JP2010160521A - Noise canceller, and communication device equipped with the same - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce generation of voice distortion in noise suppressing by performing stable noise suppression processing by avoiding influence of change, even when noise energy is changed. <P>SOLUTION: In a noise minimum value estimation circuit 29, a minimum value of acoustic noise power of each band is calculated, and a spectrum feature of the noise minimum value is used for determination of a band classified gain by a band classified gain determination section 33. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a noise canceller provided in a communication device that encodes and transmits an audio signal, such as a digital automobile / mobile phone device, a digital cordless phone, a digital wired phone device, and the like, and a communication device including the noise canceller.

  In a digital cellular phone device, a low bit rate speech encoding method such as a CELP (Code Excited Linear Prediction) method is generally used. By using this type of coding system, it is possible to perform a good voice call even in an environment where the background noise is relatively large. For details of the CELP method, see MR Schroeder and BSAtal in “Code-Excited Linear Prediction (CELP): High-Quality Speech At Very Low Bit Rates” in Proc.ICASSP, 1985.pp.937-939. It is stated.

  However, in a noisy environment such as a railway platform or a main road, background noise significantly reduces voice clarity. For this reason, various studies have been made on noise cancellers that remove noise and use only speech for encoding. As an example, there is a noise canceller defined as an option in the variable rate speech coding system standardized in the United States (see Non-Patent Document 1, for example).

  FIG. 9 is a functional block diagram of a noise canceller defined in the standard described in Non-Patent Document 1. In the figure, 51 is a fast Fourier transform (FFT), and a framed transmission signal is input to the FFT 51. Note that the framed transmission signal is generated by dividing the A / D converted transmission signal into, for example, 80-sample frames and then adjusting it to 128 samples including the overlap. The FFT 51 performs a fast Fourier transform process on the framed transmission signal, thereby obtaining, for example, transform coefficients that are frequency-divided into 16 bands.

  The 16-band conversion coefficient obtained by the FFT 51 is input to the band energy estimation unit 52. The band energy estimation unit 52 calculates the energy of the conversion coefficients of the 16 bands and inputs the calculated values of the band-specific energy to the band SNR estimation unit 53 and the noise estimation unit 54. The noise estimation unit 54 estimates the band-specific energy of the noise portion based on the estimated value of the band-specific energy, and supplies the estimated band-specific noise energy to the band SNR estimation unit 53. The band SNR estimator 53 calculates a logarithmic value (SNR) of signal energy and noise energy for each band based on the calculated value of the band-specific energy and the estimated value of the noise energy for each band. Then, this band-specific SNR is given to the voice metric calculation unit 55. The voice metric calculation unit 55 multiplies the band-specific SNR by a weighting coefficient corresponding to the size and then obtains the sum, and the calculation result is a voice metric.

  The band SNR correction unit 56 determines that the voice is not included when the voice metric is smaller than the threshold value, corrects the band SNR to a small value, and provides the band gain calculation unit 57 with the band SNR. The band gain calculation unit 57 determines the noise suppression amount of each band based on the modified SNR, and gives it to the multiplication unit 61. As a result, the multiplication unit 61 multiplies the 16-band conversion coefficient output from the FFT 51 and the noise suppression amount for each band for each band, and outputs a signal with noise suppression for each band. Then, the transform coefficient subjected to noise suppression for each band is subjected to inverse fast Fourier transform in an inverse fast Fourier transform unit (IFFT) 58, thereby returning to a time domain signal and outputting it.

The noise energy is updated after the spectrum deviation estimation unit 59 and the noise update determination unit 60 determine whether or not the update is possible.
That is, in the noise canceller, the noise suppression amount for each band is determined based on the SNR of the transmission signal, that is, the ratio between the signal energy and the noise energy. Further, in order to determine whether or not speech is included, a voice metric which is a value obtained by weighted addition of each band SNR is used.

