KR100750584B1 - Signal processing method - Google Patents

Abstract

A method for processing signals is provided to reduce/effectively remove noise which is generated from audio broadcasting of territorial DMB(Digital Multimedia Broadcasting). A method for processing signals comprises the following steps of: confirming whether magnification components of an audio signal sample having error possibility are included in a predetermined range; and using the value as it is when the magnification components are included in the predetermined range and compensating the magnification components by using a first value decided on the basis of one of the predetermined range and the magnification components when the magnification components are included in the predetermined range.

Description

Signal processing method

1 illustrates an MPEG-1 / 2 Layer-2 coding method used for a radio service of terrestrial DMB broadcasting.

2 shows a comparison between an MPEG Layer-2 frame and a DAB frame.

3 illustrates an embodiment of the present invention for correcting a scale factor of a subband by concealment level 2,

4A and 4B illustrate other embodiments of the present invention for correcting the scale factor of a subband by concealment level 2,

5 shows another embodiment of the present invention for correcting the scale factor of a subband by concealment level 2,

6 illustrates an embodiment of setting and releasing a section to which an error concealment method according to the present invention is applied.

The present invention relates to a signal processing method, and more particularly, to a method for removing noise of a digital multimedia broadcasting (DMB) radio.

Recently, a digital multimedia broadcasting service is started, and a mobile phone including a DMB terminal or a DMB function has been widely used. Digital multimedia broadcasting is a broadcasting service that digitally modulates various multimedia data such as voice, video, and data, and provides a new concept of mobile multimedia broadcasting service combining broadcasting and communication.

In addition, it is called a next-generation broadcast service because it is possible to provide sound quality and data service of CD level and to watch multimedia broadcasts of various channels while moving through small terminals such as mobile phones, PDAs, notebook computers, and car receivers.

DMB is classified into terrestrial DMB (T-DMB: Terrestrial DMB) and satellite DMB (S-DMB: Satellite DMB) according to transmission method and network configuration. Satellite DMB uses satellites, which are broadcasted by satellite DMB broadcasting centers and distributed to DMB terminals nationwide through radio waves. The terrestrial DMB transmits a program using a frequency on the ground, and uses a VHF channel 12 which is currently empty, and unlike a satellite DMB, a broadcast signal is transmitted through a base station on the ground.

Terrestrial DMB is based on the European digital audio broadcasting standard EUREKA-147 Digital Audio Broadcasting (DAB), which adds specifications for providing video services using Moving Picture Experts Group (MPEG) -4 technology. Complemented and added to the data broadcast standard. In addition, the terrestrial DMB transmission / reception standard uses the DAB transmission standard, which adopts an orthogonal frequency division multiplexing (OFDM) transmission technology, which is a digital modulation method developed for DAB, to improve transmission rate per bandwidth and prevent multipath interference. Indicates mobile reception performance.

On the other hand, the terrestrial DMB radio service is based on DAB and uses the MPEG-1 / 2 Layer-2 codec. For example, a terrestrial DMB radio service receives a DMB in a mobile station or a shaded area where a broadcast signal does not reach. Etc., the audio bit stream may be lost or distorted during the transmission process, leading to deterioration of sound quality.

The DMB terminal receiving the terrestrial DMB radio broadcast may not receive a part of the bitstream or receive a distorted bitstream in a low reception period, and completely recover data lost or distorted during such transmission or reception. This loss and distortion causes noise to appear in the output audio and degrades the sound quality.

The present invention has been made in view of the above circumstances, and an object of the present invention is to remove or reduce noise of a DMB radio caused by data loss or distortion.

The present invention was created to achieve the above object, and the signal processing method according to an embodiment of the present invention comprises the steps of checking whether a magnification component for an audio signal sample, which is error-prone, falls within a predetermined range. ; And correcting the magnification component by using the value as it is if the magnification component falls within the predetermined range, or using a first value determined based on at least one of the predetermined range or the magnification component. Characterized in that comprises a step.

