WO2018116944A1

WO2018116944A1 - Audio noise detection device, digital broadcast receiving device, and audio noise detection method

Info

Publication number: WO2018116944A1
Application number: PCT/JP2017/044832
Authority: WO
Inventors: 勇哉西牟田; 高木　和也
Original assignee: 三菱電機株式会社
Priority date: 2016-12-20
Filing date: 2017-12-14
Publication date: 2018-06-28
Also published as: JPWO2018116944A1; JP6669277B2

Abstract

An audio noise detection device (10), which detects audio noise even for digital audio signals having noise components wherein strong components are distributed over a wide range of frequency, is characterized by being provided with: an audio signal input unit (101) for inputting a digital audio signal received; a segmented audio signal generating unit (1021) for generating a segmented audio signal from the digital audio signal on the basis of a defined time interval; a high-frequency component extraction unit (1022) for extracting high-frequency components in the frequency spectrum signal from the segmented audio signal; a characteristic quantity calculation unit (1023) for extracting high-frequency values for component values from the high-frequency components in the frequency spectrum signal and generating audio characteristic data from values obtained by multiplying the component values and frequency values; and an audio noise detection unit (103) for detecting noise components in the segmented audio signal from the audio characteristic data.

Description

Audio noise detection device, digital broadcast reception device, and audio noise detection method

The present invention relates to an audio noise detection apparatus for detecting audio noise in a digital broadcast receiving apparatus that receives a digital broadcast while moving.

Digital broadcast receivers that can receive and watch digital TV and digital radio while moving have become widespread. For example, in-vehicle digital broadcast receivers that can be viewed in a car or a car navigation system with a built-in digital broadcast reception function. It has become popular in recent years.

When the digital broadcast receiving apparatus is mounted on a moving body such as an automobile, the radio wave reception environment is affected by changes in the surrounding environment or high-speed movement, so that the radio wave received by the digital broadcast receiving apparatus may be disturbed. For example, when the digital broadcast receiving apparatus is located far away from the broadcasting station, the intensity of the radio wave reaching the receiving antenna from the broadcasting station becomes weak. When the digital broadcast receiving apparatus is in an urban area surrounded by high-rise buildings, unnecessary interference waves are included in the received radio waves due to the influence of reflection on the building walls. Further, when radio waves are received during movement, the amplitude variation of the received signal and the phase variation of the phase appear due to the influence of the Doppler shift.

Digital broadcasting is performed by applying an error correction code to data obtained by compressing video data and audio data by a compression method such as MPEG (Moving Picture Experts Group) on the transmission side. The receiving side can correct the error in the received data by performing error correction using the error correction code used on the transmitting side, but if there are many errors in the received data, the error correction may not be completed. For this reason, when this received data is decoded by the compression method used on the transmission side, an abnormality may occur in the decoding result and a part of the digital audio signal may be output as noise.

Therefore, the frequency spectrum signal obtained by converting the digital audio signal into the frequency domain is divided into a plurality of bands, and the noise generation section is detected for each divided band, and the audio signal in the corresponding section is corrected. There is a method of reducing the influence of (see, for example, Patent Document 1).

JP 2010-249939 (pages 6-36, FIG. 5)

However, the method of Patent Document 1 uses a frequency peak for noise detection, and thus has a problem that noise cannot be detected correctly when a strong component is distributed over a wide range in the frequency direction. .

The present invention has been made to solve the above-described problems, and is an audio noise that detects noise even for a digital audio signal having a noise component in which a strong component is distributed over a wide range in the frequency direction. An object is to provide a detection device.

In the audio noise detection device according to the present invention, an audio signal input unit that inputs a digital audio signal, an interval audio signal generation unit that generates an interval audio signal from the digital audio signal based on a set time width, and an interval audio High-frequency component extraction unit that extracts the high-frequency component of the frequency spectrum signal from the signal, and the voice feature based on the value obtained by extracting the high-frequency component value from the high-frequency component of the frequency spectrum signal and multiplying the component value by the frequency value A feature amount calculation unit that generates data and an audio noise detection unit that detects a noise component of the section audio signal from the audio feature data are provided.