“Enhanced Variable Rate Codec, Speech Service Option 3 for Wideband Spread Spectrum Digital Systems” (TIA IS127)

  By the way, although noise (Background Noise) is generally assumed to be steady, it may fluctuate outdoors. In particular, the energy of noise generated when a vehicle passes by increases as the vehicle approaches. When the transmitted voice is input in this state, the energy difference between the voice and noise is small, and thus the suppressed voice may be distorted. Further, even when the noise spectrum shape is similar to the speech spectrum shape, if suppression is performed based on noise energy, interference with the speech spectrum is likely to occur, and thus the suppressed speech is distorted.

  Since the voice metric used for voice detection is obtained by multiplying the SNR by a weighting coefficient and taking the sum, it is likely to be regarded as voice when sudden noise is mixed.

  The present invention has been made paying attention to the above circumstances, and its purpose is to eliminate the influence of noise energy fluctuations and to perform stable noise suppression processing, thereby generating voice distortion in noise suppression. The present invention is to provide a noise canceller with reduced noise and a communication apparatus including the noise canceller.

  In order to achieve the above object, one aspect of the present invention is to divide an input signal into frames having a fixed time length, and divide the signals of these frames into a plurality of frequency bands, respectively. In a noise canceller that performs suppression processing, the power of a signal is obtained for each frequency band, noise power estimation means for estimating noise power for each band based on the band power, and the band power and the noise power for each band For at least one, a minimum value detecting means for detecting a minimum value of power over a plurality of frame periods, and obtaining a difference between the band power for each frequency band and the minimum value for each band detected by the minimum value detecting means, And a suppression amount determining means for determining a noise suppression amount for each frequency band based on this difference.

  Specifically, the suppression amount determination means performs a noise suppression amount determination process for each band based on a difference between the band power and the minimum value for each band in at least the speech section of the speech section and the noise section.

  Therefore, according to one aspect of the present invention, the minimum value of the band power or the noise power for each band is detected for each frequency band, and the noise suppression amount for each frequency band is determined based on the difference between the band power and the minimum value. It is determined. For this reason, even if, for example, a vehicle passes during a call and the noise energy temporarily increases, the noise suppression amount is determined based on the difference between the band power and the minimum value, so the suppression amount is stable. As a result, stable noise suppression processing is always possible.

  In addition, even if the noise spectrum shape is similar to the speech spectrum shape, suppression processing is performed based on the difference between the band power and the above minimum value, making it less susceptible to noise energy. Interference with the spectrum of is reduced and the occurrence of speech distortion is prevented.

  Further, according to the present invention, the suppression amount determining means generates a common adjustment value for each band different for each frame, and for each frequency band, the band power and the minimum value for each band detected by the minimum value detecting means are It is also characterized in that a difference from a value obtained by adding an adjustment value common to the bands is obtained, and a noise suppression amount for each frequency band is determined based on the difference.

Specifically, in the noise period, a common adjustment value is determined based on the difference between the average value between the minimum values for each band and the average value between the noise powers for each band. An adjustment value common to the bands is determined based on the difference between the minimum value among the plurality of band powers and the maximum value among the minimum values according to the bands.
By doing so, it becomes possible to determine a more appropriate noise suppression amount for each of the speech section and the noise section.

  That is, according to the present invention, even when the noise energy fluctuates, it is possible to eliminate the influence and perform stable noise suppression processing, thereby including a noise canceller that reduces the occurrence of voice distortion in noise suppression and the noise canceller. A communication device can be provided.

1 is a circuit block diagram showing an embodiment of a digital cellular phone device provided with a noise canceller according to the present invention. The functional block diagram which shows one Embodiment of the noise canceller concerning this invention. The flowchart which shows the processing procedure and the content in the significant value calculation part of the noise canceller shown in FIG. The flowchart which shows the process sequence and the content in the audio | voice weight calculation part of the noise canceller shown in FIG. The flowchart which shows the processing procedure in the noise minimum value estimation part of the noise canceller shown in FIG. 2, and the first half part of the content. The flowchart which shows the latter half part of the process sequence in the noise minimum value estimation part of the noise canceller shown in FIG. The figure which shows the relationship between the band power in a noise area, a noise minimum value, and the noise minimum value after adjustment. The figure which shows the relationship between the band power in a speech area, the noise minimum value, and the noise minimum value after adjustment. The figure which shows an example of a structure of the conventional noise canceller.