In one example of the present invention, the first value may be a value used as a reference for determining the predetermined range or a value close to the magnification component among the boundary values of the predetermined range. Alternatively, the first value may be used as a reference for determining the predetermined range and an average value of the magnification components. When the average value is out of the predetermined range, the first value is included in the predetermined range. A value close to the magnification component among the boundary values of may be used.

In one example of the present invention, the predetermined range may be determined based on the value of the magnification component at a time immediately preceding, or is located in front of and behind the magnification component which is likely to be in error, and there is no error. The predetermined range can be determined based on the values calculated by the interpolation based on the components as reference values.

In the above example, the predetermined range may be determined by a symmetric structure having the same width in both directions about the reference value or by an asymmetric structure having different widths in one direction and the other direction about the reference value. Can be. In the latter case, the asymmetric structure in which the ratio of the width in the one direction and the other direction is fixed is used in the predetermined range or in accordance with the relative position at which the reference value lies in the range of values that the magnification component may have. The ratio of the width in the direction and the other direction may vary.

Further, in the above example, the predetermined range may be determined as either a symmetrical structure or an asymmetrical structure depending on the position where the reference value lies in the range of values that the magnification component may have.

In one example of the present invention, the predetermined range of widths may be a fixed value or a value that is continuously updated based on a change in the magnification component value previously used without error. Can be updated based on the maximum, minimum, mean, and deviation of the difference between two magnification component values adjacent to each other, previously used. In addition, a value continuously updated for a predetermined time may be used for the width of the predetermined range, and a value updated last after the predetermined time may be used.

In accordance with another aspect of the present invention, there is provided a signal processing method comprising: determining whether an error probability is high for each section of a received audio broadcast; And correcting the audio sample on the basis of the characteristics of the audio sample decoded in the section determined to have a high probability of error.

In one example of the present invention, whether the error probability for the current section is high is based on at least one of a change in reception rate for a section already received or a continuity of a section in which an error occurs during decoding operation on already received audio data. Can be determined.

In addition, in one example of the present invention, whether to correct the audio sample may be determined by whether at least one of the energy or the magnitude of the audio sample is out of a predetermined allowable range. In this case, whether to correct the audio sample may be determined in units of bands separated by a predetermined number, and the audio sample may be corrected only for the high frequency band.

According to still another aspect of the present invention, there is provided a method of removing noise in a digital audio broadcast, comprising: determining an allowable range for a scale factor of a current time based on a scale factor used at a previous time; Confirming that a scale factor of the current time for any subband in the scale factor group where an error is detected falls within the determined tolerance range; And using the boundary value of the allowable range close to the scale factor of the current time as the corrected scale factor of the current time when the scale factor of the current time does not fall within the allowable range. .

Hereinafter, exemplary embodiments of a signal processing method according to the present invention will be described in detail with reference to the accompanying drawings.

Audio is cut off due to data loss and noise is generated at low and high frequencies due to data distortion. However, since the breakup phenomenon due to data loss does not occur frequently, low frequency and high frequency noise are mainly the main part of the sound quality degradation, and the breakdown phenomenon <low frequency noise <high frequency noise takes the share of sound quality degradation.

FIG. 1 illustrates an MPEG-1 / 2 Layer-2 coding method used for a radio service of a terrestrial DMB broadcast. The MPEG-1 / 2 Layer-2 coding will be briefly described with reference to FIG. 1.

Pulse Code Modulation (PCM) audio data, which corresponds to approximately 24 msec time, passes through a bandpass filter for each band in which audio frequencies are divided evenly (evenly in log scale) and 32 subbands (Sub -band) is temporarily stored in the bank, and each sub-band (n-band #n) data is sampled 36 times.

