JP2022011890A - Noise update circuit - Google Patents

Abstract

To provide a noise update circuit that estimates noise components accurately, and is used in a noise reduction circuit.SOLUTION: A noise update circuit has: a first update unit 111 that calculates a post-update noise spectrum when a frame to be processed is a frame including only noise components; and a second update unit 112 that calculates a post-update noise spectrum when the frame to be processed is a frame including voice components.EFFECT: Even when the frames including voice components are continued one after another as the frames to be processed, fluctuation of noise can be followed, and noise components can be accurately estimated.SELECTED DRAWING: Figure 2

Description

The present invention relates to a noise update circuit, for example, a noise update circuit that can be used in a noise reduction circuit incorporated in a radio that transmits and receives high frequency signals.

A spectral subtraction method is known as a method for suppressing a noise component contained in an audio signal (see, for example, Patent Document 1 and Non-Patent Document 1).

Japanese Unexamined Patent Publication No. 4-238399

P. Scalart and J. Vieira Filho "Speech Enhancement Based on a Priori Signal to Noise Estimation", IEEE International Conference on. Acoustics, Speech, Signal Progressing, Atlanta, GA, USA, vol. 2, pp. 629-632, 1996

By the way, in order to properly apply the spectral subtraction method, it is important to accurately estimate the noise component and subtract it from the input audio signal.

Therefore, an object of the present invention is to provide a noise update circuit capable of accurately estimating a noise component.

In order to solve the above problem, in the invention according to claim 1, when the frame to be processed is a frame having only a noise component, a signal corresponding to the amplitude spectrum of the frame to be processed is input to the input signal spectrum Yi ( As f), the first update unit that calculates the updated noise spectrum Ni (f) for each frequency f by a process including averaging (however, i: the number of orders representing the time series order), and the processing target. When the frame of is a frame including a voice component, the frame has a second update unit for calculating the updated noise spectrum Ni (f) for each frequency f by a process including averaging, and the voice component is used. When calculating the updated noise spectrum Ni (f) when the frame includes the frame, the averaging time when calculating the updated noise spectrum Ni (f) is when calculating the updated noise spectrum Ni (f) when the frame contains only the noise component. It is a noise update circuit characterized by having a longer time than the averaging time of.

According to the second aspect of the present invention, when the frame to be processed is a frame containing only noise components, the signal corresponding to the amplitude spectrum of the frame to be processed is set as the input signal spectrum Yi (f) for each frequency f. The first update unit that calculates the updated noise spectrum Ni (f) by processing including averaging (however, i: the number of orders representing the time series order), and the frame to be processed is a frame containing an audio component. In the case where the frame includes a second update unit for calculating the updated noise spectrum Ni (f) for each frequency f by a process including averaging, and the frame includes the voice component, the update is performed. An average of the input signal spectrum Yi (f), the past input signal spectrum Yi-j (f), and the past noise spectrum Ni-k (f) when calculating the later noise spectrum Ni (f). (However, j: an integer of 1 or more representing the degree of deviation from the order number i in the time series, k: an integer of 1 or more representing the degree of deviation from the order number i in the time series), said. The input signal spectrum Yi (f), the past input signal spectrum Yi-j (f), and the past when calculating the updated noise spectrum Ni (f) in the case of a frame containing only noise components. It is a noise update circuit characterized in that it is longer than the average averaging time using the noise spectrum Ni-k (f).

According to the third aspect of the present invention, when the frame to be processed is a frame containing only noise components, the signal corresponding to the amplitude spectrum of the frame to be processed is set as the input signal spectrum Yi (f) for each frequency f. The first update unit that calculates the updated noise spectrum Ni (f) by processing including averaging (however, i: the number of orders representing the time-series order), and the frame to be processed is a frame containing an audio component. In the case where the frame includes a second update unit for calculating the updated noise spectrum Ni (f) for each frequency f by a process including averaging, and the frame includes the voice component, the update is performed. The average averaging time using the input signal spectrum Yi (f) and the past noise spectrum Ni-k (f) when calculating the later noise spectrum Ni (f) (however, k: time series). The input signal spectrum Yi (f) when calculating the updated noise spectrum Ni (f) in the case of a frame containing only the noise component (an integer of 1 or more representing the degree of deviation from the order number i in ) And the past noise spectrum Ni-k (f), which is longer than the average averaging time.

According to the fourth aspect of the present invention, when the frame to be processed is a frame containing only noise components, the signal corresponding to the amplitude spectrum of the frame to be processed is set as the input signal spectrum Yi (f) for each frequency f. The first update unit that calculates the updated noise spectrum Ni (f) by processing including averaging (however, i: the number of orders representing the time-series order), and the frame to be processed is a frame containing an audio component. In the case where the frame includes a second update unit for calculating the updated noise spectrum Ni (f) for each frequency f by a process including averaging, and the frame includes the voice component, the update is performed. The average averaging time using the input signal spectrum Yi (f) and the past input signal spectrum Yi-j (f) when calculating the later noise spectrum Ni (f) (however, j: hour). An integer of 1 or more representing the degree of deviation from the order number i in the series), the input signal spectrum Yi (in the case of a frame containing only the noise component) when calculating the updated noise spectrum Ni (f). It is a noise update circuit characterized in that it is longer than the average averaging time using f) and the past input signal spectrum Yi-j (f).