Since the present invention extracts a frequency value having a high component value from a high frequency component of a frequency spectrum signal, generates voice feature data from the product of the component value and the frequency value, and detects a noise component. There is an effect that noise can be detected even for a digital audio signal having a noise component in which a strong component is distributed.

1 is a block diagram schematically showing a configuration of a digital broadcast receiving apparatus according to a first exemplary embodiment. 1 is a block diagram schematically showing a configuration of an audio noise detection apparatus according to a first exemplary embodiment. It is a figure which shows the relationship between the extraction area of the area audio signal production | generation part concerning Embodiment 1, and an overlap rate. FIG. 3 is a block diagram schematically showing a configuration of a high frequency component extraction unit according to the first exemplary embodiment. It is a figure explaining the boundary line determination in a support vector machine. 1 is a block diagram schematically showing a configuration of an audio signal processing unit according to a first embodiment; FIG. 3 is a diagram illustrating an example of processing of an audio signal processing unit according to the first embodiment. FIG. 6 is a diagram illustrating another example of the process of the audio signal processing unit according to the first embodiment. 3 is a flowchart illustrating an example of audio noise detection processing according to the first exemplary embodiment; It is a block diagram which shows roughly the structure of the audio | voice noise detection apparatus concerning Embodiment 2. FIG. It is a table | surface which shows an example of the relationship between the quality information concerning Embodiment 2, and an overlap rate. FIG. 6 is a block diagram schematically showing a configuration of a digital broadcast receiving apparatus according to a third exemplary embodiment. It is a block diagram which shows an example of a quality information map.

Embodiment 1 FIG.
FIG. 1 is a block diagram schematically showing a configuration of a digital broadcast receiving device including an audio noise detection device 10 according to the present exemplary embodiment. The digital broadcast receiving apparatus includes an audio noise detection device 10, a receiving unit 20, a demultiplexing unit 30, an audio decoding unit 40, an audio signal processing unit 50, and a control unit 60.

The receiving unit 20 receives and demodulates the selected digital broadcast radio wave. The receiving unit 20 may demodulate signals received from a plurality of antennas. Here, the digital broadcasting handled in the present embodiment compresses an audio signal, converts the compressed data into other data (for example, compression of data obtained by compressing a video signal) and multiplex processing (both multiplexing processing). It is assumed that the signal is transmitted after digital modulation. The other data is, for example, data obtained by compressing a video signal. As such digital broadcasting, not only ISDB-T (Integrated Services Digital Broadcasting-Terrestrial) which is a digital television broadcasting standard adopted in Japan, but also DVB-T (Digital Video Broadcasting-) which is a digital television broadcasting standard in Europe. Terrestrial), DTMB (Digital Terrestrial Multimedia Broadcast), which is a Chinese digital television broadcast standard, CMMB (China Mobile Multimedia Broadcasting), which is a broadcast standard for mobile terminals in China, and the like. Further, DAB (Digital Audio Broadcast) which is a digital radio broadcast standard may be targeted, or another digital broadcast standard may be targeted.

The demultiplex unit 30 performs demultiplex processing (also referred to as separation processing) on the demodulated data, acquires audio compression data, and supplies it to the audio decoding unit 40.

The audio decoding unit 40 performs a decoding process (also referred to as a decoding process) on the audio compression data from the demultiplexing unit 30 to generate a digital audio signal.

The audio noise detection device 10 receives a digital audio signal from the audio decoding unit 40 and detects a noise component of the audio signal. The detection method will be described later.

The audio signal processing unit 50 corrects the period of the noise component of the digital audio signal from the audio decoding unit 40 using the noise component information of the digital audio signal detected by the audio noise detection device 10 and outputs the audio signal. A digital audio signal is generated. The correction of the audio signal processing unit 50 will be described later.