The 36 data in one subband are separated by waveform and magnification. Since the maximum amplitude is normalized to a predetermined value (for example, 1.0) in units of 12 data, and the magnification at that time becomes a scale factor, one subband has 36 scaled samples corresponding to a waveform. And three scale factors corresponding to the magnification. Here, the scale factor is determined by an arbitrary index of a scale factor table composed of 64 values ranging from 0 to 63. The larger the amplitude of the waveform of the corresponding subband, the closer the scale factor index becomes to zero.

The scale factor is a component related to the loudness of the sound, and when the scale factor is distorted, it greatly affects high frequency and low frequency noise. Sample data has a significant impact on low frequency noise when distorted with timbre and pitch (melody) components.

As such, the main cause of noise generated in the DMB audio service is the distortion of the subband scale factor and sample data. Therefore, noise can be minimized by correcting the distorted scale factor and sample data.

To this end, first, a method of detecting a distorted portion of the scale factor and sample data is required, and when it is detected, a method of correcting the distorted portion of the scale factor and sample data is needed.

FIG. 2 illustrates a comparison between an MPEG Layer-2 frame and a DAB frame. A SCF-CRC (Scale Factor) is used to detect an error that may occur in a scale factor in a DAB frame compared to an MPEG-1 / 2 Layer-2 frame. Cyclic Redundancy Check).

Therefore, in the DMB radio service based on DAB, error correction of the scale factor is possible through the ScF-CRC included in the audio frame.

The present invention provides a method for reducing or eliminating noise generated in an audio broadcast transmitted through a DMB audio channel, which detects an error or error while correcting an error while correcting the detected error while increasing a level of concealment. To improve the level.

1) Concealment level 1

Concealment level 1 is a technique that simply detects and corrects an error by applying an error detection technique defined in the DAB standard. The ScF-CRC is used to check and correct an error of a scale factor.

Distortion and error of the scale factor is detected using ScF-CRC information included in each frame and transmitted, and correction thereof is performed by replacing (copying) with the scale factor value of the previous frame as shown in FIG. 3. This significantly reduces high frequency noise.

However, because the ScF-CRC is used in units of scale factor groups, simply copying the scale factor of the corresponding group of the previous frame reduces the maximum value of the distortion but contaminates the normal scale factor around it. It can be seen that this is dispersed. In addition, even when ScF-CRC using 8-bit CRC, errors are frequently detected.

For reference, since a predetermined number, for example, scale factors of four subbands are managed as one group, even if an error occurs in one scale factor, all remaining scale factors may be replaced with the scale factors of the corresponding group of the previous frame. .

Figure 3 shows the correction for the scale factor of any one subband, where an error occurs in the scale factors SF1 to SF4 of frame #N and frame # (N + 1), and there is no error in SF5 and SF6. Since it is determined that an error has occurred in the group, a distortion variance problem occurs in which the scale factor value of the previous frame is copied as it is and changed to the wrong value even for SF5 and SF6 which have no error.

2) Concealment level 2

Concealment level 2 improves the dispersion of distortion, which is a problem of concealment level 1, and improves the scale factor error detection method and also the scale factor correction method.

Since one frame corresponds to a very short time of about 24 msec, and one frame is divided into three intervals to obtain a scale factor, there is almost no sudden sudden change in scale between those times (about 8 msec). The range of values that the current scale factor may have compared to the scale factor is somewhat limited.

Therefore, by using the characteristics of the scale factor, it is possible to confirm and correct the error through comparison with the previous scale factor for each scale factor in the group identified as an error.

Concealment level 2 checks whether an individual scale factor in the group in which the ScF-CRC error was detected is an error. For each scale factor in the group in which the error was detected, the previous scale factor value (for the scale factor at the beginning of the frame) The tolerance is determined based on the last scale factor value used in the previous frame, and the scale factor value just used for the scale factor in the middle or the end of the frame), and whether the scale factor is within the tolerance range. Check it.

For a scale factor that is within the allowable range determined by the previous scale factor value, it is determined that it is not an error or is an acceptable value and that value is used as it is. On the other hand, it is determined that the scale factor outside the allowable range is an error, and in one embodiment, the previous scale factor value that is the basis of the allowable range may be used as it is.