Embodiments according to the present invention will be described below with reference to the drawings.
FIG. 1 is a circuit block diagram showing an embodiment of a digital cellular phone device provided with a noise canceller according to the present invention.
In the figure, a radio carrier signal transmitted from a base station (not shown) via a radio channel is received by an antenna 1 and then input to a receiving circuit (RX) 3 via an antenna duplexer (DUP) 2. Thus, the received local oscillation signal output from the frequency synthesizer (SYN) 4 is mixed and frequency-converted to an intermediate frequency signal. The received intermediate frequency signal is sampled by an A / D converter (not shown) and then input to a digital demodulator (DEM) 6.

  The digital demodulator 6 performs digital demodulation processing after establishing frame synchronization and bit synchronization for the digital reception intermediate frequency signal. The baseband digital demodulated signal obtained by this demodulation processing is input to a time division multiple access circuit (TDMA) 8, where the time slot addressed to itself is separated and extracted for each transmission frame. Information regarding frame synchronization and bit synchronization obtained in the digital demodulator 6 is input to the control circuit 18.

  The digital demodulated signal output from the TDMA circuit 8 is subsequently input to an error correction code decoding circuit (CH-COD) 9 where error correction decoding processing is performed. Then, this error-corrected decoded digital demodulated signal is input to a voice decoding circuit (DEC) 10 and subjected to voice decoding processing, whereby a digital received signal is reproduced. The digital reception signal is converted back to an analog reception signal by the D / A converter 11 and then supplied to the speaker 12 via a voice amplifier (not shown), and is output from the speaker 12 as a loud voice.

  On the other hand, the transmitted voice of the speaker is collected by the microphone 13 and converted into an electrical signal, which is then input to the A / D converter 14 and sampled by the A / D converter 14 at a predetermined sampling period. It is converted into a digital transmission signal. This digital transmission signal is passed through a noise canceller 17 which will be described later, and then input to a voice encoding circuit (COD) 16 for voice encoding.

  The encoded speech data output from the speech encoding circuit 16 is input to the error correction code decoding circuit (CH-COD) 9 together with the control signal output from the control circuit 18, and is subjected to error correction encoding. The error-encoded digital transmission signal is input to the TDMA circuit 8. In this TDMA circuit 8, a transmission frame corresponding to a time division multiple access (TDMA) system is generated, and processing for inserting the digital transmission signal into a time slot assigned to the own apparatus in this transmission frame is performed. . The digital transmission signal output from the TDMA circuit 8 is subsequently input to the digital modulator (MOD) 7.

  The digital modulator 7 generates a transmission intermediate frequency signal digitally modulated by the digital transmission signal. The transmission intermediate frequency signal is converted into an analog signal by a D / A converter (not shown), and then transmitted to a transmission circuit (TX) 5. Is input. As a digital modulation method, for example, a π / 4 shift QPSK (π / 4 shifted quadrature phase shift keying) method is used.

  In the transmission circuit 5, the modulated transmission intermediate frequency signal is first mixed with the transmission local oscillation signal output from the frequency synthesizer 4, and thereby converted to a radio carrier frequency corresponding to the speech channel. This radio modulated wave signal is controlled to a predetermined transmission power level instructed by the control signal TCS from the control circuit 18 in the transmission power amplifier, and then from the antenna 1 to the base station (not shown) via the antenna duplexer 2. Sent to.

Reference numeral 19 denotes an operation panel unit. The operation panel unit 19 includes a key input unit having a call key, an end key, a dial key, and various function keys, a liquid crystal display (LCD), and a light emitting diode (LED). And a display unit having the same.
Incidentally, the noise canceller 17 is realized by, for example, a DSP (Digital Signal Processor), and the processing program is stored in a memory in the noise canceller or a memory attached to the control circuit 18. FIG. 2 is a block diagram showing a configuration of functions realized by this processing program.