The control unit 60 controls the operations and settings of the receiving unit 20, the demultiplexing unit 30, the audio decoding unit 40, the audio signal processing unit 50, and the audio noise detection device 10. For example, control is performed by transmitting information necessary for channel selection and information to be demultiplexed from the information to each component. Further, when the label set by the audio decoding unit 40 is managed, a control signal instructing association of unit audio signals is transmitted to the audio decoding unit 40, the audio signal processing unit 50, and the audio noise detection device 10.

FIG. 2 is a block diagram schematically showing the configuration of the audio noise detection apparatus 10 according to the present exemplary embodiment. The audio noise detection device 10 includes an audio signal input unit 101, an audio feature data generation unit 102, and an audio noise detection unit 103. The audio feature data generation unit 102 includes a section audio signal generation unit 1021, a high frequency component extraction unit 1022, and a feature amount calculation unit 1023. The audio noise detection unit 103 includes a noise identification information storage area 1031, an audio noise detection processing unit 1032, and a detection result storage area 1033. Here, the noise identification information storage area 1031 and the detection result storage area 1033 may be configured to be stored in a common storage unit.

The audio signal input unit 101 inputs the digital audio signal from the audio decoding unit 40.

The section audio signal generation unit 1021 generates a section audio signal by extracting based on the time width set for the digital audio signal input by the audio signal input unit 101. Here, for example, when performing FFT (Fast Fourier Transform), the time width may be set as a time width corresponding to a power of 2 in a sampling unit.

Also, the section audio signal generation unit 1021 may set and extract a section that overlaps in the time direction with the previous extraction section when extracting the section sound signal from the digital audio signal. That is, the section voice signal may be extracted based on a section in which continuous section voice signals overlap in the time direction and the set time width. FIG. 3 is a diagram illustrating the relationship between the extraction interval of the interval audio signal generation unit 1021 and the overlap rate. FIG. 3 shows a situation in which a first interval, a second interval, and a third interval are sequentially extracted from the input digital audio signal with a time width L and an overlap rate R_o (0 ≦ R_o ≦ 0.5). ing. As shown in FIG. 3, the overlapping section between the respective extraction sections is a section obtained by multiplying the time width L by the overlap rate R_o.

FIG. 4 is a block diagram schematically showing the configuration of the high frequency component extraction unit 1022. The high frequency component extraction unit 1022 extracts a high frequency component of the frequency from the section audio signal. The high-frequency component extraction unit 1022 includes a high-

pass filter

10221 or 10223 that is a high-pass filter that removes components in the low-frequency region, and a frequency-domain conversion unit 10222, and the section generated by the section audio signal generation unit 1021 Only high frequency components are extracted from the audio signal. Thus, a high frequency component of the frequency spectrum signal, that is, a power value corresponding to the frequency (hereinafter also referred to as a component value) is acquired. Here, the extraction of the high frequency component may be performed by converting the frequency of the digital audio signal from which the low frequency component has been removed first as shown in FIG. 4A, or by converting the frequency first as shown in FIG. 4B. The low-frequency component may be removed after performing the above. The range of the low frequency to be removed is only required to suppress normal audio signal components. For example, in the case of human speech, a frequency region of 4,000 Hz or lower including main components may be suppressed. In this case, the high frequency component is a component in a frequency region exceeding 4,000 Hz. That is, the high-frequency component is a component in a frequency region higher than the frequency region in which main components of human speech are included.

The feature amount calculation unit 1023 calculates a feature amount from the high frequency component of the frequency spectrum signal received from the high frequency component extraction unit 1022. The feature amount calculation unit 1023 extracts the top N samples (N: natural number) in descending order of the power value from the frequency spectrum signal, and acquires the extracted frequency and power value. Then, each power value is weighted with a frequency to calculate a feature amount F_noise, which is voice feature data.