The invention according to claim 5 is the noise spectrum Ni (f) after the update in the case of the frame containing the voice component in the noise update circuit according to any one of claims 1 to 4. Is characterized in that the averaging time is a value in the range of 1 to 10 seconds.

前記処理対象のフレームが音声成分を含むフレームである場合に、ＩＩＲフィルタである以下の数式３もしくはＦＩＲフィルタである以下の数式４に従って周波数ｆごとに更新後のノイズスペクトルＮi(ｆ)を算出する第２の更新部と（但し、Ｋs：処理対象のフレームが音声成分を含むフレームである場合のＹi(ｆ)に対するＮi-1(ｆ)の重みづけを決定づける定数、Ｋn＜Ｋs）、

を有する、ことを特徴とするノイズ更新回路である。 The invention according to claim 6 is an IIR filter in which a signal corresponding to the amplitude spectrum of the frame to be processed is used as an input signal spectrum Yi (f) when the frame to be processed is a frame having only a noise component. The first updater for calculating the updated noise spectrum Ni (f) for each frequency f according to the following formula 1 or the following formula 2 which is an FIR filter (however, Ni-1 (f): one frame of update). Previous noise spectrum, Yi-j (f): Input signal spectrum before j-frame update, Kn: Ni-1 (f) with respect to Yi (f) when the frame to be processed is a frame containing only noise components. Constants that determine the weighting, i: the number of orders that represent the order of the time series, j: an integer greater than or equal to 0 that represents the degree of deviation from the number i of the order in the time series),

When the frame to be processed is a frame containing an audio component, the updated noise spectrum Ni (f) is calculated for each frequency f according to the following formula 3 which is an IIR filter or the following formula 4 which is an FIR filter. The second updater and (however, Ks: a constant that determines the weighting of Ni-1 (f) with respect to Yi (f) when the frame to be processed is a frame containing an audio component, Kn <Ks),.

It is a noise update circuit characterized by having.

In the invention according to claim 7, in the noise update circuit according to claim 6, the constant Ks is either the time constant of the IIR filter or the range corresponding to 1 to 10 seconds of the average interval of the FIR filter. It is characterized in that it is set to the value of.

前記入力信号スペクトルＹi(ｆ)と前記平均スペクトルレベルＭとに関して Ｙi(ｆ)≧Ｔe×Ｍ である周波数ｆについて、以下の数式６に従って更新後のノイズスペクトルＮi(ｆ)を決定する第３の更新部と（但し、Ｔe：係数、Ｎi-1(ｆ)：更新の１フレーム前のノイズスペクトル）、

をさらに有する、ことを特徴とする。 The invention according to claim 8 calculates an average spectrum level M for the input signal spectrum Yi (f) according to the following formula 5 in the noise update circuit according to any one of claims 1 to 7. (However, f1: minimum frequency in the amplitude spectrum, f2: maximum frequency in the amplitude spectrum, Fn: number of frequencies in the range from the minimum frequency f1 to the maximum frequency f2).

Regarding the frequency f in which Yi (f) ≧ Te × M with respect to the input signal spectrum Yi (f) and the average spectrum level M, the third noise spectrum Ni (f) after updating is determined according to the following formula 6. The update part and (however, Te: coefficient, Ni-1 (f): noise spectrum one frame before the update),

It is characterized by having further.

The invention according to claim 9 is characterized in that, in the noise update circuit according to claim 8, the coefficient Te is set to any value in the range of 1 to 100.

According to the first to fourth aspects of the invention, it is possible to follow the fluctuation of noise even when a frame containing an audio component continues as a frame to be processed, and the noise component can be accurately estimated. It becomes possible. Specifically, in the conventional circuit that realizes the spectrum subtraction method, the noise spectrum is updated when the frame to be processed is a frame containing only a noise component, while the noise spectrum is updated when the frame contains a voice component. Since the noise spectrum is not updated when a frame containing an audio component continues as a frame to be processed, it is not possible to accurately follow the noise fluctuation, and as a result, the noise component is accurately updated. There is a problem that it cannot be estimated. On the other hand, in the invention according to claim 1 to 4, the noise spectrum is updated even when the frame to be processed is a frame containing an audio component, so that the audio component is used as the frame to be processed. Even when the frame containing the noise continues, it is possible to follow the fluctuation of the noise, and it is possible to accurately estimate the noise component.

According to the fifth aspect of the present invention, when the frame to be processed is a frame containing an audio component, the averaging time for calculating the updated noise spectrum Ni (f) is set to an appropriate value. Therefore, it is possible to more reliably follow the fluctuation of noise when a frame containing an audio component continues as the frame to be processed, and it is possible to estimate the noise component more reliably and accurately.

According to the invention of claim 6, even when a frame containing an audio component continues as a frame to be processed, it is possible to follow the fluctuation of noise, and it is possible to accurately estimate the noise component. Become. Specifically, in the conventional circuit that realizes the spectrum subtraction method, the noise spectrum is updated when the frame to be processed is a frame containing only a noise component, while the noise spectrum is updated when the frame contains a voice component. Since the noise spectrum is not updated when a frame containing an audio component continues as a frame to be processed, it is not possible to accurately follow the noise fluctuation, and as a result, the noise component is accurately updated. There is a problem that it cannot be estimated. On the other hand, in the invention according to claim 6, since the noise spectrum is updated even when the frame to be processed is a frame containing an audio component, the frame including the audio component is used as the frame to be processed. Even if it continues, it is possible to follow the fluctuation of noise, and it is possible to accurately estimate the noise component.

