JP2002111503A - Decoding device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To define a frame of certain length, where a minimum number of frequency extraction components are obtained are obtained as an optimal frame, when a signal frequency component is extracted through a general harmonic analysis, and to decode a bit stream containing a frequency extraction component in an optimal frame. SOLUTION: A frame-setting circuit 1 sets the frame length of a waveform sample, which is generated by a waveform sample generating circuit 2 at the waveform sample generating circuit 2, on the basis header information. The waveform sample generating circuit 2 reads in the header information of the inputted bit stream and decodes signal frequency component in the frame of the inputted bit stream, through general harmonic analysis on the basis of header information. Successively, it is determined us to whether the decoded frequency component is the last frequency component, a decoding operation is continued until the decoding is the last frequency component, and a waveform sample which is the decoded result is outputted.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a decoding device,
In particular, the present invention relates to a decoding apparatus that decodes an audio signal from a coded bit stream by extracting frequency components by general harmonic analysis in frame units.

[0002]

2. Description of the Related Art A conventional encoding method corresponding to a decoding method for an audio signal is based on a time-frequency conversion method. This is a method in which an input signal in the time domain is converted into a frequency domain and then coding is performed, and is a coding method using the bias of the signal in the frequency domain and the psychological perception of human beings.

However, a frequency analysis method such as a fast Fourier transform (FFT) or a modified discrete cosine transform (MDCT) used in the encoding method based on the time-frequency transform method is a theory for analyzing a periodic and harmonic signal. It is assumed that the signal in the observation section periodically repeats outside the observation section where the signal is observed.

However, in the above-described frequency analysis method, since a large number of frequency components different from the actual ones are extracted depending on the observation section, errors are already included in the frequency domain signal, and the coding efficiency is reduced. Since the frequency resolution is inversely proportional to the length of the observation section, even when performing a transient signal analysis, the observation section length cannot be shortened too much, the frequency resolution becomes insufficient and the coding efficiency decreases, and However, since it is difficult to predict the waveform outside the observation section, there are drawbacks such as difficulty in improving the coding efficiency by coding the prediction residual.

On the other hand, the theory of Fourier analysis extended to non-harmonic signals is also known. This theory is called Generalized Harmonic Analysis (GHA),
This is an analysis method in which the most dominant sine wave with the minimum residual energy is extracted from the original waveform in the observation section, and the same processing is repeated for the residual component. For this reason, accurate frequency analysis can be extracted even for small frequency fluctuations that are not stationary, and the observation section length and frequency resolution can be freely set independently of each other. The feature is that prediction of is possible.

[0006]

Therefore, the above-mentioned GHA
Can perform frequency analysis with higher accuracy than harmonic analysis such as FFT and MDCT. Therefore, compared with the case of using FFT or the like used as a frequency analysis unit in a conventional encoding device, the There is a problem that the accuracy is high, and the amount of code required to encode the accuracy of the extracted frequency component is increased depending on the input signal as compared with the case where FFT or the like is used. In addition, it is necessary to ensure that decoding can be performed even if an audio signal is encoded so that encoding efficiency can be greatly improved using GHA.

[0007] The present invention has been made in view of the above points,
An object of the present invention is to provide a decoding device capable of reliably decoding a bit stream obtained by encoding an audio signal using GHA.

[0008]

In order to achieve the above object, the present invention sequentially selects a plurality of types of frames, that is, a longest frame and a frame having a length that is a fraction of an integer, and selects each of the selected frames. The frequency component of the input digital audio signal is extracted by general harmonic analysis, and among the number of frequency extraction components in each frame, the frame length that gives the minimum number of frequency extraction components in the same sample section is determined as the optimal frame. A bit stream output from an encoding device that generates a bit stream by adding at least header information indicating a frame length to a frequency extraction component in an optimal frame is received as an input signal via an arbitrary transmission path, and an audio signal is output. A decoding device that decodes the longest data based on the header information of the bit stream. A frame length setting circuit that sets a frame length by reference updating at a frame length cycle, and a waveform obtained by decoding a frequency component of a signal of the frame of a bit stream by general harmonic analysis while referring to information of the frame length setting circuit. And a waveform sample generation circuit that generates a sample and outputs the sampled signal as a decoded signal.