An example of feature amount calculation processing is shown. Assuming that the extracted frequency is Df and the power value is Dp, in the feature amount calculation process, an average obtained by multiplying the power value Dp of each of the N samples by the frequency Df is used as the feature amount F_noise as shown in the following equation (1). calculate. Note that the calculation of the feature amount is an example, and the power value weighted by the frequency is not based on this formula alone.

As described above, the voice feature data generation unit 102 generates a section voice signal by extracting a digital voice signal based on the set time width, converts the section voice signal into a frequency spectrum signal, and A high frequency component of the frequency is extracted, a frequency value having a high component value is extracted from the frequency spectrum signal, and a feature amount F_noise in the section audio signal is generated by multiplying the component value and the frequency value.

Next, the audio noise detection unit 103 will be described. The audio noise detection unit 103 includes a noise identification information storage area 1031, an audio noise detection processing unit 1032, and a detection result storage area 1033. From the feature amount F_noise in the corresponding section audio signal generated by the audio feature data generation unit 102. Then, it is determined whether or not there is noise in the corresponding section audio signal.

The noise identification information storage area 1031 is an area for storing noise identification information, which is information used for noise detection, and is a partial storage area in a common storage unit even if it is an independent storage unit storage area. It does not matter. Here, the noise identification information is information of a discriminator that determines whether or not there is noise from the feature amount.

The audio noise detection processing unit 1032 determines whether there is noise or no noise from the feature amount F_noise based on the noise identification information.

For example, when the discriminator is a linear discriminator, the discriminator is represented by a discriminant polynomial D shown in the following equation (2).

Here, F_noise is a feature quantity, A and B are coefficients constituting the discriminant polynomial D, and M is the number of dimensions of the feature quantity. The noise identification information storage area 1031 stores information on the coefficients A and B of each dimension in the discriminant polynomial D.

The discriminant polynomial D is determined by an algorithm for constructing a linear classifier called a support vector machine, for example. In the support vector machine, a discriminant polynomial D for identifying the presence / absence of noise is determined using a feature quantity labeled with / without noise, that is, learning data.

FIG. 5 is a diagram for explaining the determination of the boundary line that distinguishes two classes (class 1 and class 2) from the two-dimensional feature quantity by the support vector machine. The support vector machine determines the boundary based on the idea of margin maximization in order to optimally separate the two classes from the two-dimensional feature quantity. Margin maximization is to maximize the margin (distance) between classes. In the example of FIG. 5, the straight line that maximizes the distance from both the point X and the point Y is defined as the discriminant polynomial D. Ask. In the example of FIG. 5, the values of the feature quantity 1 and the feature quantity 2 are substituted into the discriminant polynomial D shown in the equation (2), and if the values are positive, the class 1 above the straight line and the value is negative. If there is, it is determined to be class 2 below the straight line.

In this embodiment, the discriminant polynomial D is determined by the support vector machine, and the coefficient A and the coefficient B are stored in advance in the noise identification information storage area 1031 as noise identification information in order to be used by the audio noise detection processing unit 1032. To do.

The audio noise detection processing unit 1032 acquires the coefficient A and the coefficient B, which are noise identification information held in the noise identification information storage area 1031, calculates the discriminant polynomial D shown in the equation (2), and determines whether the calculation result is positive or negative Determine the presence or absence of noise.

In this example, two classes are classified by the linear separation / identification plane using a support vector machine. However, an algorithm for constructing a nonlinear separation / identification plane may be used, or another algorithm such as a neural network may be used. Also good.

ノイズ The presence / absence of noise is determined for each section audio signal. Here, in order to obtain a noise detection result corresponding to the input digital audio signal, the detection result is accumulated in the detection result storage area 1033 until the noise determination of all the segment audio signals obtained by dividing the input digital audio signal is completed. After accumulating the detection results for the input digital audio signal, it is output to the outside. The method of accumulating the noise detection result may be a 1-bit signal represented by 0 or 1 that is ON only in the section audio signal noise generation section, or may be a list of the start time and end time of the noise generation section.

