布を取得するように構成され、 前記一致度判定手段は、前記基準音声信号が前記パワー取得手段に入力された場合に当該パワー取得手段により補正されたパワーと、前記基準パワーと、が一致している程度を表す一致度が前記基準一致度よりも高いか否かを前記取得された確率分布に基づいて判定するように構成された音声信号処理装置。
- 6請求項1乃至請求項5のいずれか一項に記載の音声信号処理装置であって、 前記確率分布取得手段は、前記確率分布を表す関数であって、前記確率変数に対して連続的に変化する関数である確率密度関数を推定することにより当該確率分布を取得するように構成された音声信号処理装置。
- 7請求項6に記載の音声信号処理装置であって、 前記確率密度関数は、前記確率変数が0から所定のピーク位置値へ向けて増加するにつれて単調に増加し、且つ、当該確率変数が当該ピーク位置値から増加するにつれて単調に減少する関数である音声信号処理装置。
- 8請求項7に記載の音声信号処理装置であって、 前記確率密度関数は、ガンマ分布を表す確率密度関数である音声信号処理装置。
- 9請求項1乃至請求項8のいずれか一項に記載の音声信号処理装置であって、 周囲の音声を集音し、当該集音した音声を表す音声信号を出力するマイクロフォンを複数備えるとともに、 前記パワー取得手段は、前記複数のマイクロフォンのそれぞれにより出力された音声信号が入力されるように構成された音声信号処理装置。
- 10入力された音声信号を受け付けるとともに、当該受け付けた音声信号に基づいて、その音声信号が表す音声の大きさを表すパワーを取得するパワー取得手段と、 前記取得されたパワーの大きさを確率変数とする確率分布を取得する確率分布取得手段と、 所定の基準音声信号が前記パワー取得手段に入力された場合に当該パワー取得手段により取得されるパワーと、所定の基準パワーと、が一致している程度を表す一致度が所定の基準一致度よりも高いか否かを前記取得された確率分布に基づいて判定する一致度判定手段と、 を備え、 前記一致度判定手段は、前記取得された確率分布と、所定の基準確率分布と、が一致している程度が高くなるほど小さくなる分布間距離値を取得し、当該取得した分布間距離値が予め設定された基準距離値よりも小さい場合に前記一致度が前記基準一致度よりも高いと判定するように構成され、 周囲の音声を集音し、当該集音した音声を表す音声信号を出力するマイクロフォンを複数備えるとともに、 前記パワー取得手段は、前記複数のマイクロフォンのそれぞれにより出力された音声信号が入力されるように構成され、 前記確率分布取得手段は、前記複数のマイクロフォンのうちの第1のマイクロフォンにより出力された音声信号に基づいて前記パワー取得手段により取得されたパワーの大きさを確率変数とする確率分布を取得するように構成され、 更に 、 前記複数のマイクロフォンのうちの第2のマイクロフォンにより出力された音声信号に基づいて前記パワー取得手段により取得されたパワーの大きさを確率変数とする確率分布を前記基準確率分布として取得する基準確率分布取得手段を備える音声信号処理装置。
- 11入力された音声信号を受け付けるとともに、当該受け付けた音声信号に基づいて、その音声信号が表す音声の大きさを表すパワーを取得するパワー取得手段と、 前記取得されたパワーの大きさを確率変数とする確率分布を取得する確率分布取得手段と、 所定の基準音声信号が前記パワー取得手段に入力された場合に当該パワー取得手段により取得されるパワーと、所定の基準パワーと、が一致している程度を表す一致度が所定の基準一致度よりも高いか否かを前記取得された確率分布に基づいて判定する一致度判定手段と、 を備え、 前記一致度判定手段は、前記取得された確率分布と、所定の基準確率分布と、が一致している程度が高くなるほど小さくなる分布間距離値を取得し、当該取得した分布間距離値が予め設定された基準距離値よりも小さい場合に前記一致度が前記基準一致度よりも高いと判定するように構成され、 周囲の音声を集音し、当該集音した音声を表す音声信号を出力するマイクロフォンを複数備えるとともに、 前記パワー取得手段は、前記複数のマイクロフォンのそれぞれにより出力された音声信号が入力されるように構成され、 前記確率分布取得手段は、前記複数のマイクロフォンのうちの1つにより出力された音声信号に基づいて前記パワー取得手段により取得されたパワーの大きさを確率変数とする確率分布を取得するように構成され、 更に 、 前記複数のマイクロフォンのそれぞれにより出力された音声信号に基づいて前記パワー取得手段により取得されたパワーの大きさを確率変数とする確率分布を前記基準確率分布として取得する基準確率分布取得手段を備える音声信号処理装置。
- 12請求項 10又は請求項11に 記載の音声信号処理装置であって、 前記確率分布取得手段は、前記複数のマイクロフォンのうちの1つにより出力された音声信号に基づいて前記パワー取得手段により取得されたパワーの大きさを確率変数とする確率分布を取得するように構成され、 前記一致度判定手段は、前記基準確率分布として、予め記憶された値を用いるように構成された音声信号処理装置。
- 13入力された音声信号を受け付けるとともに、当該受け付けた音声信号に基づいて、その音声信号が表す音声の大きさを表すパワーを取得し、 前記取得されたパワーの大きさを確率変数とする確率分布を取得し、 所定の基準音声信号が入力されることにより取得されるパワーと、所定の基準パワーと、が一致している程度を表す一致度が所定の基準一致度よりも高いか否かを前記取得された確率分布に基づいて判定する、音声信号処理方法。
- 14請求項13に記載の音声信号処理方法であって、 前記受け付けた音声信号を所定のフレーム間隔毎に分割し、当該分割された各部分に対して前記パワーを取得し、 前記分割された複数の部分のそれぞれに対して取得されたパワーに基づいて前記確率分布を取得する、音声信号処理方法。
- 15請求項13又は請求項14に記載の音声信号処理方法であって、 前記取得された確率分布と、所定の基準確率分布と、が一致している程度が高くなるほど小さくなる分布間距離値を取得し、当該取得した分布間距離値が予め設定された基準距離値よりも小さい場合に前記一致度が前記基準一致度よりも高いと判定する、音声信号処理方法。
- 16音声信号処理装置に、 入力された音声信号を受け付けるとともに、当該受け付けた音声信号に基づいて、その音声信号が表す音声の大きさを表すパワーを取得するパワー取得手段と、 前記取得されたパワーの大きさを確率変数とする確率分布を取得する確率分布取得手段と、 所定の基準音声信号が前記パワー取得手段に入力された場合に当該パワー取得手段により取得されるパワーと、所定の基準パワーと、が一致している程度を表す一致度が所定の基準一致度よりも高いか否かを前記取得された確率分布に基づいて判定する一致度判定手段と、 を実現させるための音声信号処理プログラム。
- 17請求項16に記載の音声信号処理プログラムであって、 前記パワー取得手段は、前記受け付けた音声信号を所定のフレーム間隔毎に分割し、当該分割された各部分に対して前記パワーを取得するように構成され、 前記確率分布取得手段は、前記分割された複数の部分のそれぞれに対して取得されたパワーに基づいて前記確率分布を取得するように構成された音声信号処理プログラム。
- 18請求項16又は請求項17に記載の音声信号処理プログラムであって、 前記一致度判定手段は、前記取得された確率分布と、所定の基準確率分布と、が一致している程度が高くなるほど小さくなる分布間距離値を取得し、当該取得した分布間距離値が予め設定された基準距離値よりも小さい場合に前記一致度が前記基準一致度よりも高いと判定するように構成された音声信号処理プログラム。
Independent claims18
101 paragraphs, as filed
The present invention relates to an audio signal processing device that processes an input audio signal.
There is known an audio signal processing device that includes a plurality of microphones, receives an audio signal input via each microphone, and processes the received audio signal.
As one of the audio signal processing devices of this type, the audio signal processing device described in Patent Document 1 has a power (amplification rate corresponding to the power) representing the magnitude of the sound represented by the voice signal received via a certain microphone. ) Is acquired for each frequency. Then, the audio signal processing device determines whether or not the power acquired at one time point (acquired power) and the predetermined reference power match for each frequency. When the audio signal processing device determines that the acquired power and the reference power do not match, it determines that the microphone is out of order.
<p num="0004"><patcit num="1"><text>Japanese Unexamined Patent Publication No. 2002-159098</text></patcit></p>
<p num="0005"> By the way, a plurality of microphones are arranged at different positions from each other. Therefore, the time when the voice generated at a certain position reaches each microphone is different for each microphone. In other words, at a certain point in time, each microphone is input with a voice signal based on the voice emitted at different time points.</p><p num="0006"> Therefore, for example, when the audio signal processing device is configured to use the power of the audio signal (reference audio signal) received at a certain time point via a certain microphone (reference microphone) as the reference power, the acquisition is performed. There was a risk that the audio signal that was the basis of the power and the reference audio signal would be relatively significantly different.</p><p num="0007"> In order to deal with this, it is considered preferable to configure the audio signal processing device so that the average value of the powers acquired at a plurality of time points is used as the acquired power and the reference power.