If the scale factor is outside the allowable range, in another embodiment, a value closest to the scale factor among the values within the allowable range, that is, a close boundary value of the allowable range may be used. In another embodiment, the mean value of the scale factor and the reference value of the tolerance range (e.g., the median value of the tolerance range) may be used as the scale factor corrected for the time, in which case the scale factor used May be located outside the allowable range. In another embodiment, an average value of a corresponding scale factor and a reference value of the allowable range is used as a scale factor corrected for the corresponding time, but the close boundary value of the allowable range when the average value is outside the allowable range. You can also use

In any of the embodiments, the value in the same direction as the probable scale factor (non-opposite value) based on the reference value of the allowable range, that is, the value of the probable scale factor and the reference value of the allowable range The value in between is used as the scale factor corrected for that time.

In FIG. 3, it is determined that an error occurs in the scale factors of frame #N and frame # (N + 1), that is, SF1 to SF6 by ScF-CRC error detection. Therefore, for SF1 to SF6, whether the value is used as it is or whether the scale factor used immediately before is copied is used depending on whether the scale factor is included in the allowable range determined based on the scale factor used immediately before. Is determined.

A range of values that SF1 may have, that is, an allowable range, is determined based on the SF0 value of the previous frame corresponding to SF1, which corresponds to R1 in FIG. 3. Since SF1 is included in R1 in a situation where it is not known whether SF1 is an error value or a normal value, SF1 is determined to be a normal scale factor value or a scale factor value included in a range of allowable values, and the value is used as a scale factor. .

Next, when judging about SF2, since SF2 is located outside the allowable range R2 determined by the value used as the scale factor immediately before (SF1 in FIG. 3), it is determined that an error has occurred in SF2 and the value is ignored and The value of the scale factor (SF1 which is the scale factor which is the basis of R2, which is the allowable range on which the error judgment is based) is copied and used.

Since SF3 and SF4 are located outside R2, which is the allowable range determined by SF1, similarly to SF2, it is determined that the error is ignored and the value thereof is ignored, and the previous scale factor value SF1 is copied and used.

On the other hand, since SF5 is within R2, which is an allowable range determined by the value used as the scale factor just before, it is determined that it is not an error or is an acceptable value, and the value is used as it is. In addition, since SF6 is also within the allowable range R3 determined by the value SF5 used as the scale factor just before, like SF5, it is determined that it is not an error or is an acceptable value and the value is used as it is.

In FIG. 3, only an embodiment in which the previous scale factor value that is the basis of the allowable range is used as it is when the scale factor is outside the allowable range. However, another embodiment using the boundary value of the allowable range or the average value of the scale factor with error of the reference value of the allowable range may be sufficiently described without change using the example of FIG. 3. .

On the other hand, the width of the permissible range may be determined experimentally or statistically so that a fixed value, for example, from 3 to +3 in a scale factor table from 0 to 63 may be used. Or continue to check the change in the scale factor value from the frame in which no error is detected among the previously decoded frames, that is, the maximum, minimum, average, and deviation of the differences between two adjacent scale factor values. The width of the allowable range may be continuously updated based on the result. Alternatively, the width of the allowable range is updated based on the change in scale factor value from a frame in which no error is detected among previously decoded frames for a predetermined time, and after the predetermined time, the last updated value is updated within the allowable range. It can also be used in width.

In addition, although it has been described that a reference value of an allowable range, for example, a value used in a previous scale factor as a center value, may be used as the basis of an error for individual scale factors in a group determined to have an error, but in another embodiment, The reference value of the allowable range may be determined using the scale factors of the corresponding groups of the front and back frames where no error is found, in which case the allowable range may be set to be slightly narrower than in the previous embodiment.