As described above, the audio signal processing unit 50 uses the information on the noise component detected by the audio noise detection device 10, and the noise component information detected by the audio noise detection device 10 for the digital audio signal from the audio decoding unit 40. Can be used to correct the period in which there was a noise component, and a digital audio signal for audio output can be generated.

FIG. 6 is a block diagram schematically showing the configuration of the audio signal processing unit 50. The audio signal processing unit 50 includes a buffer control unit 501, a past signal storage area 502, a corrected audio signal generation unit 503, and an audio signal correction unit 504, and a noise component detected by the audio noise detection device 10 from a digital audio signal. Perform correction based on.

When the buffer control unit 501 stores the digital audio signal from the audio decoding unit 40 in the past signal storage area 502 and performs correction from the corresponding noise detection result, the buffered control unit 501 converts the stored digital audio signal into the corrected audio signal generation unit 503. Output to.

The past signal storage area 502 is an area for storing a digital audio signal from the audio decoding unit 40, and may be a storage area of an independent storage unit or a part of a common storage unit. .

The corrected audio signal generation unit 503 inputs the digital audio signal stored from the past signal storage area 502 and the information of the noise component detected by the audio noise detection device 10, and corrects the section in which the noise component is detected. A corrected audio signal is generated. The audio signal correction unit 504 receives the corrected audio signal from the corrected audio signal generation unit 503. The audio signal correcting unit 504 switches the digital audio signal from the audio decoding unit 40 to the corrected audio signal and outputs the corrected audio signal in a section where noise is detected.

FIG. 7 is a diagram illustrating an example of processing of the audio signal processing unit. The upper diagram of FIG. 7 shows a digital audio signal from the audio decoding unit 40, and shows the result of detecting that there is a noise component in the audio noise detection device 10 in the section from time ta to time tb. The lower diagram of FIG. 7 shows the corrected audio signal that the corrected audio signal generation unit 503 has corrected for the section from time ta to time tb of the digital audio signal from the audio decoding unit 40. As shown in FIG. 7, a signal obtained by switching the time tb from the section time ta where the noise is detected to a signal having no amplitude is generated as a corrected sound signal. A signal without amplitude becomes a silence signal.

FIG. 8 is a diagram illustrating another example of the processing of the audio signal processing unit. The upper diagram of FIG. 8 is a digital audio signal from the audio decoding unit 40, and shows a result of detecting that there is a noise component in the audio noise detection device 10 in the section from time ta to time tb. The lower diagram of FIG. 8 shows a corrected audio signal that the corrected audio signal generation unit 503 has corrected for the section from time ta to time tb of the digital audio signal from the audio decoding unit 40. The section from time tc to time td is a section having the same length (set time width) as the section from time ta to time tb and indicating that there is no noise component from the audio noise detection device 10. The section from time tc to time td is immediately before the noise component is generated. As shown in FIG. 8, a digital audio signal having no amplitude at time td is copied from the section time tc indicated that the section has no noise component, and is replaced from the section time ta at which noise is detected to time tb. Is generated as a corrected audio signal. In particular, in a section where a sound having a small amplitude continues, the section having no noise component immediately before the generation of the noise component is repeated, thereby producing an effect that a sense of incongruity is reduced as compared with the silence signal.

FIG. 9 is a flowchart showing an example of the audio noise detection process. The audio signal input unit 101 inputs the digital audio signal from the audio decoding unit 40 (step S1). The section audio signal generation unit 1021 generates a section audio signal by extracting based on the time width set for the digital audio signal input by the audio signal input unit 101 (step S2). The high frequency component extraction unit 1022 extracts a high frequency component of the frequency from the section audio signal (step S3).