</p><p num="0008"> In addition, the power of background noise fluctuates with the passage of time. Therefore, even when the audio signal processing device is configured to acquire the acquired power and the reference power based on the background noise, the value obtained by averaging the powers acquired at a plurality of time points is used as the acquired power and the reference. It is considered preferable to configure the audio signal processing device so that it can be used as power.</p><p num="0009"> However, when the audio signal processing device is configured in this way, the audio signal processing device has, for example, power P1 and power P0 that are smaller than the power P0 by a predetermined amount ΔP when the power P0 is acquired only N times. The same acquisition power P0 / N is acquired when each of the power P2s which is larger than a predetermined amount ΔP is acquired N / 2 times.</p><p num="0010"> That is, in this case, the audio signal processing device can determine with high accuracy whether or not the power acquired when the predetermined reference audio signal is input and the predetermined reference power match. There was a problem that it could not be done.</p><p num="0011"> Therefore, an object of the present invention is to determine with high accuracy whether or not the power acquired when a predetermined reference audio signal is input and the predetermined reference power match, which is the above-mentioned problem. An object of the present invention is to provide an audio signal processing device capable of solving "the inability to determine".</p>
<p num="0012"> The audio signal processing device, which is one embodiment of the present invention, is used to achieve such an object. A power acquisition means that accepts an input audio signal and acquires power that represents the magnitude of the audio represented by the audio signal based on the received audio signal. A probability distribution acquisition means for acquiring a probability distribution using the magnitude of the acquired power as a random variable, and When a predetermined reference audio signal is input to the power acquisition means, the degree of coincidence indicating the degree to which the power acquired by the power acquisition means and the predetermined reference power match is greater than the predetermined reference match degree. A concordance determination means for determining whether or not the signal is high based on the obtained probability distribution, and To be equipped.</p><p num="0013"> Further, the audio signal processing method, which is another embodiment of the present invention, is In addition to accepting the input audio signal, based on the received audio signal, the power indicating the loudness of the audio represented by the audio signal is acquired, and the power is acquired. Acquire a probability distribution with the magnitude of the acquired power as a random variable, Whether or not the degree of coincidence indicating the degree to which the power acquired by inputting the predetermined reference audio signal and the predetermined reference power match is higher than the predetermined reference coincidence is obtained. It is a method of making a judgment based on a probability distribution.</p><p num="0014"> In addition, the audio signal processing program, which is another embodiment of the present invention, For audio signal processing equipment A power acquisition means that accepts an input audio signal and acquires power that represents the magnitude of the audio represented by the audio signal based on the received audio signal. A probability distribution acquisition means for acquiring a probability distribution using the magnitude of the acquired power as a random variable, and When a predetermined reference audio signal is input to the power acquisition means, the degree of coincidence indicating the degree to which the power acquired by the power acquisition means and the predetermined reference power match is greater than the predetermined reference match degree. A concordance determination means for determining whether or not the signal is high based on the obtained probability distribution, and It is a program to realize.</p>
<p num="0015"> With the above configuration, the present invention determines with high accuracy whether or not the power acquired when a predetermined reference audio signal is input and the predetermined reference power match. can do.</p>
<figref num="1">It is a block diagram which shows the outline of the function of the audio signal processing apparatus which concerns on 1st Embodiment of this invention.</figref><figref num="2">It is a flowchart which showed the audio signal processing program executed by the CPU of the audio signal processing apparatus shown in FIG.</figref><figref num="3">It is a graph which shows the probability distribution which uses the magnitude of the power of the audio signal input through each microphone as a random variable.</figref><figref num="4">It is a graph which shows the probability distribution in the case where the probability distribution for each microphone is relatively greatly different from each other.</figref><figref num="5">It is a graph which shows the probability distribution when the probability distribution for each microphone substantially matches each other.</figref><figref num="6">It is a block diagram which shows the outline of the function of the audio signal processing apparatus which concerns on 2nd Embodiment of this invention.</figref>