FIG. 4A illustrates a case where an error occurs in a scale factor of any group of frame #N, and there is no error in frame # (N-1) and frame # (N + 1). Another embodiment of the present invention uses frame #N by using the last scale factor SF0 of the corresponding group of frame # (N-1) and the first scale factor SF7 of the corresponding group of frame # (N + 1). The scale factor values (I1, I2, I3) corresponding to the times (t1, t2, t3) are temporarily calculated by an interpolation method, and the allowable ranges (R4, R5, R6) are obtained as reference values. If it is within the allowable range, it checks whether the scale factor (decoded scale factor including error) of the corresponding group of frame #N is included in the allowable range, and uses the scale factor if it is included in the range. (I1, I2, I3) are used as it is.

In FIG. 4A, since the decoded scale factor of frame #N is not included in the allowable range, scale factors I1 and I2 calculated using the scale factors of frame # (N-1) and frame # (N + 1). , I3) is used.

In FIG. 4B, the scale factor of a neighboring frame having no error as a reference value of an allowable range, which is the basis for determining whether each decoded scale factor value is an error (whether or not a decoded scale factor value is included in a corresponding allowable range). Using the value calculated for the time using or the scale factor used in the previous time may be used as a reference value.

In FIG. 4B, among the decoded scale factors of frame #N, for a scale factor of time t1, a value calculated using a corresponding allowable range R4 (scale factor of an adjacent error-free frame (a value corresponding to the corresponding time t1). Scale factor I1 calculated for t1 is used, and for the scale factor at time t2, the allowable range R5 (I2 is set as the reference value). The value (decoded value) is used as it is.

However, for the scale factor of time t3, it is determined whether or not it is within the allowable range R7 obtained based on the scale factor value used for the time t2 (the decoded scale factor is used for time t2). Since the decoded scale factor of t3 time is included in the allowable range R7, the value is used as it is. Alternatively, for the scale factor of t3 time, as in FIG. 4A, an allowable range in which I3 (a value calculated for t3 time using the scale factor of an error-free neighboring frame) is defined as a reference value (R6 in FIG. 4A). ) May be determined whether or not an error.

In Figures 3 and 4, a symmetric structure centering on the reference value, i. An example is given where the permissible range of is used to correct a scale factor that is prone to error.

However, the present invention is not limited thereto, and the width of the allowable range may be different from the reference value in the upward direction and the downward direction, that is, the allowable range of the asymmetric structure may be used. Depending on the scale factor value (the position of the reference value in the scale factor table of 0 to 63 indexes), the width of the allowable range may vary in the up and down directions.

As a simple example, divide the index between 0 and 63 into two sections centered on the median 31 or 32.If any value in the upper section is used as the reference value, make the width in the upper direction smaller than the width in the lower direction. When any value of is used as a reference value, the width in the downward direction may be smaller than the width in the upward direction.

Alternatively, the index of 0 to 63 is divided into two sections, for example, around 16, and the reference value of the upper section uses the allowable range of the symmetrical structure having the same width in the upper direction and the lower direction, and the reference of the lower section. For a value, you can also use a tolerance of an asymmetric structure where the width in the downward direction is smaller than the width in the upward direction, which is caused by a scale factor error by stricting the tolerance in the downward direction (by making its width smaller than the upward direction). This is because the larger the scale factor index is to 0, the louder the amplitude of the waveform becomes.

As another example, the index of 0 to 63 is divided into three intervals, for example, if any value of the middle interval is used as the reference value, the allowable range is symmetrical, and any value of the upper interval is used as the reference value. If the width of the upper direction is smaller than the width of the lower direction, and if any value of the lower section is used as a reference value, the width of the lower direction may be smaller than the width of the upper direction.

In this case, the 0 to 63 index may be divided into three equal parts so that the section to which the allowable range of the symmetric structure is applied and the section to which the allowable range of the asymmetric structure is applied are the same size, or the middle section is wider than the upper and lower sections. You can also divide the index from 0 to 63.

