KR100713452B1 - Apparatus and method for coding of audio signal - Google Patents

Abstract

The present invention allocates audio signals to at least two subbands having a specific frequency domain, and encodes and transmits the audio signals when the gain of the audio signals allocated to the respective subbands exceeds a masking threshold. Measure the gain of audio signals, readjust the audio signal in proportion to the importance, reset the masking threshold of each subband in inverse proportion to the measured gain, and exceed the reset masking threshold Encode only the signal.

MPEG-4, AAC, shaping function, psychoacoustic model, coding

Description

Apparatus and method for encoding an audio signal {APPARATUS AND METHOD FOR CODING OF AUDIO SIGNAL}             

1 is a diagram showing the configuration of an encoder for encoding a conventional audio signal.

2 is a diagram illustrating a process of encoding a speech signal transmitted according to a masking threshold in a subband.

3 is a diagram illustrating a process of encoding a speech signal transmitted according to a masking threshold in one subband.

4 is a diagram illustrating a configuration of an encoder for encoding an audio signal according to the present invention.

5 is a diagram illustrating a forming function generated in accordance with the present invention.

6 is a diagram illustrating a process of reconstructing a voice signal using the generated function according to the present invention.

7 is a diagram illustrating a process of encoding a speech signal transmitted according to a reset masking threshold according to the present invention.

The present invention relates to audio encoding, and more particularly, to an apparatus and a method for increasing the compression efficiency of audio encoding.

The audio signal is a video data, audio data, and other data that have been generated and processed heterogeneously and independently in accordance with the recent expansion of digital communication technology, high integration semiconductor (VLSI), and signal processing (DSP) technology. It can be processed and used in a wide variety of forms without distinguishing between sources and media. In this environment, there is a need for standardization of international transmission standards of digital data for smooth information transmission and sharing between different devices, and according to such needs, Joint Picture Experts Group (JPEG) and moving picture MPEG has been standardized for the transmission of.

According to the MPEG standardization, a method of compressing and multiplexing digital video data and audio data and transmitting the same, and a method of decoding the compressed and transmitted video data and audio data, and the like are prescribed. According to this provision, digital data is further compressed by Discrete Cosine Transform, Quantization, Scan, and finally by Variable Length Coding. The compressed and transmitted data is restored after undergoing a variable length decoding, inverse quantization, and inverse discrete cosine transform.

1 shows a configuration of an encoder for encoding an audio signal. Hereinafter, a configuration of an encoder for encoding an audio signal will be described with reference to FIG. 1. The encoder includes a time / frequency mapping unit 102, a psychoacoustic unit 100, a time domain noise shaping unit 104, an intensity stereo processor 106, a predictor 108, A mid / side (M / S) stereo processor 110, a quantization unit 112, a bit division arithmetic coding unit 114, and a bitstream forming unit 116 are provided.

Among the auditory characteristics of human, the most important phenomenon in encoding of digital audio signal is the property of masking effect and critical hand. The masking effect is a phenomenon in which a signal (sound) is not heard by another signal (sound). For example, if you talk at a low level when a train passes by at a train station, the conversation is not buried in the sound of the train. The critical band is a phenomenon in which a signal is recognized by a certain band within the human audio frequency range, and has various characteristics. One of these characteristics is that humans perceive different amounts of noise when they are within a certain threshold for the same amount of noise and when they exceed the threshold range. In this case, the perception of noise when leaving the critical band is higher.

The encoding using auditory characteristics basically calculates how much noise can be allocated within one critical band using the two characteristics, and then generates the quantization noise during quantization, thereby losing information due to encoding. To minimize

The time / frequency mapping unit 102 converts an input audio signal in a time domain into a signal in a frequency domain. The psychoacoustic unit 100 bundles the audio signals converted from the time domain into the frequency domain by the time / frequency mapping unit 102 into signals of appropriate scale factor bands, and interacts with each other. Masking thresholds in each scale factor band are calculated using the masking phenomenon generated.

The temporal noise shaping (TNS) unit 104 is used to control the temporal shape of the quantization noise in each window of the transform. By applying the filtering process of the frequency data, time domain noise shaping is possible. The module is a module that can be optionally used in the encoder. The intensity stereo processor 106 is a module for more efficiently processing a stereo signal, and encodes only quantized information about a scalefactor band for one of two channels, and transmits only the scalefactor for the remaining channels. The module is not a module that must be used in the encoder, and it is possible to determine whether to use each scalefactor band unit in consideration of various matters.