  The digital transmission signal output from the A / D converter 14 is first input to the frame dividing unit 21. The frame dividing unit 21 divides the digital transmission signal into, for example, 80-sample frames, and then overlaps the frame ends by performing windowing. As a result, the frames adjusted to 128 samples including the overlaps are arranged. Output. The digital transmission signal frame is input to the fast Fourier transform unit (FFT) 22.

  The FFT 22 performs a fast Fourier transform process on the input digital transmission signal frame, thereby obtaining, for example, transform coefficients that are frequency-divided into 16 bands from low to high. Note that the number of conversion coefficients for each band may not be the same. The band-divided conversion coefficient is input to the noise suppression multiplier 23 and also to the band power calculation unit 26.

  The band power calculation unit 26 obtains energy (root mean square value of the conversion variable) for each band, takes a logarithm, and outputs band power channel_power (m, k). Here, m is a frame number, and k is a band number (k = 0,..., 15). The band power is input to a noise leak integrated value update unit 32 and a band-specific gain determination unit 33, which will be described later, and also input to the significant value calculation unit 27.

  For each band, the significant value calculator 27 calculates a difference tmp between a noise leak integrated value noise_power (m, k) output from a noise leak integrated value updater 32 described later and the band power channel_power (m, k). The difference tmp for each band is compared with a predetermined threshold value. When it is determined that the band-specific differences tmp of adjacent bands among the band-specific differences tmp arranged in order of frequency exceed the threshold value, the band-specific differences tmp are given a predetermined weight. And then add to each other. Then, the weighted value suby (m, k) and the conditional sum of the suby are output as a significant value y.

FIG. 3 is a flowchart showing the processing procedure and contents of the significant value calculation unit 27.
In the figure, the significant value calculator 27 first resets the frame number m to 0 in step 3a, then increments the group number m in step 3b, and exceeds the significant value y, the band number k, and the threshold value. The continuous number flag of the band-specific difference tmp is initialized to 0 respectively.

Next, the significant value calculation unit 27 weights the difference tmp between the band power and the noise leak integrated value for the band k = 0 in step 3c, and the value suby (m, k) obtained by weighting the band-specific difference tmp. And are calculated as follows.
tmp = channel_power (m, k) −noise_power (m, k)
suby (m, k) = {200- (k-1) 2} / 100 * (tmp-1)
However, {200- (k-1) 2} is a weighting coefficient.

  When the band-specific difference tmp in the band k = 0 is calculated, the significant value calculating unit 27 compares the band-specific difference tmp with the threshold “1” in step 3d, and the band-specific difference tmp is the threshold. If the value exceeds “1”, it is determined that there is a possibility of voice, and the process proceeds to step 3i through steps 3e and 3g, where the continuous number flag = 1 is set. Then, after incrementing the band number k in step 3k to set k = 1, the process returns to step 3c and the same processing is executed for the band k = 1.

Now, also in this band k = 1, it is assumed that the band-specific difference tmp exceeds the threshold value “1” following the band k = 0. Then, since significant value calculation unit 27 has already flag = 1, the process proceeds from step 3e to step 3f, where
y = y + suby (m, k-1)
Perform the following operation. That is, let suby (m, k-1) in the band k = 0 be a significant value y. Then, after setting the continuous number flag = 2, the process proceeds to step 3h through step 3g.
y = y + suby (m, k)
Thus, the suby (m, k) obtained in the current band k = 1 is added to the suby (m, k-1) in the band k = 0. In step 3k, the band number k is further incremented to become k = 2, and then the process returns to step 3c to execute processing for the band k = 2.

  Thereafter, similarly, every time the band-specific difference tmp of the adjacent bands k = 2, k = 3, k = 4,... Exceeds the threshold value “1”, the suby (m, k) of that band is decreased by one. Are sequentially added to the significant value y obtained up to the band of, thereby obtaining the weighted addition value y of the band-specific difference tmp.