According to the invention of claim 7, when the frame to be processed is a frame containing an audio component, the noise spectrum Ni-1 (f) one frame before the update is weighted with respect to the input signal spectrum Yi (f). Since the constant Ks that determines It is possible to estimate.

According to the invention of claim 8, it is possible to prevent the tone signal from being suppressed. Specifically, in the conventional circuit that realizes the spectrum subtraction method, the component that is constantly present in the frequency spectrum is judged to be noise and suppressed, so that the tone signal is also noise because it is stationary. There is a problem that it is judged and becomes a target of suppression, and a tone signal (for example, a Morse signal) necessary for the user is also suppressed. On the other hand, in the invention according to claim 8, since the frequency component whose amplitude spectrum is significantly larger than the average spectrum level is excluded from the update of the noise spectrum, it is possible not to suppress the tone signal. It will be possible.

According to the invention of claim 9, since the coefficient Te is set to an appropriate value, a frequency component whose amplitude spectrum is significantly larger than the average spectrum level can be more reliably excluded from the update of the noise spectrum. It is possible to prevent the tone signal from being suppressed more reliably.

It is a functional block diagram which shows the schematic structure of the noise reduction circuit including the noise update circuit which concerns on embodiment of this invention. It is a functional block diagram which shows the schematic structure of the noise update circuit which concerns on Embodiment 1. FIG. It is a functional block diagram which shows the schematic structure of the noise update circuit which concerns on Embodiment 2. FIG.

Hereinafter, the present invention will be described based on the illustrated embodiment.

(Embodiment 1)
FIG. 1 is a functional block diagram showing a schematic configuration of a noise reduction circuit 1 including a noise update circuit 11 according to an embodiment of the present invention. FIG. 2 is a functional block diagram showing a schematic configuration of the noise update circuit 11 according to the first embodiment.

The noise reduction circuit 1 is, for example, a circuit incorporated in a radio device that transmits and receives high-frequency signals to realize a spectrum subtraction method (Spectral Reduction), which is a method of suppressing noise components contained in an audio signal, and is mainly a pre. Emphasis circuit 2, window processing unit 3, time-frequency conversion unit 4, conversion result output unit 5, subtraction unit 6, synthesis unit 7, frequency-time conversion unit 8, de-emphasis circuit 9, and audio section. It has a detection unit 10 and a noise update circuit 11.

The pre-emphasis (PE) circuit 2 performs high-frequency enhancement processing for pre-amplifying the relative intensity of high-frequency components on an audio signal demodulated from a high-frequency signal received from an antenna, and after high-frequency enhancement processing. Output a signal.

The window processing unit 3 receives the input of the signal after the high frequency enhancement processing output from the pre-emphasis circuit 2, and extracts a frame having a predetermined time length from the input signal (for example, every 12.5 ms). A time waveform for 25 ms is extracted), and each frame is subjected to window processing by multiplying it by a window function such as a Hanning window. The window processing unit 3 outputs the frame after the window processing each time the window processing is performed on each frame.

The time-frequency conversion unit 4 receives the input of the window-processed frame output from the window processing unit 3, and each time the input of the frame is received, the time-frequency conversion unit 4 changes the signal in the time domain to the signal in the frequency domain with respect to the frame. The conversion process is performed, a frequency spectrum including an amplitude component and a phase component for each of a plurality of frequencies is calculated, and a signal of the frequency spectrum of a real number and an imaginary number is output. The time-frequency transform unit 4 executes a time-frequency transform and calculates a frequency spectrum by, for example, a discrete Fourier transform or a fast Fourier transform.

The conversion result output unit 5 receives the input of the signal of the frequency spectrum for each frame (for example, at intervals of about 12.5 ms) output from the time-frequency conversion unit 4, and for each frame, the input of the frequency spectrum. The signal corresponding to the amplitude spectrum including the amplitude component of each frequency is output to the subtraction unit 6, and the signal corresponding to the phase spectrum including the phase component of each frequency in the input frequency spectrum is synthesized. Output to unit 7.

The subtraction unit 6 receives the input of the signal corresponding to the amplitude spectrum for each frame output from the conversion result output unit 5, and the signal corresponding to the updated noise spectrum for each frame output from the noise update circuit 11. Upon receiving the input, for each frame, the signal corresponding to the input updated noise spectrum is subtracted from the input signal corresponding to the amplitude spectrum for each frequency (in other words, for each spectrum). As a result, the noise component contained in the audio signal is suppressed. The subtraction unit 6 outputs a signal corresponding to the amplitude spectrum after the subtraction process for each frame output from the conversion result output unit 5.

The synthesizing unit 7 receives the input of the signal corresponding to the phase spectrum for each frame output from the conversion result output unit 5, and the signal corresponding to the amplitude spectrum after the subtraction processing for each frame output from the subtraction unit 6. Upon receiving an input, for each frame, the input signal corresponding to the phase spectrum and the signal corresponding to the amplitude spectrum are combined to generate a frequency spectrum, and a signal having a frequency spectrum of a real number and an imaginary number is output. ..