The predictor 108 predicts values of frequency coefficients of the current frame. The amount of bits used can be reduced by quantizing and encoding the difference between the predicted value and the actual frequency component. However, the prediction unit 108 may be selectively used in units of frames. In other words, the prediction unit 108 may not be used because it increases the complexity of predicting the next frequency coefficient. In addition, in some cases, since the prediction difference has a greater probability than the original signal, the prediction may cause the generated bit generation to be larger than when the prediction is not performed. In this case, the prediction unit 108 is not used. .

The M / S stereo processor 110 processes the stereo signal more efficiently. The M / S stereo processor 110 processes the signal after converting the left channel signal and the right channel signal into the added and subtracted signals, respectively. The M / S stereo processor 110 is not necessarily used in the encoder, but may determine whether the encoder is used in each scalefactor band unit in consideration of various matters.

The quantization unit 112 scalar quantizes the frequency signals of each band so that the magnitude of quantization noise of each band is smaller than a masking threshold value so that a human cannot feel it. The quantization is performed such that the NMR value of the entire band is 0 dB or less by using a noise-to-mask ratio (NMR), which is a ratio of a masking threshold calculated by the acoustic psychology unit 100 and noise generated in each band. An NMR value of 0 dB or less indicates a higher masking value than quantization noise, meaning that humans cannot hear the quantization noise.

The bit division arithmetic encoder 114 is a core module of the present invention, and since an existing audio codec such as MPEG-2 AAC cannot provide scalability, the bit split arithmetic encoder 114 may be used as an alternative to the lossless encoder of AAC. In order to implement scalable audio codec, the quantization unit 112 encodes the quantized frequency data value by combining the additional information of the corresponding band and the quantization information of the audio data according to a bit rate corresponding to each layer. In addition to these scalability features, the top layer can provide performance similar to AAC. When the function of the bit division arithmetic encoder 114 is described in more detail, a band restriction corresponding to the base layer is performed to enable bit rate adjustment, and side information is encoded. Information about the quantized values is encoded in order from the most significant bit to the least significant bit, and then in order from the lowest frequency component to the highest frequency component. In addition, the left and right channels are alternately encoded in a predetermined vector unit to encode the base layer. After the encoding of the base layer is finished, the bitstream has a hierarchical structure by encoding quantization values of additional information and audio data of the next layer. The bitstream forming unit 116 collects information generated in each module of the encoder and constructs the bitstream according to a predetermined syntax made to fit the scalable codec.

2 illustrates a process of processing an input signal using a masking threshold. According to FIG. 2, the input signal is divided into four bands. The four bands are composed of subband 1 to subband 4. The encoder obtains the gain of subband 1 to subband 4. 2 illustrates gains measured in subbands 2 and 3, for example. The gain of each subband is compared with a masking threshold transmitted from the psychoacoustic unit. As a result of the comparison, it can be seen that the gain in subband 2 is higher than the masking threshold and the gain in subband 3 is smaller than the masking threshold. The process of calculating the masking threshold is irrelevant to the present invention and will be omitted. As described above, if the gain measured in each subband is smaller than the masking threshold, it is excluded from the encoding process, thereby increasing the compression efficiency.

3 is a diagram showing a problem according to the prior art. Hereinafter, the problems of the prior art will be described with reference to FIG. 3. Referring to FIG. 3, subband 3 shows gains for two signal components. Since two signal components of the subband 3 are higher than the masking threshold, an encoding process is performed. However, in order to perform the encoding process on the two signal components, a bit rate for transmitting the encoded signal components should be guaranteed. If the bit rate for transmitting the two signal components is not guaranteed, the encoder does not perform an encoding process on the signal components of the subband 3. The bandwidth of the encoded audio signal is continuously changed because the encoding process for the entire signal component of the subband 3 is not performed. The change in bandwidth results in degradation of sound quality. Therefore, a method for solving the above problem is discussed.

Accordingly, an object of the present invention for solving the above problems of the prior art is to propose an apparatus and method for performing efficient data transmission even under a low bit rate.