The feature amount calculation unit 1023 extracts a frequency value having a high component value from the high frequency component of the frequency spectrum signal received from the high frequency component extraction unit 1022, and obtains voice feature data from a value obtained by multiplying the component value and the frequency value. Generate (step S4). The audio noise detection unit 103 determines whether the corresponding interval audio signal has noise or no noise from the audio feature data in the corresponding interval audio signal generated by the feature amount calculation unit 1023, and the noise of the interval audio signal A component is detected (step S5). Then, the audio signal processing unit 50 corrects the digital audio signal from the audio decoding unit 40 for a period in which there is a noise component using information on the noise component detected by the audio noise detection unit 103, and outputs the audio. An audio signal is generated (step S6).

As described above, the audio noise detection apparatus 10 according to the present embodiment can detect noise even for a signal having a noise component in which strong components are distributed over a wide range in the frequency direction. Further, by correcting a section having a noise component detected by the audio noise detection device 10, it is possible to improve the quality of the digital audio signal output from the digital broadcast receiving device without outputting the detected noise component. It becomes possible.

In addition, the audio noise detection device according to the present exemplary embodiment can perform noise detection using, for example, a single processor and a recording unit such as a RAM (Random Access Memory) or an HDD (Hard Disk Drive). There is.

Embodiment 2. FIG.
The audio noise detection apparatus according to the second exemplary embodiment further includes a quality information input unit that inputs quality information, and a parameter determination unit that changes the overlap rate used in the section audio signal generation unit based on the input quality information. Prepare.

FIG. 10 is a block diagram schematically showing the configuration of the audio noise detection apparatus 11 according to the present embodiment. The audio noise detection device 11 further includes a quality information input unit 114 and a parameter determination unit 115 in addition to the audio signal input unit 101, the audio feature data generation unit 102, and the audio noise detection unit 103. About the component with the same code | symbol, since a structure and an effect | action are the same as the above, description is abbreviate | omitted.

The quality information input unit 114 inputs quality information. Here, the quality information is information related to the quality of the input digital audio signal, and is, for example, the CNR estimated from the radio wave intensity of the digital broadcast radio wave received and selected by the receiving unit 20 and information obtained at the time of demodulation. (Carrier to Noise Ratio) or SNR (Signal to Noise Ratio), and packet error rate.

The parameter determination unit 115 sets the overlap rate of the extraction section set by the section audio signal generation unit 1021 based on the input quality information. The overlap ratio is set to be larger as the input quality information is a value indicating that the reception state is poor.

FIG. 11 is a table showing an example of the relationship between the quality information and the overlap rate. In FIG. 11, an example is shown in which a threshold is set for the input quality information, the classification of the quality information is classified into binary values of “good (G)” or “bad (B)”, and the overlap rate is set. showed that. In FIG. 11, when the quality information classification is G (voice quality: good), the overlap rate is set to 0.0, that is, the section is divided without overlapping. On the other hand, when the quality information classification is B (voice quality: bad), the overlap rate is set to 0.5, that is, half of the divided section overlaps with the respective extracted sections.

It should be noted that the quality information classification is not limited to binary, and it is obvious that the number of classifications may be increased and set in stages. The lower the voice quality based on the quality information, the higher the overlap rate ( Set the overlap section longer) and extract. Alternatively, a final classification result may be determined based on a combination of results obtained by inputting a plurality of types of quality information and classified by respective threshold values, and an overlap rate may be set based on the final classification result.

Thus, it is possible to reduce noise detection miss by increasing the overlap period extracted by the section audio signal generation unit 1021 as the reception state is worse.

On the other hand, the better the reception state, the more efficiently noise can be detected by reducing the overlapping period extracted by the section audio signal generation unit 1021. For example, when noise detection is performed using one processor and a recording unit such as RAM (Random Access Memory) or HDD (Hard Disk Drive), the load on other processors can be reduced by reducing the load on the processor related to noise detection. There is an effect that a load can be assigned to the process execution.