The frequency-time conversion unit 8 receives the input of the frequency spectrum signal for each frame output from the synthesis unit 7, and for each frame, the signal in the frequency region to the signal in the time region with respect to the input signal of the frequency spectrum. The audio signal is output by performing the conversion process to, that is, the reverse conversion process of the conversion process in the time-frequency conversion unit 4. The frequency-time conversion unit 8 executes frequency-time conversion by, for example, an inverse discrete Fourier transform or an inverse fast Fourier transform to generate an audio signal.

The De-Emphasis (DE) circuit 9 receives the input of the audio signal output from the frequency time conversion unit 8 and attenuates the relative intensity of the high frequency component with respect to the input audio signal. That is, the attenuation processing by the inverse filter of the pre-emphasis circuit 2 is performed, and the audio signal after the high frequency attenuation processing is output.

The voice section detection unit 10 receives the input of the signal corresponding to the amplitude spectrum for each frame output from the conversion result output unit 5 and branched, and the noise of the input signal corresponding to the amplitude spectrum for each frame. It is determined whether the frame contains only components or the frame contains audio components.

The method of determining whether the frame to be processed contains only a noise component or a voice component in the voice section detection unit 10 is not limited to a specific procedure or method, and is not limited to a specific procedure or method, but is a conventional or new procedure or method. An appropriate procedure or method can be appropriately selected from the above.

As a method of determining whether the frame to be processed contains only a noise component or a voice component in the voice section detection unit 10, for example, focusing on the non-constancy of the voice, the magnitude of the amplitude of the amplitude spectrum for each frequency is large. When the average or dispersion value related to the amplitude is less than a predetermined threshold multiple times (for example, about 3 to 5 times) in a row in the latest frame, it is determined that the frame to be processed is only a noise component, and other than the above. In the case of, the method of determining that the frame to be processed has an audio component, or when the value of the average or dispersion regarding the magnitude of the amplitude of each frequency of the amplitude spectrum is less than a predetermined threshold value, the frame to be processed is A method may be used in which it is determined that only the noise component is present, and when the average or dispersion value is equal to or greater than the threshold value, it is determined that the frame to be processed has an audio component.

The voice section detection unit 10 outputs a noise frame signal when it is determined that the frame to be processed has only a noise component, and when it is determined that the frame to be processed has a voice component, the voice frame is output. Output a signal. The voice section detection unit 10 outputs a noise frame signal or a voice frame signal as a voice section detection result for each frame.

Then, in the noise update circuit 11 according to the embodiment, when the frame to be processed is a frame containing only noise components, the signal corresponding to the amplitude spectrum of the frame to be processed is set as the input signal spectrum Yi (f) and has a frequency. The first update part that calculates the noise spectrum Ni (f) after updating for each f by a process including averaging (however, i: the number of orders representing the order in the time series), and the frame to be processed contains the audio component. In the case of a frame including, it has a second update part for calculating the updated noise spectrum Ni (f) for each frequency f by a process including averaging, and is updated in the case of a frame containing an audio component. When calculating the updated noise spectrum Ni (f) when the averaging time of the input signal spectrum Yi (f) is a frame containing only noise components when calculating the later noise spectrum Ni (f). , It is longer than the averaging time of the input signal spectrum Yi (f).

The noise update circuit 11 adds the amplitude spectrum of the current frame (in other words, the frame to be processed, the latest frame) to the noise spectrum representing the noise component for each frequency calculated in the past, so as to obtain the latest noise. It updates to the spectrum and has a first updating unit 111 and a second updating unit 112.

The noise update circuit 11 receives the input of the signal corresponding to the amplitude spectrum for each frame output from the conversion result output unit 5 and branched, and the noise section detection result for each frame output from the voice section detection unit 10. The noise spectrum Ni (f) after the update is assumed to update the noise spectrum Ni-1 (f) one frame before the update for each frequency f by receiving the input and using the signal corresponding to the input amplitude spectrum. f) is calculated according to the following formula 7 or formula 8 or formula 9 or formula 10 according to the content of the input voice section detection result. The subscript i in the following mathematical formulas is an ordinal number representing the order of the time series, and represents the order that is commonly applied to all the mathematical formulas. Further, the subscript j in the following mathematical expressions is a variable representing the degree of deviation from the ordinal number i in the time series, and is an integer of 0 or more.

When the voice section detection result input to the noise update circuit 11 is a noise frame signal, the first update unit 111 inputs a signal corresponding to the input amplitude spectrum to the input signal spectrum Yi (f). The noise after updating for each frequency f according to the following formula 7 which is an IIR (abbreviation of Infinite Impulse Response; infinite impulse response) filter or the following formula 8 which is an FIR (abbreviation of Finite Impulse Response; finite impulse response) filter. The spectrum Ni (f) is calculated. In the following equations, Ni-1 (f) represents the noise spectrum one frame before the update, and Yi-j (f) represents the input signal spectrum one frame before the update.

Kn in Equation 7 and Equation 8 is an input signal spectrum Yi which is an amplitude spectrum of the frame to be processed when the frame to be processed (in other words, the current frame and the latest frame) is a frame containing only a noise component. It is a constant that determines the weighting of the noise spectrum Ni-1 (f) one frame before the update with respect to (f), and is called "the weight constant Kn before the update at the time of noise".