Another object of the present invention is to propose an apparatus and method for preventing degradation of sound quality by minimizing a change in bandwidth of transmitted data.

Another object of the present invention is to propose an apparatus and method for performing data rate transmission by differently setting a masking threshold for encoding according to the gain of each band.

In order to achieve the above object of the present invention, a method for allocating audio signals to at least two subbands having a specific frequency region and encoding if the gain of the audio signal allocated to each subband exceeds a masking threshold, Measuring a gain of the audio signals allocated to each subband, readjusting the audio signal in proportion to the importance, resetting the masking threshold of each subband in inverse proportion to the measured gain, And encoding only an audio signal exceeding the reset masking threshold.

An apparatus for allocating audio signals to at least two subbands having a specific frequency range to achieve the objects of the present invention, and encoding and transmitting the audio signals when the gain of the audio signals allocated to each subband exceeds a masking threshold. A psychoacoustic unit for measuring the gains of the audio signals allocated to the subbands and re-adjusting the audio signal in proportion to the importance, and resetting the masking thresholds of the subbands in inverse proportion to the measured gains And a forming function for outputting only an audio signal exceeding the reset masking threshold.

Hereinafter, the present invention will be described in detail with reference to the accompanying drawings. In describing the present invention, when it is determined that a detailed description of a related known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted.

4 shows a configuration of an encoder for encoding an audio signal according to the present invention. Hereinafter, a configuration of an encoder for encoding an audio signal will be described with reference to FIG. 4. In FIG. 4, unlike in FIG. 1, a shaping function generator 118 and a forming unit 120 are added. Other configurations except for the formation function generator 118 and the formation unit 120 are as described with reference to FIG. 1. Hereinafter, the forming function generator 118 and the forming unit 120 will be described. The shaping function generator 118 generates a shaping function in consideration of the gain of each received subband. The shaping function may set a fixed value for each subband or may be expressed in the form of a function. 5 shows an example of a formation function modified in the frequency domain. Referring to FIG. 5, the subband 1 has a higher formation function than the subband 2, and the subband 2 has a higher formation function than the subband 3. The subband 3 has a higher formation function than the subband 4. The resulting formation function is related to the importance of each subband. The higher the importance of the signal component included in the subband, the larger the magnitude of the formation function. Referring to FIG. 5, it can be seen that the signal component included in the subband 1 has the greatest importance.

Equation 1 below represents a formula for forming a signal transmitted from the forming unit using the generated forming function.

X_freq '[i] = w (i) * X_freq [i]

X_freq '[i] means a signal formed by the formation function, and w (i) means a formation function. The X_freq [i] means a signal output from the time / frequency mapping section. Fig. 6 shows an example in which a signal output from the time / frequency trapping portion is formed by the forming function. As the formation function, the formation function of FIG. 5 was used. It can be seen that the formed signal has a smaller value than the original signal. That is, in FIG. 5, the shape function in subband 1 has a value 1, and the shape function in subbands 2 through 4 has a value smaller than 1. FIG. However, it is obvious that the shape function may have a value greater than 1 in some cases.

Hereinafter, a process of generating a mask threshold in the psychoacoustic unit 100 will be described. The psychoacoustic unit 100 bundles the audio signals converted from the time domain into the frequency domain by the time / frequency mapping unit 102 into signals of an appropriate scale factor band, and due to the interaction of the respective signals. Masking thresholds in each scale factor band are calculated using the generated masking phenomenon. The masking threshold is reset using the calculated masking threshold and the gain in each received subband. Equation 2 below represents an equation for resetting a masking threshold in the psychoacoustic unit.

mask '[i] = mask [i] * ((max_gain) / (gain (i))

The mask '[i] represents a reset masking threshold and the mask [i] represents a masking threshold before resetting. "I" denotes the i-th subband, and (max_gain) represents the highest gain value among the gains of the received subband. The gain (i) represents a gain value in the i th subband.

By resetting the masking threshold as described above, the reset masking threshold has a higher value than the previous value. An encoding process is performed on the received signal using the reset masking threshold.