In addition, when the 0 to 63 scale factor index is divided into 2 or 3 sections and the allowable range of the symmetric and / or asymmetric structure is used, the degree of asymmetry of the allowable range is constant regardless of the reference value in each section ( For example, the ratio of the width of the up (or down) direction to the down (or up) direction may be fixed at 3: 2, or the degree of asymmetry of the tolerance may vary depending on the reference value (e.g., As the value approaches 0 or 63, the degree of asymmetry increases.

FIG. 5 illustrates an example of correcting a potential scale factor using an allowable range of an asymmetric structure according to another embodiment of the present invention.

In FIG. 5, the scale factor index of 0 to 63 is divided into three intervals. For the reference value of 0 to 15, an allowable range of an asymmetric structure having a lower width than that of the upper direction is applied (Asymmetric section # 1). For a reference value of 48 to 63, a permissible range of asymmetric structures with a smaller width in the upper direction than that in the downward direction is applied (Asymmetric section # 1), and a width and an upward direction for the reference values 16 to 47. The allowable range of symmetrical structures with the same width is applied (Symmetric section).

5, the allowable range of the symmetrical structure is 3 in the up and down directions, the allowable range of the asymmetrical structure is 2 in one direction and 3 in the other direction, regardless of the position of the reference value. The degree of asymmetry is fixed, of course, the allowable range is set within 0 and 63.

The example of FIG. 5 is for illustrative purposes only, and other forms of modification are possible.

In FIG. 5, it is determined that an error occurs in the scale factors of frame # (N-1) and frame # (N + k + 1) by ScF-CRC error detection. The scale factor used at time t0, which is the reference value that defines the permissible range to be used for t1, belongs to an interval (Asymmetric section # 1) to which the permissible range of asymmetric structures is applied, with a width of 2 in the downward direction and 3 width in the upward direction. .

Therefore, it is checked whether the scale factor of the t1 time is within an allowable range R11, which is 2 down and 3 up from the reference value, since the scale factor of the t1 time is not within the R11 and is located in the outer downward direction. According to an embodiment of the present invention, the lower boundary value (value indicated by x) of R11 is used as a corrected scale factor at the time t1.

Since the scale factor corrected and used in the t0 time belongs to the Asymmetric section # 1, the scale factor of the t2 time is 2 and the width in the up direction is based on the scale factor used in the t0 time. It is checked whether or not the width is within the allowable range R12 of the asymmetric structure of 3, and since the scale factor of the t2 time is located in the R12, the value is used as it is.

The scale factor of time t3 is the scale used at time t2 because the scale factor used at time t2, which is a reference value that defines the allowable range to be used at t3, belongs to a symmetric section to which the tolerance range of symmetry is applied. It is checked whether it is within the allowable range R13 of the symmetrical structure having a width of 3 in the up and down directions based on the factor.

Since the scale factor of the t3 time is not in the R13 but located outward, in the t3 time, the upper boundary value (value indicated by x) of the R13 is corrected in the t3 time according to an embodiment of the present invention. Used.

On the other hand, since the scale factor used at time t4, which is a reference value for determining the allowable range to be used at t5, belongs to a symmetric section to which the allowable range of symmetry is applied, the scale factor of time t5 is used at time t4. Based on the scale factor, it is checked whether or not it is within an allowable range R14 of an asymmetric structure having a width in the up and down directions of three. Since the scale factor of the t5 time is not in the R14 and is located outward, in the t5 time, the upper boundary value (value indicated by x) of the R14 is corrected in the t5 time according to an embodiment of the present invention. Used.

The scale factor of t6 time is corrected and used because it belongs to an interval (Asymmetric section # 2) to which an allowable range of an asymmetric structure with a width of 2 in the upper direction and a width of 3 in the downward direction is applied. Is a reference value based on the scale factor used at time t5 is determined whether or not within the allowable range (R15) of the asymmetric structure having a width in the upper direction is 2 and a width in the lower direction is 3, the scale factor of the t6 time is Since it is located in R15, the value is used as it is.