The weight constant Kn before update at the time of noise is not limited to a specific value as long as it is an integer of 0 or more. Specifically, the weight constant Kn before updating during noise is Kn in a range corresponding to, for example, the time constant of the IIR filter or the average interval of the FIR filter of about 0.06 to 0.20 seconds (for example, at a frame interval of 12.5 ms). It is conceivable that it is set to any value in the range of about 5 to 16), and in particular, a value corresponding to about 0.1 seconds of the time constant of the IIR filter or the average interval of the FIR filter (for example, frame). It is conceivable that Kn = 8) is set at an interval of 12.5 ms. When the time constant of the IIR filter or the average interval of the FIR filter is, for example, 0.1 seconds, the specific value of the constant Kn in Equation 7 and the specific value of the constant Kn in Equation 8 are different.

When the voice section detection result input to the noise update circuit 11 is a voice frame signal, the second update unit 112 inputs a signal corresponding to the input amplitude spectrum to the input signal spectrum Yi (f). The updated noise spectrum Ni (f) is calculated for each frequency f according to the following formula 9 which is an IIR filter or the following formula 10 which is an FIR filter.

Ks in Equation 9 and Equation 10 is an input signal spectrum Yi which is an amplitude spectrum of the frame to be processed when the frame to be processed (in other words, the current frame and the latest frame) is a frame containing an audio component. It is a constant that determines the weighting of the noise spectrum Ni-1 (f) one frame before the update with respect to (f), and is called "the weight constant Ks before the update during voice".

The weight constant Ks before update during voice is not limited to a specific value as long as it is an integer larger than the weight constant Kn before update during noise, preferably sufficiently large. Specifically, the weight constant Ks before update during voice is, for example, a range corresponding to about 1 to 10 seconds of the time constant of the IIR filter or the average interval of the FIR filter (for example, Ks = 80 to 800 at a frame interval of 12.5 ms). Any value in the range of degrees), more specifically, the time constant of the IIR filter or the range corresponding to about 1 to 6 seconds of the average interval of the FIR filter (for example, Ks = 80 to 500 at a frame interval of 12.5 ms). It is conceivable that it is set to any value within the range of degree), and in particular, a value corresponding to about 3.7 seconds of the time constant of the IIR filter or the average interval of the FIR filter (for example, the frame interval of 12.5 ms). It is conceivable that Ks = about 295). When the time constant of the IIR filter or the average interval of the FIR filter is, for example, 3.7 seconds, the specific value of the constant Ks in Equation 9 and the specific value of the constant Ks in Equation 10 are different.

By setting the pre-update weighting constant Ks to a sufficiently large value during audio, the audio has non-homeostasis, which prevents the audio component from being significantly reflected in the noise spectrum update, and therefore the audio component. Is not suppressed as a noise component.

As an averaging process for calculating the updated noise spectrum Ni (f), an IIR filter (general form is Ni (f) = ΣAjYi-j (f) + ΣBkNi-k (f); however, A and B Is a coefficient, i is an order number representing the order of the time series, j is an integer of 0 or more representing the degree of deviation from the order number i in the time series, and k is 1 indicating the degree of deviation from the order number i in the time series. When the above integer) is used, when the frame to be processed is a frame containing only noise components, the signal corresponding to the amplitude spectrum of the frame to be processed is used as the input signal spectrum Yi (f) and updated for each frequency f. The first update unit 111 that calculates the subsequent noise spectrum Ni (f) by processing including averaging, and the noise spectrum Ni after updating for each frequency f when the frame to be processed is a frame containing an audio component. An input signal for calculating the updated noise spectrum Ni (f) in the case of a frame containing an audio component and having a second update unit 112 for calculating (f) by a process including averaging. After updating when the average averaging time using the spectrum Yi (f), the past input signal spectrum Yi-j (f), and the past noise spectrum Ni-k (f) is a frame containing only noise components. Average of the average of the input signal spectrum Yi (f), the past input signal spectrum Yi-j (f), and the past noise spectrum Ni-k (f) when calculating the noise spectrum Ni (f) of Applicable when it is longer than the conversion time.

Further, when using the IIR filter in which Aj = 0 in j ≧ 1 for the general form of the above IIR filter, it corresponds to the amplitude spectrum of the frame to be processed when the frame to be processed is a frame containing only noise components. The signal to be processed is the input signal spectrum Yi (f), and the first update unit 111, which calculates the updated noise spectrum Ni (f) for each frequency f by a process including averaging, and the frame to be processed have audio components. When the frame includes a second update unit 112 which calculates the updated noise spectrum Ni (f) for each frequency f by a process including averaging, and includes a voice component. The average averaging time using the input signal spectrum Yi (f) and the past noise spectrum Ni-k (f) when calculating the updated noise spectrum Ni (f) is in the frame of only the noise component. In some cases, it is longer than the average averaging time using the input signal spectrum Yi (f) and the past noise spectrum Ni-k (f) when calculating the updated noise spectrum Ni (f). Corresponds to.

Regarding the general form of the above IIR filter, when Aj = 0 in j ≧ 1, A0 = 1 / (1 + Kn), B1 = Kn / (1 + Kn), and Bk = 0 (k ≧ 2). Corresponds to the above formula 7, and the case where A0 = 1 / (1 + Ks), B1 = Ks / (1 + Ks), and Bk = 0 (k ≧ 2) corresponds to the above formula 9.