The audio signal processing device, which is one embodiment of the present invention, is A power acquisition means that accepts an input audio signal and acquires power that represents the magnitude of the audio represented by the audio signal based on the received audio signal. A probability distribution acquisition means for acquiring a probability distribution using the magnitude of the acquired power as a random variable, and When a predetermined reference audio signal is input to the power acquisition means, the degree of coincidence indicating the degree to which the power acquired by the power acquisition means and the predetermined reference power match is greater than the predetermined reference match degree. A concordance determination means for determining whether or not the signal is high based on the obtained probability distribution, and To be equipped.
According to this, in the audio signal processing device, the power acquired when the reference audio signal is input and the reference power are one, based on the probability distribution using the magnitude of the acquired power as a random variable. Determine if you are doing it. As a result, it is possible to determine with high accuracy whether or not the power acquired when the reference audio signal is input and the reference power match.
In this case, the power acquisition means is configured to divide the received audio signal at predetermined frame intervals and acquire the power for each of the divided portions. It is preferable that the probability distribution acquisition means is configured to acquire the probability distribution based on the power acquired for each of the plurality of divided portions.
In this case, the matching degree determining means acquires an inter-distribution distance value that becomes smaller as the degree of coincidence between the acquired probability distribution and the predetermined reference probability distribution increases, and the acquired inter-distribution distance. It is preferable that the matching degree is determined to be higher than the reference matching degree when the value is smaller than the preset reference distance value.
In this case, the power acquisition means is configured to acquire the power for each frequency. It is preferable that the probability distribution acquisition means is configured to acquire the probability distribution for each predetermined frequency range.
By the way, the probability distribution using the magnitude of power as a random variable differs for each frequency range. Therefore, by configuring the audio signal processing device as described above, it is determined with higher accuracy whether or not the power acquired when the reference audio signal is input and the reference power match. can do.
In this case, the power acquisition means is configured to correct the acquired power so as to approach the reference power. The probability distribution acquisition means is configured to acquire the probability distribution based on the corrected power. In the matching degree determining means, when the reference audio signal is input to the power acquiring means, the matching degree indicating the degree to which the power corrected by the power acquiring means and the reference power match is described. It is preferable that it is configured to determine whether or not it is higher than the reference degree of agreement based on the obtained probability distribution.
According to this, when the reference audio signal is input to the power acquisition means, it can be determined with high accuracy whether or not the power corrected by the power acquisition means and the reference power match. .. That is, it can be determined whether or not the power is appropriately corrected by the power acquisition means.
In this case, the probability distribution acquisition means acquires the probability distribution by estimating a probability density function, which is a function representing the probability distribution and is a function that continuously changes with respect to the random variable. It is preferable to be configured.
In this case, the probability density function is a function that increases monotonically as the random variable increases from 0 toward a predetermined peak position value, and decreases monotonically as the random variable increases from the peak position value. Is preferable.
In this case, the probability density function is preferably a probability density function representing a gamma distribution.
The probability distribution with the power of background noise as a random variable is better represented by the gamma distribution. Therefore, by configuring the audio signal processing device as described above, when the audio signal representing the background noise is used as the reference audio signal, the audio signal processing device has the magnitude of the power acquired by the power acquisition means. It is possible to estimate the probability density function that well represents the probability distribution with.
In this case, the audio signal processing device is It is equipped with multiple microphones that collect ambient sound and output audio signals that represent the collected sound. It is preferable that the power acquisition means is configured so that an audio signal output by each of the plurality of microphones is input.
In this case, the probability distribution acquisition means obtains a probability distribution in which the magnitude of the power acquired by the power acquisition means is used as a random variable based on the audio signal output by the first microphone among the plurality of microphones. Configured to get, The above-mentioned audio signal processing device further A reference probability distribution that acquires a probability distribution with the magnitude of the power acquired by the power acquisition means as a random variable based on the voice signal output by the second microphone among the plurality of microphones as the reference probability distribution. It is preferable to provide an acquisition means.