The scale factor of time t7 is the scale used at time t6, because the scale factor used at time t6, which is a reference value that determines the allowable range to be used at t7, belongs to a symmetric section to which the tolerance range of symmetry is applied. It is checked whether or not the value is within the allowable range R16 of the symmetrical structure having a width of 3 in the up and down direction based on the factor value. Since the scale factor of t7 time is within the value R16, the value is used as it is.

According to this Concealment level 2, it is determined that an error-free scale factor is considered to be an error, thereby reducing the dispersal of distortion that is changed to an incorrect value. As the error of the scale factor decreases, the degree of elimination of high frequency noise is improved. The tone is improved.

However, even at concealment level 2, the limitation of ScF-CRC described above as concealment level 1 problem still remains, and there is an error that ScF-CRC cannot detect.

3) Concealment level 3

At concealment level 3, in order to overcome the limitation of ScF-CRC, which is a problem of concealment levels 1 and 2, subband samples are used to detect and correct errors.

There is no way to detect and correct errors in the scale factor that the ScF-CRC cannot detect. Therefore, at Concealment level 3, scale factor x sampled data is defined as a sub-band sample, and the characteristics of the sub-band sample are used, for example, amplitude and energy. Set the error tolerance for the error and correct subband samples outside the tolerance. That is, it is possible to reduce the size or energy of subband samples that are too large or too high in energy.

In this case, since an error-free component may also be distorted, it is mainly applied only to a high frequency component. In addition, it is necessary to limit the section to which error detection and correction according to Concealment level 3 is applied to compensate for the disadvantage that the error-free component is distorted.

The section to which Concealment level 3 is applied can be set using the reception rate information input from the tuner or modem, or the section where the error is concentrated can be estimated while monitoring the error information such as ScF-CRC when decoding the frame. have.

As shown in Fig. 6, when the state where the reception rate is lower than the predetermined value (or the error rate is higher than the predetermined value) lasts for a predetermined time or when the frames in which an error occurs continuously exceeds a predetermined number, an error occurs for subsequent frames. Determines that the probability of occurrence is high and sets the error interval and applies the error detection and correction method according to Concealment level 3.

In addition, while the error detection and correction method according to Concealment level 3 is applied, a state in which the reception rate is higher than the predetermined value (or the error rate is lower than the predetermined value) lasts for a predetermined time or the error does not occur continuously. If more than the predetermined number, it is determined that there is a low probability that the error occurs for the subsequent frame, the set error interval is released and error detection and correction method according to Concealment level 3 is not applied.

The value (the predetermined value, the predetermined number) that is the basis of the determination when setting or canceling the error interval is a value used when setting as shown in FIG. 6 (value 1, L1) and a value used when canceling. 2, L2) may be different.

According to the concealment level 3, high frequency noise can be almost eliminated. However, low frequency noise still remains.

4) Concealment level 4

Concealment level 4 is to mute the entire waveform instead of compensating for low frequency noise, which cannot be perfectly compensated for. The waveform can be muted for the error section set by the Concealment level 3 method. The reception rate value, the number of frames where an error occurs, can be set more strictly than the Concealment level 3.

In addition, it is possible to set whether to apply Concealment level 4 according to the user's preference. Even if there is a lot of noise, the user who wants to listen to the broadcast may not set Concealment level 4, and in case of a lot of noise, it is better to be quiet. In case of judging user, Concealment level 4 can be set. Therefore, the user can determine whether to apply Concealment level 4.

In addition, it is possible to set whether to apply Concealment level 4 for each content or radio broadcast type. In the case of a music broadcast, the Concealment level 4 is set, and in the case of a content including a human voice, the Concealment level 4 is not set. Can be.

Concealment level 4, although there is still a problem that the audio playback section is reduced as much as the muted section, it can remove all the noise of the low frequency and high frequency form.