Further, as an averaging process for calculating the updated noise spectrum Ni (f), an FIR filter (general form is Ni (f) = ΣAjYi-j (f); where A is a coefficient and i is a time series. When using an order number indicating the order, j is an integer of 0 or more indicating the degree of deviation from the order number i in the time series; that is, Bk = 0 for the general form of the above IIR filter), the processing target is used. When the frame of is a frame containing only noise components, the signal corresponding to the amplitude spectrum of the frame to be processed is set as the input signal spectrum Yi (f), and the updated noise spectrum Ni (f) is averaged for each frequency f. When the frame to be processed is a frame containing an audio component, the first update unit 111 calculated by the process including The input signal spectrum Yi (f) and the past input signal spectrum when calculating the updated noise spectrum Ni (f) in the case of a frame containing an audio component and having a second updating unit 112. The input signal spectrum Yi (f) and the past when calculating the updated noise spectrum Ni (f) when the average averaging time using Yi-j (f) is a frame containing only noise components. It corresponds to the case where it is longer than the average averaging time using the input signal spectrum Yi-j (f) of.

Regarding the general form of the above FIR filter, the case where Aj = 1 / Kn corresponds to the above formula 8 and the case where Aj = 1 / Ks corresponds to the above formula 10.

Each time the noise update circuit 11 receives the signal corresponding to the amplitude spectrum for each frame and the input of the voice interval detection result, the noise update circuit 11 inputs the updated signal corresponding to the noise spectrum Ni (f) for each frequency f to the subtraction unit 6. Output to. The subtraction unit 6 uses the signal corresponding to the updated noise spectrum Ni (f) output from the noise update circuit 11 for each frame, and the signal corresponding to the amplitude spectrum output from the conversion result output unit 5. Is processed to subtract the signal corresponding to the updated noise spectrum Ni (f).

According to the noise update circuit 11 as described above, it is possible to follow the fluctuation of noise even when a frame containing an audio component continues as a frame to be processed, and it is possible to accurately estimate the noise component. It becomes. Specifically, in the conventional circuit that realizes the spectrum subtraction method, the noise spectrum is updated when the frame to be processed is a frame containing only a noise component, while the noise spectrum is updated when the frame contains a voice component. Since the noise spectrum is not updated when a frame containing an audio component continues as a frame to be processed, it is not possible to accurately follow the noise fluctuation, and as a result, the noise component is accurately updated. There is a problem that it cannot be estimated. On the other hand, in the noise update circuit 11 as described above, since the noise spectrum is updated even when the frame to be processed is a frame containing an audio component, the frame including the audio component as the frame to be processed is used. It is possible to follow the fluctuation of noise even when the noise component continues, and it is possible to accurately estimate the noise component.

(Embodiment 2)
FIG. 3 is a functional block diagram showing a schematic configuration of the noise update circuit 11 according to the second embodiment of the present invention.

This embodiment is different from the first embodiment in that the noise update circuit 11 has an additional configuration as compared with the configuration of the first embodiment, but there are also common configurations and processing contents, and the embodiment is implemented. The same reference numerals are given to the configurations and processing contents equivalent to those in No. 1, and the description thereof will be omitted.

The noise update circuit 11 according to this embodiment has an average value calculation unit 113 that calculates an average spectrum level M for the input signal spectrum Yi (f), and Yi (for the input signal spectrum Yi (f) and the average spectrum level M. f) Further has a third update unit 114 for determining the updated noise spectrum Ni (f) for the frequency f for which ≧ Te × M.

The noise update circuit 11 receives the input of the signal corresponding to the amplitude spectrum for each frame output from the conversion result output unit 5 and branched, and the noise section detection result for each frame output from the voice section detection unit 10. The noise spectrum Ni (f) after the update is assumed to update the noise spectrum Ni-1 (f) one frame before the update for each frequency f by receiving the input and using the signal corresponding to the input amplitude spectrum. f) is calculated according to any one of the above-mentioned formula 7 or formula 8, formula 9 or formula 10, and formula 12 according to the content of the input voice section detection result and the like.

The mean value calculation unit 113 calculates the mean spectrum level M according to the following equation 11 with the signal corresponding to the amplitude spectrum for each frame output from the conversion result output unit 5 and branched as the input signal spectrum Yi (f). .. The average spectrum level M is, that is, the average value of the amplitude spectrum in the frequency axis direction.

In Equation 11, f represents the frequency in the amplitude spectrum, f1 is the minimum frequency in the amplitude spectrum, f2 is the maximum frequency in the amplitude spectrum, and Fn is from the minimum frequency f1 to the maximum frequency f2. Represents the number of frequencies in the range. The number of frequencies Fn is n / 2 + 1, that is, where n is the number of sample points included in one frame in the time-frequency conversion in the time-frequency conversion unit 4.

The mean value calculation unit 113 calculates the mean spectrum level M each time a signal corresponding to the amplitude spectrum for each frame is input.

Then, when the voice section detection result input to the noise update circuit 11 is a noise frame signal, the signal corresponding to the input amplitude spectrum is set as the input signal spectrum Yi (f), and the above is described for each frequency f. The following processing of <a> or <b> is performed according to the relationship between the input signal spectrum Yi (f) and the average spectrum level M.

<A> For the frequency f where Yi (f) <Te × M, the first updating unit 111 calculates the updated noise spectrum Ni (f) according to the above formula 7 or formula 8.