According to the present invention, the frequency components of the input digital audio signal are extracted by general harmonic analysis, and of the number of frequency extraction components of each frame, the frame length of the frame length at which the minimum number of frequency extraction components is obtained in the same sample section is obtained. The bit stream output from the encoding device that determines a signal as an optimum frame and adds a header information to a frequency extraction component in the optimum frame to generate a bit stream is a waveform sample from a frequency component of a signal having a longest frame length. Generate

[0010]

Next, an embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing an embodiment of a decoding apparatus according to the present invention. The decoding device of this embodiment decodes a bit stream obtained by encoding by the encoding device proposed by the present inventor in Japanese Patent Application No. 11-75171. Before describing the mode, the encoding apparatus will be described first.

FIG. 5 shows that the inventor of the present invention disclosed in Japanese Patent Application No. 11-75171.
FIG. 1 shows a block diagram of an example of an encoding device proposed in No. 1;
In the figure, the encoding device comprises a general harmonic analysis circuit 10
A determination circuit 13 for determining whether all frames have been selected, a determination circuit 14 for determining an optimal frame and outputting a frequency extraction component, a bit stream generation circuit 15 for generating a bit stream, and a longest frame And a frame selection circuit 16 that prepares a plurality of types of frames, that is, a frame having a length that is a fraction of the integer, and sequentially selects the frames. General harmonic analysis circuit 10
Comprises a framing circuit 11 for dividing an input digital audio signal into frames composed of a plurality of samples, and a frequency component extraction circuit 12 for extracting frequency components by general harmonic analysis.

Next, the operation of the encoding apparatus will be described with reference to FIG.
This will be described with reference to the flowchart of FIG. For example, a digital audio signal having a sampling frequency of 48 kHz is supplied to the framing circuit 11 in FIG. 1, where it is divided into observation sections (frames) including a plurality of samples. At this time, the frames do not necessarily have to overlap. At this time, the longest frame is set in the frame selection circuit 16 for sequentially selecting the frames. The digital audio signal (multiple samples) in the frame is
The frequency component is supplied to the frequency component extraction circuit 12, where the frequency component of the signal in the frame is extracted by the general harmonic analysis (GHA) according to the procedure of the flowchart shown in FIG.

That is, first, when a digital audio signal of one frame is input (step 10 in FIG. 6).
1) Extract only one frequency component that minimizes the residual signal from the signal of the one frame (step 102 in FIG. 6). Subsequently, it is determined whether or not the original signal excluding the extracted frequency component, that is, the residual is sufficiently small (step 103 in FIG. 6). If the residual is larger than a predetermined threshold, the processing in step 102 is performed. Is performed again on the residual, and the analysis is terminated when the residual becomes sufficiently small. Then, the obtained analysis result is determined by the determination circuit 1 shown in FIG.
3 (step 104 in FIG. 6).

With the extracted sine wave (frequency component), the signal x (t) in the frame is expressed as f ₁ ,
f ₂ ,. . . , Represented by superposition of the f _N, a total of N number of sine wave (frequency extracting component). These sine waves are not necessarily harmonic (frequency that is an integral multiple of the fundamental wave). In addition,
In the expression (1), x (t), Sk and Ck are amplitudes, t is time, f is an arbitrary frequency, and n is an integer.

[0015]

(Equation 1) Here, the condition for terminating the analysis in the frequency component extraction is that the residual is sufficiently small as shown in step 103 in FIG. 6, and is not the number of frequency extraction components. “The residual is sufficiently small” means that the square error value of the amplitude per sample is a sufficiently small value and smaller than a fixed value dt (> 0) regardless of the frame length.
That is, there is no correlation between the frame length and the number of frequency extraction components. However, to improve coding efficiency,
The smaller the number of frequency extraction components, the better.