Further, in another aspect of the above-mentioned audio signal processing device, The probability distribution acquisition means is configured to acquire a probability distribution using the magnitude of the power acquired by the power acquisition means as a random variable based on an audio signal output by one of the plurality of microphones. Being done The above-mentioned audio signal processing device further Provided with a reference probability distribution acquisition means for acquiring a probability distribution having the magnitude of power acquired by the power acquisition means as a random variable based on the audio signals output by each of the plurality of microphones as the reference probability distribution. Is preferable.
In this case, the probability distribution acquisition means acquires a probability distribution using the magnitude of the power acquired by the power acquisition means as a random variable based on the audio signal output by one of the plurality of microphones. Is configured as It is preferable that the matching degree determining means is configured to use a value stored in advance as the reference probability distribution.
Further, the audio signal processing method, which is another embodiment of the present invention, is In addition to accepting the input audio signal, based on the received audio signal, the power indicating the loudness of the audio represented by the audio signal is acquired, and the power is acquired. Acquire a probability distribution with the magnitude of the acquired power as a random variable, Whether or not the degree of coincidence indicating the degree of coincidence between the power acquired by inputting the predetermined reference audio signal and the predetermined reference power is higher than the predetermined reference coincidence is acquired above. It is a method of making a judgment based on a probability distribution.
In this case, the above audio signal processing method is The received audio signal is divided at predetermined frame intervals, and the power is acquired for each of the divided parts. It is preferable to acquire the probability distribution based on the power acquired for each of the plurality of divided portions.
In this case, the above audio signal processing method is An inter-distribution distance value that becomes smaller as the degree of coincidence between the acquired probability distribution and the predetermined reference probability distribution becomes higher is acquired, and the acquired inter-distribution distance value is larger than a preset reference distance value. When is smaller than the above, it is preferable to determine that the degree of agreement is higher than the degree of reference.
In addition, the audio signal processing program, which is another embodiment of the present invention, For audio signal processing equipment A power acquisition means that accepts an input audio signal and acquires power that represents the magnitude of the audio represented by the audio signal based on the received audio signal. A probability distribution acquisition means for acquiring a probability distribution using the magnitude of the acquired power as a random variable, and When a predetermined reference audio signal is input to the power acquisition means, the degree of coincidence indicating the degree to which the power acquired by the power acquisition means and the predetermined reference power match is greater than the predetermined reference match degree. A concordance determination means for determining whether or not the signal is high based on the obtained probability distribution, and It is a program to realize.
In this case, the power acquisition means is configured to divide the received audio signal at predetermined frame intervals and acquire the power for each of the divided portions. It is preferable that the probability distribution acquisition means is configured to acquire the probability distribution based on the power acquired for each of the plurality of divided portions.
In this case, the matching degree determining means acquires an inter-distribution distance value that becomes smaller as the degree of coincidence between the acquired probability distribution and the predetermined reference probability distribution increases, and the acquired inter-distribution distance. It is preferable that the matching degree is determined to be higher than the reference matching degree when the value is smaller than the preset reference distance value.
Even the invention of the audio signal processing method or the audio signal processing program having the above-described configuration can achieve the above-mentioned object of the present invention in order to have the same operation as the above-mentioned audio signal processing apparatus. it can.
Hereinafter, embodiments of an audio signal processing device, an audio signal processing method, and an audio signal processing program according to the present invention will be described with reference to FIGS. 1 to 6.
<First Embodiment> (Constitution) As shown in FIG. 1, the audio signal processing device 1 according to the first embodiment is an information processing device. The audio signal processing device 1 includes a central processing unit (CPU), a storage device (memory and hard disk drive (HDD)), and an input device (not shown).
The input device is connected to a plurality of (six in this example) microphones MC1 to MC6. Each microphone MC1 to MC6 collects ambient sound and outputs an audio signal representing the collected sound to an input device. The input device inputs the audio signals output from the microphones MC1 to MC6 and receives the input audio signals. The input device constitutes a part of the power acquisition means.
The function of the audio signal processing device 1 configured as described above is realized by executing the audio signal processing program or the like represented by the flowchart shown in FIG. 2 described later by the CPU of the audio signal processing device 1. .. Note that this function may be realized by hardware such as a logic circuit.
The audio signal processing device 1 operates in the same manner for each of the plurality of microphones MC1 to MC6. Therefore, the functions and operations of the audio signal processing device 1 with respect to the microphone MCk (where k is an integer of 1 to 6), which is any one of the plurality of microphones MC1 to MC6, will be described below.
The functions of the audio signal processing device 1 are a power acquisition unit (power acquisition means) 10, a probability distribution acquisition unit (probability distribution acquisition means, reference probability distribution acquisition means) 20, and a degree of agreement determination unit (match degree determination means). Including 30 and.
The power acquisition unit 10 receives the audio signal input from the microphone MCk. The power acquisition unit 10 converts the audio signal from the analog signal to the digital signal by performing A / D (analog-digital) conversion processing on the received audio signal.
Further, the power acquisition unit 10 divides the converted audio signal into predetermined (constant in this example) frame intervals. The power acquisition unit 10 performs the following processing on each part (frame signal) of the divided audio signal.
The power acquisition unit 10 performs predetermined pre-processing (for example, pre-emphasis processing, window-hanging processing for applying a window function, etc.) on the frame signal. Next, the power acquisition unit 10 acquires a frame signal (complex number consisting of a real part and an imaginary part) in the frequency domain by performing a fast Fourier transform (FFT) process on the frame signal.