Te in the relationship between the input signal spectrum Yi (f) and the average spectrum level M is a coefficient for excluding frequency components whose amplitude spectrum is significantly larger than the average spectrum level M from updating the noise spectrum. The coefficient Te is not limited to a specific value. For example, the tone signal is not subtracted by the subtractor 6 and is not suppressed as a noise component so that the frequency component corresponding to the tone signal is excluded from the update of the noise spectrum. It is set to an appropriate value after considering the above. Specifically, the coefficient Te may be set to any value in the range of about 1 to 100, and in particular, it may be set to about 10.

<A> With respect to the frequency f in which Yi (f) ≧ Te × M, the third update unit 114 determines the updated noise spectrum Ni (f) according to the following equation 12.

Ni-1 (f) in Equation 12 represents the noise spectrum one frame before the update.

When the voice section detection result input to the noise update circuit 11 is a voice frame signal, the signal corresponding to the input amplitude spectrum is set as the input signal spectrum Yi (f), and the above is described for each frequency f. The following processing <c> or <d> is performed according to the relationship between the input signal spectrum Yi (f) and the average spectrum level M.

<C> For the frequency f where Yi (f) <Te × M, the second updating unit 112 calculates the updated noise spectrum Ni (f) according to the above formula 9 or formula 10.

<D> With respect to the frequency f in which Yi (f) ≧ Te × M, the third updating unit 114 determines the updated noise spectrum Ni (f) according to the above equation 12.

In the above processing, that is, for a frequency component in which Yi (f) ≧ Te × M and the amplitude spectrum is significantly larger than the average spectrum level M, the noise spectrum Ni (f) after the update is updated to 1. The noise spectrum before the frame is Ni-1 (f) and is excluded from the update.

Each time the noise update circuit 11 receives the signal corresponding to the amplitude spectrum for each frame and the input of the voice interval detection result, the noise update circuit 11 inputs the updated signal corresponding to the noise spectrum Ni (f) for each frequency f to the subtraction unit 6. Output to. The subtraction unit 6 uses the signal corresponding to the updated noise spectrum Ni (f) output from the noise update circuit 11 for each frame, and the signal corresponding to the amplitude spectrum output from the conversion result output unit 5. Is processed to subtract the signal corresponding to the updated noise spectrum Ni (f).

According to the noise update circuit 11 as described above, it is possible not to suppress the tone signal. Specifically, in the conventional circuit that realizes the spectrum subtraction method, the component that is constantly present in the frequency spectrum is judged to be noise and suppressed, so that the tone signal is also noise because it is stationary. There is a problem that it is judged and becomes a target of suppression, and a tone signal (for example, a Morse signal) necessary for the user is also suppressed. On the other hand, in the noise update circuit 11 as described above, since the frequency component whose amplitude spectrum is significantly larger than the average spectrum level is excluded from the update of the noise spectrum, the tone signal should not be suppressed. Is possible.

Regarding the noise update circuit 11 according to the second embodiment, the mode is selected in which the user selects whether or not to operate the average value calculation unit 113 and the third update unit 114 according to the operation at that time. At the same time, the method of updating the noise spectrum may be adjusted so that the method of suppressing the noise component may be adjusted. For example, when the user mainly transmits and receives an audio signal while suppressing a noise component, the user selects a mode in which the mean value calculation unit 113 and the third update unit 114 do not function to suppress the tone signal, while suppressing the tone signal. When transmitting and receiving signals, it is conceivable to select a mode in which the mean value calculation unit 113 and the third update unit 114 function so as not to suppress the tone signal.

Although the embodiment of the present invention has been described above, the specific configuration is not limited to the above-described embodiment, and even if there is a design change or the like within a range that does not deviate from the gist of the present invention. Included in the invention. For example, in the above embodiment, the case where the noise update circuit 11 according to the present invention is applied to the noise reduction circuit 1 whose schematic configuration is shown in FIG. 1 is described as an example, but the present invention is applied. The possible configuration of the noise reduction circuit is not limited to the example shown in FIG. Furthermore, the circuits to which the present invention can be applied are not limited to noise reduction circuits. That is, the present invention can be applied to various circuits in which the noise spectrum needs to be updated in time series.

Further, in the process including averaging for calculating the noise spectrum Ni (f) after the update, the coefficients corresponding to the noise constant before update Kn and the voice constant before update Ks in the above embodiment are used. Any method may be used as long as it is a method. That is, the updated noise spectrum Ni (f) calculation formula is not limited to the formula 7 to the formula 10 in the above embodiment.

1 Noise reduction circuit 2 Pre-emphasis circuit 3 Window processing unit 4 Time frequency conversion unit 5 Conversion result output unit 6 Subtraction unit 7 Synthesis unit 8 Frequency time conversion unit 9 De-emphasis circuit 10 Voice section detection unit 11 Noise update circuit 111 First Update unit 112 Second update unit 113 Average value calculation unit (Embodiment 2)
114 Third update section (Embodiment 2)

Claims (9)