As described above, the method according to the present invention for reducing or removing noise generated in an audio broadcast may be applied to a DMB terminal for receiving a DMB, a mobile phone with a DMB function, a portable multimedia player (PMP), and the like.

Preferred embodiments of the present invention described above have been disclosed for the purpose of illustration, and those skilled in the art will improve, change, and substitute various other embodiments within the technical spirit and the technical scope of the present invention disclosed in the appended claims below. Or addition may be possible.

According to the embodiment of the present invention as described above, it is possible to reduce or effectively remove the noise generated in the audio broadcasting of terrestrial digital multimedia broadcasting.

Claims (24)

Confirming that a magnification component for the audio signal sample, which is error prone, falls within a predetermined range; And Correcting the magnification component using the value as it is if the magnification component falls within the predetermined range, or using a first value determined based on at least one of the predetermined range or the magnification component if not. It is made, including Here, the predetermined range is determined by reference to the value of the magnification component at the immediately preceding time or by interpolation based on a magnification component that is located before and after the potential magnification component in error and is free of errors. The signal processing method, characterized in that determined based on the calculated value. The method of claim 1, And a reference value for determining the predetermined range is used for the first value. The method of claim 1, And a value close to the magnification component among the boundary values in the predetermined range is used as the first value. The method of claim 1, And a reference value for determining the predetermined range and an average value of the magnification component are used for the first value. The method of claim 4, wherein And when the average value is out of the predetermined range, a value close to the magnification component among the boundary values of the predetermined range is used as the first value. delete delete The method of claim 1, The signal processing method characterized in that the predetermined range is determined with the magnification component used as it is as the center value when the magnification component that is temporally ahead of the error potential magnification component is used as it is. . delete The method of claim 1, The predetermined range is determined by a symmetrical structure having the same width in both directions with respect to the reference value. The method of claim 1, And the predetermined range is determined by an asymmetric structure having different widths in one direction and the other direction with respect to the reference value. The method of claim 11, And a non-symmetrical structure having a fixed ratio of widths in the one direction and the other direction within the predetermined range. The method of claim 11, And a ratio of a width in one direction to another in accordance with a relative position at which the reference value lies within a range of values that the magnification component can have. The method of claim 1, And the predetermined range is determined by one of a symmetrical structure and an asymmetrical structure according to a position where the reference value lies in a range of values that the magnification component can have. The method of claim 1, And the width of the predetermined range is fixed. The method of claim 1, And the width of the predetermined range is continuously updated based on a change in the magnification component value previously used without error. The method of claim 16, Wherein the width of the predetermined range is updated based on a combination of any one or more of the maximum, minimum, average, and deviation of the difference between two magnification component values adjacent to each other, previously used without error. The method of claim 1, A value which is continuously updated for a predetermined time is used for the width of the predetermined range, and a value which is last updated after the predetermined time is used. Determining whether an error probability is high for each section of a received audio broadcast; And And correcting the audio sample based on an energy or magnitude characteristic of the audio sample to be decoded in the section determined as the section having a high probability of error. The method of claim 19, Whether the probability of an error is high is determined based on at least one of a change in a reception rate for a section already received or a continuity of a section in which an error occurs during a decoding operation on already received audio data. The method of claim 19, Whether or not to correct the audio sample is determined by determining whether at least one of the energy or the magnitude of the audio sample is outside a predetermined allowable range set to an absolute value for each. The method of claim 21, And determining whether to correct the audio sample in units of bands separated by a predetermined number. The method of claim 22, A signal processing method characterized in that an audio sample is corrected only for a high frequency band. Determining an allowable range for the scale factor of the current time based on the scale factor used at the previous time; Confirming that a scale factor of the current time for any subband in the scale factor group where an error is detected falls within the determined tolerance range; And Using the boundary value of the allowable range close to the scale factor of the current time as the corrected scale factor of the current time if the scale factor of the current time does not fall within the allowable range. How to remove it.

Images