Therefore, a plurality of types of frame lengths, which are division units in the framing circuit 11, are prepared in advance, and the determination circuit 13 sends a signal for sequentially specifying a plurality of types of frame lengths to the framing circuit 11 through the frame selection circuit 16. Output. Then, the determination circuit 13 calculates the number of frequency extraction components at each frame length in the section of the same number of samples sequentially extracted from the frequency component extraction circuit 12, and determines whether all frame lengths have been selected. .

The decision circuit 14 receives the frequency-extracted components of the frame length and the number of frequency-extracted components at each frame length extracted from the decision circuit 13, and determines the number of frequency-extracted components corresponding to each frame length. By comparison, the frame length that provides the minimum number of frequency extraction components is determined as the optimum frame length, and the frequency extraction component having the optimum frame length is output to the bit stream generation circuit 15.

For example, as a result of analyzing a section of 1024 samples, when the number of frequency extraction components when 1024 samples are one frame length (first frame length) is 10
0, which is 1 when 1024 samples are divided into two to make a frame length of 512 samples (second frame length).
The number of frequency extraction components for 024 samples is 85 (= 4
0 + 45) lines, and 1024 samples were divided into four
When the frame length of 56 samples (the third frame length) is 1, the number of frequency extraction components for 1024 samples is 1
If there are 10 (= 35 + 25 + 30 + 20),
The second frame length is determined as the optimum frame length.

In practice, the longest frame length is 1024
About 4096 samples, 1/2, 1
Provisional analysis is performed for each of a plurality of frame lengths of about ４, ８, and 1/16 to determine an optimum frame length.
In this way, the processing of the adaptive frame length for determining the optimum frame length according to the characteristics of the digital audio signal is performed, and as few frequency extraction components as possible per unit time are extracted from the determination circuit 14. The processing of the adaptive frame length is performed for each section length that is an integral multiple of the longest frame length.

The above-described frequency extraction component is supplied to the bit stream generation circuit 15 in FIG. 1, and the extraction component having a power level lower than the set power level is not output, thereby reducing the amount of information and further setting. The amount of information is reduced by lossless compression coding such as Huffman coding for frequency extraction components at power levels higher than the power level, and header information that records the number of frequency extraction components and frame length information is added. And transmitted to an arbitrary transmission path as a bit stream. (1)
The power level P (f _k ) of the signal x (t) represented by the equation
Is represented by S ² _k + C ² _k .

Thus, for example, taking three frame lengths from the first frame length to the third frame length as an example, the longest first frame length (1024 samples)
If the above adaptive frame length processing is performed for each section of 1024 samples as a preset section length of an integral multiple of the above, frame length information indicating which of the three frame lengths has been coded is included in the header. In the decoding device (decoder) side, when the second frame length information is obtained from the frame length information, the frame length continues twice, and the third frame length information , It is understood that the frame length continues four times.

Next, the configuration and operation of one embodiment of the decoding apparatus according to the present invention shown in FIG. 1 will be described. As shown in FIG. 1, a decoding device according to the present embodiment includes a frame setting circuit 1 to which a bit stream output from a bit stream generation circuit 15 in FIG. 5 is input via an arbitrary transmission path, And a waveform sample generating circuit 2 to which one output signal is input.

The frame setting circuit 1 converts a waveform sample generated by the waveform sample generation circuit 2 based on the header information out of the bit stream input from the encoding apparatus shown in FIG. The frame length is set in the waveform sample generation circuit 2. Here, the frame length is the longest frame or a frame of a length that is an integral number of the longest frame. If the frame length is a frame length other than the longest frame, at least two or more frames of the same length are continuously set. In addition, since the number of continuations is uniquely determined by the frame length, it is not always necessary to refer to the header information for each frame.