Then, the power acquisition unit 10 powers the sum of the squared value of the real part of the acquired frame signal and the squared value of the imaginary part of the acquired frame signal for each frequency (power of the audio signal). Calculate as.
For example, when a signal having a sampling frequency of 44.1 kHz and quantized by 16 bits is used as a digital signal, when the frame interval is 10 ms and FFT processing is performed at 1024 points, it is approximately about. Power x every 43Hz<sub>i</sub>(t) is calculated. Where i is the number corresponding to the frequency (in this example, i is incremented by 1 and the frequency is incremented by about 43 Hz), and t is the frame signal on the time axis. A number representing the position (for example, a frame number for identifying a frame).
In this way, the power acquisition unit 10 divides the audio signal received via the microphone MCk at predetermined frame intervals, and applies power to each portion (frame signal) of the divided audio signal for each frequency. Calculate to.
The power acquisition unit 10 is the calculated power x<sub>i</sub>(t) is corrected so as to approach a predetermined reference power based on the following equation (1). That is, the power acquisition unit 10 has a correction coefficient f stored in advance in the storage device for each frequency.<sub>i</sub>The above calculated power x<sub>i</sub>By multiplying by (t), the power x<sub>i</sub>Correct (t).<maths num="1"><img id="000002" he="13" wi="159" file="JP5772591B2_D0001.tif" img-format="tif" img-content="drawing" /></maths>
Then, the power acquisition unit 10 determines the corrected power y.<sub>i</sub>Output (t). Here, the correction coefficient f<sub>i</sub>Is a value set for each number i (that is, frequency) corresponding to the frequency and information for identifying the microphones MC1 to MC6. Correction coefficient f<sub>i</sub>Is the calculated power x<sub>i</sub>By correcting (t), the power x<sub>i</sub>(t) is set to approach the above reference power.
The probability distribution acquisition unit 20 is the power y output by the power acquisition unit 10.<sub>i</sub>Obtain a probability distribution with the magnitude of (t) as a random variable. That is, it can be said that the probability distribution acquisition unit 20 acquires the probability distribution based on the power corrected by the power acquisition unit 10.
Specifically, the probability distribution acquisition unit 20 acquires the probability distribution when the audio signal received by the power acquisition unit 10 is an audio signal representing background noise, while the audio received by the power acquisition unit 10 When the signal is an audio signal representing audio other than background noise, the probability distribution is not acquired. In addition, in this specification, an audio signal representing background noise is also referred to as a reference audio signal.
Here, the background noise is a sound collected by the microphones MC1 to MC6 in a state where no sound source exists in the vicinity of the microphones MC1 to MC6. In this example, the probability distribution acquisition unit 20 uses the power y output by the power acquisition unit 10.<sub>i</sub>When the value obtained by averaging the magnitude of (t) over a predetermined period is smaller than the preset threshold value, it is determined that the audio signal received by the power acquisition unit 10 is an audio signal representing background noise.
First, the probability distribution acquisition unit 20 uses the power y output by the power acquisition unit 10 for each preset power range.<sub>i</sub>Of (t), the power y existing within that range<sub>i</sub>Count the number of (t) (ie, the frequency with which power within that range appears).
FIG. 3 is a graph showing a probability distribution in which the magnitude of the power of the audio signal input via the microphones MC1 to MC6 is used as a random variable. The bar graph in FIG. 3 has a length proportional to frequency.
The probability distribution acquisition unit 20 has acquired power y for each of a plurality of (100 in this example) frame signals (plural parts of the divided audio signal).<sub>i</sub>The above frequency is counted based on (t). Therefore, in this example, the probability distribution acquisition unit 20 has 51200 (= 512 × 100) powers y.<sub>i</sub>The above frequency is counted based on (t).
The power y used to count the frequency<sub>i</sub>As the number of frame signals on which (t) is based increases, the statistical variation in the counted frequency decreases. On the other hand, as the number of the frame signals increases, the possibility that the background noise includes suddenly generated noise increases. Therefore, the power y used to count the frequency<sub>i</sub>The number of frame signals on which (t) is based is preferably a number corresponding to 1 second to 10 seconds.
Next, the probability distribution acquisition unit 20 estimates a probability density function, which is a function representing the probability distribution and is a function that continuously changes with respect to the random variable, based on the counted frequency. According to this, the processing load for calculating the inter-distribution distance value, which will be described later, can be reduced. Further, the probability distribution for the range where the frequency is not counted can be easily obtained.
By the way, as shown in FIG. 3, the distribution of this frequency increases monotonically as the random variable increases from 0 toward a predetermined peak position value, and the random variable increases from the peak position value. It decreases monotonously as it goes on. This frequency distribution (that is, the probability distribution with the power of background noise as a random variable) is better represented by the gamma distribution. The gamma distribution is represented by the probability density function expressed by the following equation (2).<maths num="2"><img id="000003" he="19" wi="159" file="JP5772591B2_D0001.tif" img-format="tif" img-content="drawing" /></maths>
The probability density function P (y) expressed by the above equation (2) increases monotonically as the random variable y increases from 0 toward a predetermined peak position value, and the random variable y increases to the peak position value. It is a function that decreases monotonically as it increases from.
Here, in the above equation (2), the corrected power y<sub>i</sub>(t) is set as a random variable y. Also, Γ (λ) is a gamma function. λ is the shape parameter of the gamma distribution. σ is a measure parameter of the gamma distribution.