Note that, in any band k = i, when the band-specific difference tmp is equal to or less than the threshold value “1”, the significant value calculation unit 27 proceeds from step 3d to step 3j and resets flag to 0 here.
Then, when the processing for all 16 bands k = 0 to k = 15 constituting one frame is completed, the significant value calculation unit 27 proceeds from step 3m to step 3n, where the significant value y And the weighted subband difference suby (m, k) (k = 0, 1,..., 15) calculated for each band are output.

  Thus, for each frame, the weighted addition value y of the band-specific differences tmp of a plurality of continuous bands whose threshold value exceeds “1” is obtained, and this weighted addition value y is used for the voice weight calculation unit 28 described later. This is used for calculating the weight, that is, determining whether the frame is a speech frame, a noise frame, or an intermediate transition frame. That is, when the band-specific difference tmp exceeds the threshold value “1” in only one band, this is regarded as noise and eliminated, and the voice / noise / transient area in the voice weight calculation unit 28 is excluded. It is not used for judgment.

  When the weighted addition value y is supplied from the significant value calculation unit 27, the audio weight calculation unit 28 calculates the audio weight sp used for determining the noise suppression gain. The sound weight sp is a numerical value representing the degree of sound included in one frame in a range of 0 ≦ sp ≦ 6, and is calculated from the weighted addition value y. Note that sp = 0 represents a noise section, and sp = 6 represents a voice section.

FIG. 4 is a flowchart showing the procedure for calculating the voice weight sp in the voice weight calculator 28 and the processing content thereof.
The voice weight calculator 28 first resets the frame number m to 0 in step 4a, and then increments the group number m in step 4b. Next, in step 4c, the weighted addition value y is compared with an arbitrary threshold value “13”. If y <13, it is determined as a noise frame, and the process proceeds to step 4d.
sp (m) = sp (m-1) −0.5
Set to. On the other hand, if y ≧ 13, the process proceeds to step 4e where
z = (y−13) * 1.5 + 1
Calculate That is, the voice weight z is temporarily set in the range of 1 to 6 based on y.

  Subsequently, the voice weight calculation unit 28 determines sp (m−1) ≦ 0.5 in Step 4f. That is, it is determined whether or not the voice weight sp (m−1) one frame before is a noise frame. If it is a noise frame, the process proceeds to step 4g, where the speech weight sp (m) of the current frame is set to z. On the other hand, if the speech weight sp (m−1) one frame before is not a noise frame, the process proceeds to step 4h where z> sp (m−1) +0.5 is determined, and z If> sp (m-1) +0.5, the voice weight sp (m) of the current frame is set to sp (m-1) +0.5 in step 4i. On the other hand, if z> sp (m-1) +0.5, the process proceeds to step 4j, where z> sp (m-1) -0.5 is determined, and z> sp (m-1) -0.5. If there is, the voice weight sp (m) of the current frame is set to sp (m-1) -0.5 in step 4k. If z> sp (m-1) −0.5 is not satisfied, the process proceeds to step 4m where the voice weight of the current frame is set to sp (m) = MIN (sp (m), 6) or sp (m ) = Set to MAX (sp (m), 0).

  That is, in step 4f to step 4m, the temporary audio weight z calculated in the current frame is corrected in consideration of the audio weight sp (m-1) set in the previous frame. Therefore, by using the speech weight sp (m) obtained in this way, it is possible to determine speech / noise / transient range in consideration of continuity between frames.

The speech weight sp (m) obtained by the speech weight calculation unit 28 is output in step 4n and input to the noise minimum value estimation unit 29 and the band-specific gain determination unit 33.
The noise minimum value estimation unit 29 checks the minimum value of the noise leakage integral value noise_power (m, k) in each band for every 100 frames in which the voice weight is sp = 0. Then, this minimum value is used as the minimum noise value noise_min (m, k) in the next 100 frames. At the same time, the average value min_all of the minimum noise values in each band is obtained.