The waveform sample generating circuit 2 generates waveform samples of the bit stream input from the encoding apparatus shown in FIG. 5 via an arbitrary transmission line according to the flowcharts shown in FIGS. Output signals in the same order as in the above. That is, first, the header information of the input bit stream is read (step 201 in FIG. 2), and the frequency component of the signal in the frame of the input bit stream is decoded by general harmonic analysis based on the header information (step 202 in FIG. 2). Subsequently, it is determined whether the decoded frequency component is the last frequency component (Step 203 in FIG. 2),
It decodes up to the last frequency component and outputs a waveform sample as a decoding result (step 204 in FIG. 2). In addition,
The header information makes it possible to know how many frequency components exist in the frame.

FIG. 3 shows a detailed flowchart of step 202 in FIG. The frequency component decoding process in step 202 will be described in more detail with reference to the flowchart of FIG. 3. One frequency component is selected from the input bit stream (step 301 in FIG. 3), and the sine wave of the frequency component is selected. A (sin wave) waveform is generated (step 302 in FIG. 3), and a cosine wave (cos wave) waveform is generated (step 303 in FIG. 3).

Then, those waveforms in the frame are added to decode the frequency component (step 304 in FIG. 3).
In this manner, the waveform sample generation circuit 2 decodes the frequency component of the signal in the frame, generates a waveform sample of the signal, and outputs it as a decoded signal. FIG. 4 shows an example of the waveform of the decoded signal output from the waveform sample generation circuit 2. Sections A and B each indicate a frame.

[0027]

As described above, according to the present invention,
When extracting frequency components of a signal by general harmonic analysis,
By determining the frame length at which the minimum number of frequency extraction components can be obtained in the same sample section as the optimal frame, and encoding the frequency extraction components in the optimal frame, efficient and independent signal steadiness is achieved. The frequency component extraction can be performed, and the frequency component constituting the bit stream is reduced, so that the bit stream generated by the coding apparatus that has the effect of greatly improving the coding efficiency is used as the input bit stream, By generating waveform samples from the frequency components of the signal at the longest frame length of the stream, it is possible to reliably decode the audio signal from the bit stream.

[Brief description of the drawings]

FIG. 1 is a block diagram of an embodiment of a decoding device according to the present invention.

FIG. 2 is a flowchart for explaining the operation of the waveform sample generation circuit in FIG. 1;

FIG. 3 is a flowchart for explaining details of a main part in FIG. 2;

FIG. 4 is a diagram illustrating an example of a waveform of a decoded signal.

FIG. 5 is a block diagram of an example of an encoding device that outputs a bit stream to be decoded by the device of the present invention.

FIG. 6 is a flowchart for explaining an operation of a main part in FIG. 5;

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Frame setting circuit 2 Waveform sample generation circuit 10 General harmonic analysis circuit 11 Frame conversion circuit 12 Frequency component extraction circuit 13 Judgment circuit 14 Decision circuit 15 Bit stream generation circuit 16 Frame selection circuit

Claims (1)

[Claims] 1. A method for sequentially selecting a plurality of types of frames including a longest frame and a frame having a length equal to an integral number thereof, and extracting a frequency component of an input digital audio signal from each selected frame by general harmonic analysis. Out of the number of frequency extraction components of each frame, a frame length having the minimum number of frequency extraction components in the same sample interval is determined as an optimal frame, and the frequency extraction component in the optimal frame includes at least a frame length. A bit stream output from a coding apparatus that generates a bit stream by adding header information indicating the bit stream as an input signal via an arbitrary transmission path, and decodes an audio signal. Based on the header information of A frame length setting circuit for setting a system length, and generating a waveform sample by decoding frequency components of a signal of the frame of the bit stream by general harmonic analysis while referring to information of the frame length setting circuit. A waveform sample generating circuit for outputting the signal as a signal.