Specifically, the probability distribution acquisition unit 20 estimates the probability density function by determining the shape parameter λ and the scale parameter σ based on the counted frequency. In this example, the probability distribution acquisition unit 20 determines the shape parameter λ and the scale parameter σ by performing maximum likelihood estimation. As a result, the probability distribution acquisition unit 20 estimates the probability density function as shown by the solid line in FIG.
That is, the probability distribution acquisition unit 20 is configured to acquire the probability distribution by estimating a probability density function, which is a function representing the probability distribution and is a function that continuously changes with respect to the random variable. Has been done.
The matching degree determination unit 30 calculates (acquires) the inter-distribution distance value for each of any two combinations of the microphones MC1 to MC6. The inter-distribution distance value becomes smaller as the degree of agreement between the first probability distribution acquired by the probability distribution acquisition unit 20 and the second probability distribution acquired by the probability distribution acquisition unit 20 increases. Is.
The first probability distribution is a random variable of the magnitude of the power output by the power acquisition unit 10 based on the audio signal output by the first microphone that constitutes any combination of two microphones MC1 to MC6. It is a probability distribution. The second probability distribution is a probability distribution in which the magnitude of the power output by the power acquisition unit 10 is used as a random variable based on the voice signal output by the second microphone constituting the two combinations of the microphones MC1 to MC6. (Reference probability distribution).
The concordance determination unit 30 determines the inter-distribution distance value D based on the following equation (3).<sub>KL</sub>Is calculated. In this example, the inter-distribution distance value D<sub>KL</sub>Is a value also called KL distance (Kullback-Leibler divergence). Here, p (y) is a probability density function representing the first probability distribution, and q (y) is a probability density function representing the second probability distribution.<maths num="3"><img id="000004" he="19" wi="159" file="JP5772591B2_D0001.tif" img-format="tif" img-content="drawing" /></maths>
The inter-distribution distance value may be a value indicating the degree to which a plurality of probability distributions match each other, and may be a value called a Bhattacharyya distance.
Then, the matching degree determination unit 30 determines the inter-distribution distance value D calculated for each of any two combinations of the microphones MC1 to MC6.<sub>KL</sub>Get the maximum value of. Next, the matching degree determination unit 30 determines the acquired inter-distribution distance value D.<sub>KL</sub>It is determined whether or not the maximum value of is smaller than the preset reference distance value.
The concordance determination unit 30 uses the acquired inter-distribution distance value D.<sub>KL</sub>When the maximum value of is smaller than the reference distance value, it is determined that the degree of agreement is higher than the reference degree of agreement. Here, the degree of coincidence is the power output by the power acquisition unit 10 when the reference audio signal (that is, the audio signal representing the background noise) is input to the power acquisition unit 10 via the first microphone, and the reference. Indicates the degree to which the power (reference power) output by the power acquisition unit 10 when the audio signal is input to the power acquisition unit 10 via the second microphone matches.
In this way, it is said that the match degree determination unit 30 determines whether or not the match degree is higher than the preset reference match degree based on the probability distribution acquired by the probability distribution acquisition unit 20. be able to.
When the concordance degree determination unit 30 determines that the concordance degree is higher than the reference concordance degree, it outputs a normal signal indicating that the power correction by the power acquisition unit 10 is normally performed. On the other hand, when the matching degree determination unit 30 determines that the matching degree is lower than the reference matching degree, it outputs an error signal indicating that the power correction by the power acquisition unit 10 is not normally performed.
(Operation) Next, the operation of the audio signal processing device 1 configured as described above will be described. The CPU of the audio signal processing device 1 executes the audio signal processing program shown by the flowchart in FIG. 2 every time it receives an audio signal via the microphone MCk.
Specifically, when the CPU starts processing the audio signal processing program, the CPU divides the received audio signal at each frame interval in step 205, and powers each part (frame signal) of the divided audio signal. x<sub>i</sub>Calculate (t). In addition, the CPU has the calculated power x<sub>i</sub>Power y after correction by correcting (t) based on the above equation (1)<sub>i</sub>Calculate (acquire) (t) (power acquisition process).
Next, in step 210, the CPU determines whether or not the received audio signal is an audio signal representing background noise. Now, the description will be continued assuming that the received audio signal is an audio signal representing background noise. In this case, the CPU determines "Yes" and proceeds to step 215.
Then, the CPU has the power y calculated in step 205 above.<sub>i</sub>Obtain a probability distribution with the magnitude of (t) as a random variable. Specifically, the CPU has the calculated power y for each preset power range.<sub>i</sub>Power y within that range of (t)<sub>i</sub>Count the number (frequency) of (t). Then, the CPU estimates the probability density function represented by the above equation (2) by determining the shape parameter λ and the scale parameter σ of the gamma distribution based on the counted frequency. In this way, the CPU power y<sub>i</sub>Acquire a probability distribution with the magnitude of (t) as a random variable (probability distribution acquisition process).
Next, the CPU is based on the acquired probability distribution and the above equation (3), and the inter-distribution distance value D for each of any two combinations of microphones MC1 to MC6.<sub>KL</sub>Is calculated (step 220, part of the matching degree determination process).
Then, the CPU is the inter-distribution distance value D calculated for each of any two combinations of microphones MC1 to MC6.<sub>KL</sub>Get the maximum value of. Next, the CPU obtains the inter-distribution distance value D.<sub>KL</sub>It is determined whether or not the maximum value of is smaller than the above reference distance value (0.01 in this example). As a result, the CPU determines whether or not the match degree is higher than the reference match degree (step 225, part of the match degree determination step).