7 illustrates a process of encoding a received signal using a reset masking threshold according to the present invention. FIG. 7 assumes a case in which a signal to be encoded is limited to subband 3 as shown in FIG. 3. FIG. 7 illustrates an encoding process by using a non-reset masking threshold and a reset masking threshold. Hereinafter, the masking threshold to be reset will be described as an example. It is assumed that the received signal is composed of subbands 1 to 4.

benefit Subband 1 100 Subband 2 90 Subband 3 85 Subband 4 80

When the gain according to each subband is received as shown in Table 1, the psychoacoustic unit resets the masking threshold as shown in Table 2 below in Equation 2.

Masking threshold before reset Masking Threshold After Reset Subband 1 mask [1] mask [1} * (100/100) Subband 2 mask [2] mask [2] * (100/90) Subband 3 mask [3] mask [3] * (100/85) Subband 4 mask [4] mask [4] * (100/80)

As shown in Table 2, it can be seen that the masking threshold value after reset is set to be larger than the masking threshold value before reset. Only the masking threshold for subband 1 having the maximum gain is reset the same.

As shown in FIG. 7, it can be seen that the two signal components in subband 3 have a value higher than the masking threshold before resetting. Therefore, the code part must perform an encoding process on two signal components located in subband 3. However, as described in the related art, when a bit rate that is sufficient to transmit both signal components is not guaranteed, the encoder does not perform an encoding process on both signal components located in the subband 3.

However, according to the reset masking threshold, only one signal component among two signal components located in subband 3 has a value greater than the masking threshold. Accordingly, the encoder performs an encoding process only on signal components having a value larger than the reset masking threshold. The encoder does not perform an encoding process on a signal component having a value smaller than the reset masking threshold. By performing the encoding process only on signal components having a value greater than the reset masking threshold, the voice signal can be transmitted even under a limited bit rate.

As described above, the present invention increases the compression efficiency by applying the subband gain to audio encoding and keeps the bandwidth of the encoded signal constant. In addition, the degradation of sound quality is minimized by keeping the encoded bandwidth constant. The present invention can also efficiently transmit an audio signal by using it in an audio encoder requiring a low bit rate.

Claims (10)

A method for allocating audio signals to at least two subbands having a specific frequency domain and encoding and transmitting audio signals when a gain of the audio signals allocated to each subband exceeds a masking threshold, The following equations are used to measure the gains of the audio signals assigned to each subband, and a shaping function indicating the importance of each subband relative to the importance of the subband having the maximum gain according to the measured gains. Re-adjusting the audio signal by And resetting a masking threshold of each subband in inverse proportion to the measured gains, and encoding only a readjusted audio signal having a gain exceeding the reset masking threshold. X_freq '[i] = w (i) * X_freq [i] X_freq '[i]: Signal readjusted by the shaping function w (i): formation function X_freq [i]: signal before readjustment i: subband indicator delete delete The method as claimed in claim 1, wherein the resetting of the masking thresholds of the subbands is determined by the following equation.  mask '[i] = mask [i] * ((max_gain) / (gain (i)) mask '[i]: reset masking threshold mask [i]: masking threshold before reset max_gain: maximum gain gain (i): gain of the i-th subband The method as claimed in claim 1, wherein only one of the audio signals included in each subband exceeds the reset masking threshold. 10. An apparatus for allocating audio signals to at least two subbands having a specific frequency domain, and encoding and transmitting audio signals when a gain of the audio signal allocated to each subband exceeds a masking threshold. The following equations are used to measure the gains of the audio signals allocated to the respective subbands, and use the Shape function to indicate the importance of each subband relative to the importance of the subband having the maximum gain according to the measured gains. A psychoacoustic unit which readjusts the audio signal by A shaping function generator that resets masking thresholds of the subbands in inverse proportion to the measured gains; And a forming unit for outputting only the readjusted audio signal having a gain exceeding the reset masking threshold. X_freq '[i] = w (i) * X_freq [i] X_freq '[i]: Signal readjusted by the shaping function w (i): formation function X_freq [i]: signal before readjustment i: subband indicator delete delete The method of claim 6, wherein the forming unit, And resetting the masking threshold by the following equation and outputting only the audio signal exceeding the reset masking threshold. mask '[i] = mask [i] * ((max_gain) / (gain (i)) mask '[i]: reset masking threshold mask [i]: masking threshold before reset max_gain: maximum gain gain (i): gain of the i-th subband The method of claim 6, wherein the formation function generating unit, And wherein only one of the audio signals included in each subband generates a shaping function to exceed the reset masking threshold.

Images