Embodiment 3 FIG.
The quality information input by the audio noise detection apparatus according to the second embodiment is the quality information received and estimated by the digital broadcast reception apparatus, but the audio noise detection apparatus according to the third embodiment is external to the digital broadcast reception apparatus. The position information is input from, and the quality information is supplied to the audio noise detection device based on the quality information map stored in association with the position information and the quality information.

FIG. 12 is a block diagram schematically showing a configuration of a digital broadcast receiving apparatus provided with the audio noise detection apparatus 11 according to the present embodiment. The digital broadcast receiving apparatus includes an input unit (also referred to as a position information input unit) in addition to the audio noise detection device 11, the receiving unit 20, the demultiplexing unit 30, the audio decoding unit 40, the audio signal processing unit 50, and the control unit 60. 70 and a quality information map storage area 80. About the component with the same code | symbol, since a structure and an effect | action are the same as the above, description is abbreviate | omitted.

The input unit 70 acquires information regarding the environment in which radio waves are received from the outside. For example, position information is acquired from a moving body such as a car equipped with a digital broadcast receiver. The position information includes GPS (Global Positioning System) information acquired by an in-vehicle device such as a car navigation system.

The input unit 70 inputs position information from the outside, and supplies the quality information corresponding to the input position information to the audio noise detection device 11 based on the quality information map stored in the quality information map storage area 80. To do.

The quality information map storage area 80 is an area for storing location information and quality information in association with each other, and may be a storage area of an independent storage unit or a part of a storage area in a common storage unit. Absent.

The quality information includes the radio wave intensity when a mobile body such as a car equipped with a digital broadcast receiver is actually received, the CNR or SNR estimated from information obtained at the time of demodulation, and the packet error rate.

FIG. 13 is a block diagram showing an example of the quality information map stored in the quality information map storage area 80. In FIG. 13, the quality information map is map information divided in a grid pattern, and has quality information at positions corresponding to the grids. The quality information possessed by each grid holds, for example, the aforementioned binary values of “good (G) and bad (B)”.

For example, the quality information obtained by the digital broadcast receiving apparatus when it is located in a block having a quality information map is stored in the quality information map storage area 80 as quality information for the entire block. As a result, even when the signals are received at different positions in the same block, the quality information can be set using the quality information at the previously received position, so that the quality information can be obtained without always estimating the quality information.

Also, with respect to a block for which reception quality estimation has not been performed, the quality information of the block may be estimated from the quality information in the neighboring blocks and stored. Furthermore, when there is a difference between the result of re-estimating the reception quality in the same block and the quality information stored in the quality information map storage area 80, it may be stored again with new information.

In FIG. 12, the quality information map storage area 80 is stored on the digital broadcast receiving apparatus side. However, when the digital broadcast receiving apparatus is connected to the Internet, the quality information map storage area 80 is stored in a cloud such as an external server. It goes without saying that the intended purpose can be achieved even if the quality information in the input location information block is acquired by accessing the quality information map storage area 80 via the Internet.

In Embodiments 1 to 3 described above, part of the audio noise detection device and the digital broadcast reception device is realized by a processing circuit. The processing circuit may be dedicated hardware or a CPU that executes a program stored in a memory. For example, in FIG. 1, the functions of the receiving unit 20, the demultiplexing unit 30, the audio decoding unit 40, the audio signal processing unit 50, and the control unit 60 may be realized by separate processing circuits. The functions of a plurality of parts may be realized by a single processing circuit.

Similarly, in FIG. 2, the functions of the audio signal input unit 101, the audio feature data generation unit 102, and the audio noise detection unit 103 may be realized by separate processing circuits, or the functions of the above-described plurality of parts. May be realized by a single processing circuit.