Now, as shown in FIG. 4, the explanation will be continued assuming an example in which the probability distributions acquired for each microphones MC1 to MC6 are relatively significantly different from each other. In this example, the inter-distribution distance value D<sub>KL</sub>The maximum value of is 4.5. Therefore, in this case, the CPU determines that the match degree is lower than the reference match degree and outputs an error signal. After that, the CPU ends the execution of this audio signal processing program.
Next, as shown in FIG. 5, the description will be continued assuming an example in which the probability distributions acquired for each microphones MC1 to MC6 are substantially the same as each other. In this example, the inter-distribution distance value D<sub>KL</sub>The maximum value of is 0.0044. Therefore, in this case, the CPU determines that the matching degree is higher than the reference matching degree, and outputs a normal signal. After that, the CPU ends the execution of this audio signal processing program.
If the received audio signal is not an audio signal representing background noise, the CPU determines "No" in step 210, and the audio signal processing program of this audio signal processing program does not execute the processes of steps 215 to 225. End the execution.
As described above, according to the first embodiment of the audio signal processing device according to the present invention, the audio signal processing device 1 is based on a probability distribution in which the magnitude of the acquired power is used as a stochastic variable. The power acquired when the signal is input through the first microphone and the power (reference power) acquired when the reference audio signal is input through the second microphone match. Judge whether or not. As a result, it is possible to determine with high accuracy whether or not the power acquired when the reference audio signal is input and the reference power match.
Further, in the first embodiment, the audio signal processing device 1 is configured to acquire a probability distribution based on the corrected power and determine whether or not the degree of matching is higher than the reference degree of matching. ing.
According to this, when the reference audio signal is input to the power acquisition unit 10, it is possible to determine with high accuracy whether or not the power corrected by the power acquisition unit 10 and the reference power match. it can. That is, it can be determined whether or not the power is appropriately corrected by the power acquisition unit 10.
Further, in the first embodiment, the audio signal processing device 1 is configured to use a probability density function representing a gamma distribution as a function representing a probability distribution in which the magnitude of power is a random variable. As a result, the audio signal processing device 1 can estimate a probability density function that well represents a probability distribution with the magnitude of power as a random variable.
<Second embodiment> Next, the audio signal processing device according to the second embodiment of the present invention will be described with reference to FIG. The functions of the audio signal processing device 100 according to the second embodiment are the power acquisition unit (power acquisition means) 110, the probability distribution acquisition unit (probability distribution acquisition means) 120, and the degree of agreement determination unit (match degree determination means) 130. And, including.
The power acquisition unit 110 receives the input audio signal and, based on the received audio signal, acquires the power indicating the magnitude of the audio represented by the audio signal. The probability distribution acquisition unit 120 acquires a probability distribution in which the magnitude of the power acquired by the power acquisition unit 110 is used as a random variable.
The matching degree determination unit 130 indicates the degree of matching between the power acquired by the power acquisition unit 110 and the predetermined reference power when a predetermined reference audio signal is input to the power acquisition unit 110. Is higher than a predetermined reference degree of agreement based on the probability distribution acquired by the probability distribution acquisition unit 120.
According to the audio signal processing device 100 according to the second embodiment, when the audio signal processing device 100 inputs a reference audio signal based on a probability distribution using the magnitude of the acquired power as a random variable. It is determined whether or not the acquired power and the reference power match. As a result, it is possible to determine with high accuracy whether or not the power acquired when the reference audio signal is input and the reference power match.
Although the present invention has been described above with reference to each of the above embodiments, the present invention is not limited to the above-described embodiments. Various modifications that can be understood by those skilled in the art can be made within the scope of the present invention in the configuration and details of the present invention.
For example, in the above embodiment, the probability distribution acquisition unit 20 may be configured to acquire the probability distribution for each predetermined frequency range. By the way, the probability distribution using the magnitude of power as a random variable differs for each frequency range. Therefore, by configuring the audio signal processing device in this way, it is determined with higher accuracy whether or not the power acquired when the reference audio signal is input and the reference power match. be able to.
In the modified example of the above embodiment, the probability distribution acquisition unit 20 may be configured to use the counted frequency as the probability distribution without estimating the probability density function. Further, the probability distribution acquisition unit 20 is configured to use a probability density function representing a gamma distribution as a function representing a probability distribution, but a probability density function representing a distribution other than the gamma distribution (for example, a normal distribution). May be configured to use.
Further, in the modification of the above embodiment, when the audio signal processing device 1 determines that the degree of coincidence is lower than the reference degree of coincidence, the correction coefficient f<sub>i</sub>May be configured to notify the user to reconfigure. Further, when the audio signal processing device 1 determines that the matching degree is lower than the reference matching degree, the correction coefficient f<sub>i</sub>May be configured to change.
Further, in the above embodiment, the audio signal processing device 1 calculates the inter-distribution distance value for all combinations of any two of the microphones MC1 to MC6, and is based on the calculated maximum value of the inter-distribution distance value. It was configured to determine whether the degree of coincidence was higher than the degree of reference.
By the way, in the modification of the above embodiment, the audio signal processing device 1 defines one of the microphones MC1 to MC6 as a reference microphone, and for a combination of the reference microphone and each of the other microphones MC1 to MC6. It may be configured to calculate the inter-distribution distance value and determine whether or not the degree of agreement is higher than the reference degree of agreement based on the maximum value of the calculated inter-distribution distance value.
Further, in the above embodiment, the audio signal processing device 1 is configured to determine whether or not the matching degree is higher than the reference matching degree based on the calculated maximum value of the inter-distribution distance value. It may be configured to determine whether or not the degree of agreement is higher than the reference degree of agreement based on the value obtained by averaging the calculated inter-distribution distance values.