5 and 6 are flowcharts showing the procedure and contents of the minimum value estimation process executed in the noise minimum value estimation unit 29.
In the figure, the noise minimum value estimation unit 29 first resets the frame number m to m = 0 in step 5a, sets the frame counter value to fc = 96, and sets the noise minimum value to noise_min (k) = 36. Bands are initially set to k = 0,..., 15, respectively, and noise_min (k) _h (k) = MAX (noise_power (m, 2k), noise_power (m, 2k + 1)), k = 0,. 7 shows the average noise minimum value between bands min_all.
7
min_all ＝ Σ noise_min (k) _h (n) / 8
n = 0
Initialize each.

  Next, the noise minimum value estimation unit 29 increments the frame number m in step 5b, and then determines in step 5c whether or not the speech weight is sp = 0, that is, whether or not it is a noise frame. If it is a noise frame, the process returns to step 5b to increment the frame number m, and the noise frame is determined in step 5c. In other words, the audio frame or the transient frame is detected by the above steps 5b and 5c.

When a voice frame or a transient frame is detected, the minimum noise estimation unit 29 proceeds to step 5d, where the frame counter fc is incremented and the band k = 0 is selected. And in step 5e
x = MAX (noise_power (m, 2k), noise_power (m, 2k + 1))
After moving to step 5f, it is determined whether or not noise_min (k) _h (k)> x. If noise_min (k) _h (k)> x, the process proceeds to step 5g where Set the noise minimum value to noise_min (k) _h (k) = x. Then, the process proceeds to step 5h.

  On the other hand, if noise_min (k) _h (k)> x is not satisfied, the process proceeds directly to step 5h to select the next band k = 1, and until the band k = 8 is reached, the noise generated in steps 5e to 5g. The setting process of the minimum value noise_min (k) _h (k) is repeated.

  When the band k reaches 8, the noise minimum value estimation unit 29 determines whether or not the frame counter fc reaches 100 in step 5j. Until the number of frames reaches 100, the process returns to step 5b to select the next frame, and the processes in steps 5c to 5i are repeated for the selected frame.

On the other hand, when the processing for the 100 frames is completed, the noise minimum value estimation unit 29 proceeds to step 6a shown in FIG. 6, where the noise minimum value interband average min_all is calculated.
7
min_all ＝ Σ noise_min (k) _h (n) / 8
n = 0
Ask for.

Along with that, noise_min (0) and noise_min (1)
noise_min (0) = noise_min_h (0)
noise_min (1) = 0.75 noise_min_h (0) +0.25 noise_min_h (1)
And the bandwidth is k = 1.

Further, the noise minimum value estimation unit 29 proceeds to step 6b, where the remaining bands k = 8 to k are based on the eight noise minimum values previously obtained for the bands k = 0 to k = 7. Noise minimum value for = 15
noise_min (2k) = 0.75 noise_min_h (k) +0.25 noise_min_h (k-1)
noise_min (2k + 1) = 0.75 noise_min_h (k) +0.25 noise_min_h (k + 1)
Calculate as follows.

When the above calculation is completed, the noise minimum value estimation unit 29 proceeds from step 6d to step 6e, where
noise_min (14) = 0.75 noise_min_h (7) +0.25 noise_min_h (6)
noise_min (15) ＝ noise_min_h (7)
Is calculated.

That is, the noise minimum value estimation unit 29 interpolates 16 min_alls based on the 8 min_alls in Steps 6a to 6e.
After calculating 16 min_all, the noise minimum value estimation unit 29 resets the frame counter fc to 0 in step 6f, sets the noise minimum value to noise_min_h (k) = 36, and sets the band to k = 0. , ..., reset to 7. Then, in step 6g, the average value min_all of the minimum noise values calculated previously and the minimum noise values noise_min (m, k), k = 0,..., 15 are output, and then the process returns to step 5b to return to the next. The same processing for calculating the minimum noise value and the average value between the bands is repeated for the frame m = m + 1.