When the processing circuit is a CPU, the functions of the plurality of parts described above are realized by software, firmware, or a combination of software and firmware. Software or firmware is described as a program and stored in a memory. The CPU implements the functions of the respective units by reading and executing the program stored in the memory. Also, some of the functions of the plurality of parts of the digital broadcast receiving apparatus may be realized by dedicated hardware, and the other part of the functions may be realized by software or firmware.

10, 11 audio noise detection device, 20 receiving unit, 30 demultiplexing unit, 40 audio decoding unit, 50 audio signal processing unit, 60 control unit, 70 input unit, 80 quality information map storage area, 101 audio signal input unit, 102 voice feature data generation section, 103 voice noise detection section, 114 quality information input section, 115 parameter determination section, 501 buffer control section, 502 past signal storage area, 503 correction voice signal generation section, 504 voice signal correction section, 1021 section Audio signal generation unit, 1022, high frequency component extraction unit, 1023 feature quantity calculation unit, 1031 noise identification information storage area, 1032 audio noise detection processing unit, 1033 detection result storage area, 10221, 10223 high pass filter , 10222 frequency domain conversion unit.

Claims

An audio signal input unit for inputting a digital audio signal;
An interval audio signal generation unit for generating an interval audio signal from the digital audio signal based on a set time width;
A high-frequency component extraction unit that extracts a high-frequency component of a frequency spectrum signal from the section audio signal;
A feature amount calculating unit that extracts a frequency value having a high component value from a high frequency component of the frequency spectrum signal and generates voice feature data from a value obtained by multiplying the component value and the frequency value;
An audio noise detection apparatus comprising: an audio noise detection unit that detects a noise component of the section audio signal from the audio feature data.
The high-frequency component extraction unit further includes a high-pass filter that removes a component in a low-frequency region using the frequency spectrum signal as an input,
The audio noise detection apparatus according to claim 1, wherein the feature amount calculation unit extracts a frequency value having a high component value from an output of the high-pass filter.
The section audio signal generation unit generates the section audio signal based on a section in which the continuous section audio signals overlap in the time direction and the set time width. Item 3. The audio noise detection device according to Item 2.
4. The audio noise detection apparatus according to claim 3, wherein the interval overlapping in the time direction is set longer as the audio quality indicated by the quality information of the digital audio signal is lower.
A receiver that demodulates the received radio wave to generate a digital audio signal;
The audio noise detection device according to any one of claims 1 to 3,
A digital broadcast receiving apparatus comprising: an audio signal processing unit that outputs a digital audio signal generated by performing correction based on a noise component detected by the audio noise detection device from the digital audio signal.
A receiver that demodulates the received radio wave to generate a digital audio signal;
An audio noise detection device according to claim 4;
An audio signal processing unit that outputs a digital audio signal generated by performing correction based on a noise component detected by the audio noise detection device from the digital audio signal;
The receiving unit generates the quality information using the radio wave intensity of the radio wave received by the receiving unit, information obtained by estimating a CN ratio from the demodulated signal, or a packet error rate obtained from the demodulated signal. A digital broadcast receiver characterized by the above.
A position information input unit for inputting position information;
The digital broadcast receiving apparatus according to claim 6, further comprising a storage unit that stores the position information and the quality information at the position in association with each other.
The audio signal processing unit is configured to switch between a silence signal or a digital audio signal having the time width immediately before the noise component is generated for a section in which the audio noise detection device detects a noise component in the section audio signal. The digital broadcast receiving apparatus according to claim 5, wherein correction is performed.
An audio signal input step for inputting a digital audio signal;
An interval audio signal generating step for generating an interval audio signal from the digital audio signal based on a set time width;
A high-frequency component extracting step of extracting a high-frequency component of a frequency spectrum signal from the section audio signal;
Extracting a frequency value having a high component value from a high frequency component of the frequency spectrum signal, and generating a feature value calculating step from the value obtained by multiplying the component value and the frequency value; and
An audio noise detection method comprising: an audio noise detection step of detecting a noise component of the section audio signal from the audio feature data.