Further, in the above embodiment, the audio signal processing device 1 is configured to determine whether or not the matching degree is higher than the reference matching degree based on the corrected power, but the power before the correction is used. Based on this, it may be configured to determine whether or not the degree of agreement is higher than the degree of reference. According to this, it is possible to determine whether or not the frequency characteristics of the microphones MC1 to MC6 match.
Further, in the above embodiment, the number of microphones included in the audio signal processing device 1 is 6, but it may be any number of 1 or more.
Further, in the above embodiment, the probability distribution acquisition unit 20 uses a probability distribution in which the magnitude of the power acquired by the power acquisition unit 10 based on the audio signal output by one microphone is used as a random variable as a reference probability distribution. It was configured to get.
By the way, the probability distribution acquisition unit 20 is configured to acquire a probability distribution having the magnitude of power acquired by the power acquisition unit 10 as a random variable based on audio signals output by a plurality of microphones as a reference probability distribution. It may have been done. For example, the probability distribution acquisition unit 20 may be configured to acquire a reference probability distribution based on all the power acquired for the plurality of microphones MC1 to MC6.
Further, the matching degree determination unit 30 may be configured to use a value stored in advance in the storage device as the reference probability distribution.
Further, in the above embodiment, the probability distribution acquisition unit 20 is configured to acquire the probability distribution when the voice represented by the received voice signal is background noise, but the received voice signal represents the probability distribution. It may be configured to acquire a probability distribution when the voice is a predetermined voice other than background noise.
Further, although the program is stored in the storage device in the above embodiment, it may be stored in a computer-readable recording medium. For example, the recording medium is a portable medium such as a flexible disk, an optical disk, a magneto-optical disk, and a semiconductor memory.
Further, as another modification of the above embodiment, any combination of the above-described embodiment and modification may be adopted.
The present invention enjoys the benefit of priority claim based on the patent application of Japanese Patent Application No. 2009-065443 filed in Japan on March 18, 2009, and is disclosed in the patent application. All of the content is included herein.
The present invention is applicable to an audio signal processing device or the like that includes a plurality of microphones, receives an audio signal input via each microphone, and processes the received audio signal.
1 Audio signal processor 10 Power acquisition unit 20 Probability distribution acquisition section 30 Matching degree judgment unit 100 Audio signal processor 110 Power acquisition unit 120 Probability distribution acquisition section 130 Matching degree judgment unit MC1 ~ MC6 Microphone
9 sheets
Sheet 1 Sheet 2 Sheet 3 Sheet 4 Sheet 5 Sheet 6 Sheet 7 Sheet 8 Sheet 9
Every citation, both ways
| Document | Relation | Office | Cited during |
|---|---|---|---|
| JP2002149190A | Cites | Japan | Search report |
| JP2002159086A | Cites | Japan | Search report |
| WO2007130766A2 | Cites | World Intellectual Property Organization (WIPO) | Search report |
| JP2002159086A | Cites | Japan | – |
| JP2002149190A | Cites | Japan | – |
| WO2007130766A2 | Cites | World Intellectual Property Organization (WIPO) | – |
5 members in 3 offices
Priority claims11
| Document | Office | Kind | Date |
|---|---|---|---|
| 2009065443 | Japan | A | |
| 2009065443 | Japan | A | |
| 2009065443 | Japan | – | |
| 2010001016 | Japan | W | |
| 2010001016 | Japan | W | |
| 2011504722 | Japan | A | |
| 2009065443 | – | – | – |
| JP20090065443 | – | – | – |
| JP2010001016 | – | – | – |
| JP20110504722 | – | – | – |
| WO2010JP01016 | – | – | – |
Members5
| Document | Office | Kind | |
|---|---|---|---|
| WO2010106734A1 | World Intellectual Property Organization (WIPO) | A1 | |
| US2012004916A1 | United States of America | A1 | |
| JPWO2010106734A1 | Japan | A1 | |
| US8738367B2 | United States of America | B2 | |
| JP5772591B2This record | Japan | B2 |
10 legal events, as the office reported them to INPADOC
Over the term
Point at a mark for the eventEvents
| Event | Code | |
|---|---|---|
| Certificate of patent or registration of utility modelJAPANESE INTERMEDIATE CODE: R150R150 | R150 | |
| First payment of annual fees (during grant procedure)JAPANESE INTERMEDIATE CODE: A61A61 | A61 | |
| Written decision to grant a patent or to grant a registration (utility model)JAPANESE INTERMEDIATE CODE: A01A01 | A01 | |
| Decision of grant or rejection writtenTRDD | TRDD | |
| Written amendmentJAPANESE INTERMEDIATE CODE: A523A521 | A521 | |
| Notification of reasons for refusalJAPANESE INTERMEDIATE CODE: A131A131 | A131 | |
| Written amendmentJAPANESE INTERMEDIATE CODE: A523A521 | A521 | |
| Notification of reasons for refusalJAPANESE INTERMEDIATE CODE: A131A131 | A131 | |
| Written request for application examinationJAPANESE INTERMEDIATE CODE: A621A621 | A621 | |
| Notification of extinguishment of power of attorneyJAPANESE INTERMEDIATE CODE: A7427RD07 | RD07 |
Numbers
- Publication
- 5772591
- Publication, DOCDB
- 5772591
- Publication, EPODOC
- JP5772591B
- Application
- 2011504722
- Application, DOCDB
- 2011504722
- Application, EPODOC
- JP20110504722
Titles2
- Japanese
- 音声信号処理装置
- English
- Audio signal processor
Classification
- CPC, 1
- H04R5/04
- IPC, 2
- H04R3 00
- H04R29 00