In addition, the update determination unit 31 and the noise leak integration value update unit 32 perform update processing of the noise leak integration value noise_power (m, k). That is, the update determination unit 31 determines that updating is possible when y <15, and updating is not possible otherwise. When updating is possible, the noise leak integrated value updating unit 32 sets the noise power noise_power to, for example,
noise_power (m + 1, k) = noise_power (m, k) * 0.9 + channel_power (m, k) * 0.1, k = 0, ..., 15
Update like this.

  The band-specific gain determination unit 33 includes the band power channel_power (m, k) output from the band power calculation unit 26, the noise power noise_power (m, k) output from the noise leak integration value update unit 32, and a voice weight calculation. Based on the speech weight sp (m, k) output from the unit 28 and the noise minimum value noise_min (m, k) output from the noise minimum value estimation unit 29, the gain by band gain (m, k) is obtained. decide.

First, the band average value noise_all of the noise leak integrated value noise_power (m, k)
15
noise_all = Σ noise_power (m, k) / 16
k = 0
Ask for.

Subsequently, the band minimum value min_band of the band power channel_power (m, k) and the band maximum value max_band of the noise minimum value noise_min (m, k), respectively,
min_band = MIN (channel_power (m, k), k = 2,…, 11
max_band = MAX (noise_power (m, k), k = 0,…, 15)
Ask for.

Next, the adjustment value md common to all bands
md = (noise_all−min_all) * (1−sp / 6) + (min_band−max_band) * sp / 6
Determined by According to this formula:
When sp = 0, that is, in the noise interval, md = noise_all−min_all
sp = 6, that is, md = min_band−max_band when the voice interval
Thus, it can be seen that the transition region takes an intermediate value between these.

Examples of frequency versus power characteristics in the case of a noise frame and in the case of a speech frame are shown in FIGS. 7 and 8, respectively.
In the noise frame, as shown in FIG. 7, the band power is close to the noise minimum value. The value obtained by adding the adjustment value to the minimum noise value is obtained by changing the average value to the noise power average value noise_all without changing the spectral characteristics of the minimum noise value.

On the other hand, in the case of an audio frame, as shown in FIG. 8, the value obtained by adding the adjustment value to the minimum noise value is the same as the minimum value of the band power while the minimum spectral characteristics remain unchanged. Will be adjusted as follows.
The gain by band gain (m, k) is determined from the band power channel_power (m, k), the minimum noise value noise_min (m, k), and the adjustment value as follows. That is, in the band k,
tmp = channel_power (m, k) −noise_min (m, k) −md−1.625
gain (m, k) = {sqrt (1.4 + 0.49 * tmp2) + 0.7 * tmp−9.5} * 2
Are independently obtained for k = 0, ..., 15.
In addition, since it is noise when Sp = 0, the gain may be set to a constant in all bands regardless of the above formula, for example, gain (m, k) = − 20.

  Then, the gain by band gain (m, k) obtained as described above is multiplied by the conversion coefficient for each band in the multiplier 23, and noise cancellation is thereby performed. The noise-canceled transform coefficients for each band are subjected to inverse fast Fourier transform in IFFT 24 and returned to the signal frame on the time axis, and then frame-synthesized in the frame synthesizer 25 to the speech encoding circuit 16. Supplied.

  As described above, according to this embodiment, the noise minimum value estimation circuit 29 obtains the minimum value of the noise power of each band, and the spectrum shape of this noise minimum value is determined by the band-specific gain determination unit 33. Therefore, it is possible to realize a noise canceling process that is not affected by a short-term change in the noise spectrum, such as when passing through an automobile, and that hardly distorts the audio spectrum.

  In addition, according to this embodiment, the significant value calculation unit 27 obtains a weighted addition value of a difference for each band of a plurality of continuous bands whose threshold value exceeds “1”, and uses the weighted addition value as a voice weight calculation unit 28. Is used for the calculation of the voice weight in, that is, whether the frame is a voice frame, a noise frame, or an intermediate transition frame. For this reason, when the difference between the bands exceeds the threshold value “1” with only one band, it can be regarded as noise and eliminated, thereby accurately determining the voice / noise / transient range. Thus, noise cancellation performance can be improved.