When the frame to be processed is a frame containing only noise components, the signal corresponding to the amplitude spectrum of the frame to be processed is set as the input signal spectrum Yi (f), and the noise spectrum Ni (f) updated for each frequency f. With the first update part calculated by processing including averaging (however, i: the number of orders representing the order of the time series),
When the frame to be processed is a frame containing an audio component, it has a second update unit for calculating the updated noise spectrum Ni (f) for each frequency f by a process including averaging.
When the average time for calculating the updated noise spectrum Ni (f) is a frame containing only the noise component, the updated noise spectrum Ni (f) is used. Longer than the averaging time when calculating,
A noise update circuit characterized by that. When the frame to be processed is a frame containing only noise components, the signal corresponding to the amplitude spectrum of the frame to be processed is set as the input signal spectrum Yi (f), and the noise spectrum Ni (f) updated for each frequency f. With the first update part calculated by processing including averaging (however, i: the number of orders representing the order of the time series),
When the frame to be processed is a frame containing an audio component, it has a second update unit for calculating the updated noise spectrum Ni (f) for each frequency f by a process including averaging.
The input signal spectrum Yi (f), the past input signal spectrum Yi-j (f), and the past noise when calculating the updated noise spectrum Ni (f) in the case of a frame containing the voice component. The average averaging time using the spectrum Ni-k (f) (where j: an integer of 1 or more representing the degree of deviation from the order number i in the time series, k: the order number i in the time series). The input signal spectrum Yi (f) and the past input when calculating the updated noise spectrum Ni (f) in the case of a frame containing only the noise component (an integer of 1 or more representing the degree of separation). It is longer than the average averaging time using the signal spectrum Yi-j (f) and the past noise spectrum Ni-k (f).
A noise update circuit characterized by that. When the frame to be processed is a frame containing only noise components, the signal corresponding to the amplitude spectrum of the frame to be processed is set as the input signal spectrum Yi (f), and the noise spectrum Ni (f) updated for each frequency f. With the first update part calculated by processing including averaging (however, i: the number of orders representing the order of the time series),
When the frame to be processed is a frame containing an audio component, it has a second update unit for calculating the updated noise spectrum Ni (f) for each frequency f by a process including averaging.
An average using the input signal spectrum Yi (f) and the past noise spectrum Ni-k (f) when calculating the updated noise spectrum Ni (f) in the case of a frame containing the voice component. (However, k: an integer of 1 or more representing the degree of deviation from the order number i in the time series), and in the case of a frame containing only the noise component, the updated noise spectrum Ni (f) is obtained. It is longer than the average averaging time using the input signal spectrum Yi (f) and the past noise spectrum Ni-k (f) in the calculation.
A noise update circuit characterized by that. When the frame to be processed is a frame containing only noise components, the signal corresponding to the amplitude spectrum of the frame to be processed is set as the input signal spectrum Yi (f), and the noise spectrum Ni (f) updated for each frequency f. With the first update part calculated by processing including averaging (however, i: the number of orders representing the order of the time series),
When the frame to be processed is a frame containing an audio component, it has a second update unit for calculating the updated noise spectrum Ni (f) for each frequency f by a process including averaging.
The input signal spectrum Yi (f) and the past input signal spectrum Yi-j (f) used in calculating the updated noise spectrum Ni (f) in the case of a frame containing the voice component are used. The updated noise spectrum Ni (f) when the average averaging time (where j: an integer of 1 or more representing the degree of deviation from the order number i in the time series) and the frame containing only the noise component is used. Is longer than the average averaging time using the input signal spectrum Yi (f) and the past input signal spectrum Yi-j (f) in calculating.
A noise update circuit characterized by that. In the case of a frame containing the voice component, the averaging time when calculating the updated noise spectrum Ni (f) is any value in the range of 1 to 10 seconds.
The noise update circuit according to any one of claims 1 to 4, wherein the noise update circuit is characterized in that. When the frame to be processed is a frame containing only noise components, the signal corresponding to the amplitude spectrum of the frame to be processed is set as the input signal spectrum Yi (f), and the following equation 1 or FIR filter which is an IIR filter is used. The first update unit that calculates the updated noise spectrum Ni (f) for each frequency f according to the following formula 2 (however, Ni-1 (f): the noise spectrum one frame before the update, Yi-j ( f): Input signal spectrum before j-frame update, Kn: Constant that determines the weighting of Ni-1 (f) to Yi (f) when the frame to be processed is a frame containing only noise components, i: Hour Order number representing the order of the series, j: an integer of 0 or more representing the degree of deviation from the order number i in the time series),

When the frame to be processed is a frame containing an audio component, the updated noise spectrum Ni (f) is calculated for each frequency f according to the following formula 3 which is an IIR filter or the following formula 4 which is an FIR filter. The second updater and (however, Ks: a constant that determines the weighting of Ni-1 (f) with respect to Yi (f) when the frame to be processed is a frame containing an audio component, Kn <Ks),.

A noise update circuit characterized by having. The constant Ks is set to a value in either the time constant of the IIR filter or the range corresponding to 1 to 10 seconds of the average interval of the FIR filter.
The noise update circuit according to claim 6. For the input signal spectrum Yi (f), an average value calculation unit for calculating the average spectrum level M according to the following formula 5 (however, f1: minimum frequency in the amplitude spectrum, f2: maximum frequency in the amplitude spectrum, Fn: Number of frequencies in the range from the minimum frequency f1 to the maximum frequency f2),

Regarding the frequency f in which Yi (f) ≧ Te × M with respect to the input signal spectrum Yi (f) and the average spectrum level M, the third noise spectrum Ni (f) after updating is determined according to the following formula 6. The update part and (however, Te: coefficient, Ni-1 (f): noise spectrum one frame before the update),

The noise update circuit according to any one of claims 1 to 7, further comprising. The coefficient Te is set to any value in the range of 1 to 100.
The noise update circuit according to claim 8.

Images