The present invention is not limited to the above embodiment. For example, in the above-described embodiment, the digital mobile phone device adopting the TDMA system has been described as an example. However, the present invention can also be applied to a digital mobile phone device adopting the CDMA system.
In addition, the processing procedure and processing contents of each functional unit in the noise canceller and the circuit configuration or processing program for realizing the processing can be variously modified without departing from the gist of the present invention.

  DESCRIPTION OF SYMBOLS 1 ... Antenna, 2 ... Antenna duplexer (DUP), 3 ... Reception circuit (RX), 4 ... Frequency synthesizer (SYN), 5 ... Transmission circuit (TX), 6 ... Digital demodulator (DEM), 7 ... Digital modulation Unit (MOD), 8 ... time division multiple access circuit (TDMA), 9 ... error correction code decoding circuit (CH-COD), 10 ... voice decoding circuit (DEC), 11 ... D / A converter, 12 ... speaker, DESCRIPTION OF SYMBOLS 13 ... Microphone, 14 ... A / D converter, 16 ... Voice coding circuit (COD), 17 ... Noise canceller, 18 ... Control circuit, 19 ... Operation panel part, 21 ... Frame division part, 22 ... Fast Fourier transform part ( FFT), 23 ... multiplier, 24 ... inverse fast Fourier transform (IFFT), 25 ... frame synthesis unit, 26 ... band power calculation unit, 27 ... significant value calculation unit, 28 ... voice weight calculation unit, 29 The minimum noise value estimation unit, 31 ... update determination unit, 32 ... Noise leaky integration value updating unit, 33 ... band-specific gain determination unit.

Claims (5)

In the noise canceller that divides the input signal into frames of a certain time length, divides the signals of these frames into a plurality of frequency bands, and performs noise suppression processing for each of these frequency bands,
Noise power estimating means for obtaining signal power for each frequency band and estimating noise power for each band based on the band power;
Minimum value detecting means for detecting a minimum value of power over a plurality of frame periods for at least one of the band power and the noise power for each band;
For each frequency band, a difference between the band power and the minimum value for each band detected by the minimum value detecting means is obtained, and a suppression amount determining means for determining a noise suppression amount for each frequency band based on the difference. A noise canceller characterized by comprising. Further comprising means for determining a speech interval and a noise interval for the input signal;
The suppression amount determination means performs a noise suppression amount determination process for each band based on a difference between the band power and the minimum value for each band in at least a speech section of a speech section and a noise section. Item 2. The noise canceller according to Item 1.   The suppression amount determination unit further includes an adjustment value generation unit that generates an adjustment value common to different bands for each frame, and for each frequency band, the band power and the band detected by the minimum value detection unit The noise suppression amount for each frequency band is determined based on a difference between a minimum value and a value obtained by adding an adjustment value common to the band generated by the adjustment value generation unit. The noise canceller according to 1.   The adjustment value generation means determines an adjustment value common to the band based on the difference between the average value between the minimum values for each band and the average value between the noise powers for each band in the noise section, 4. The band common adjustment value is determined based on a difference between a minimum value among a plurality of band powers in one frame and a maximum value among a plurality of minimum values for each band. Noise canceller. In a communication apparatus that encodes and transmits a transmission input signal by a voice encoding unit,
The transmission input signal is divided into frames of a certain time length, and the signals of these frames are divided into a plurality of frequency bands, respectively, and a noise canceller that performs noise suppression processing for each of these frequency bands is provided,
The noise canceller is
Noise power estimating means for obtaining signal power for each frequency band and estimating noise power for each band based on the band power;
Minimum value detecting means for detecting a minimum value of power over a plurality of frame periods for at least one of the band power and the noise power for each band;
For each frequency band, a difference between the band power and the minimum value for each band detected by the minimum value detecting means is obtained, and a suppression amount determining means for determining a noise suppression amount for each frequency band based on the difference. A communication device comprising:

Images