Encoders and decoders

Abstract

To provide an encoder and decoder that can encode and decode an input signal as a high-quality audio signal with sufficient noise reduction, even when the amplitude fluctuation range of the input signal is large. [Solution] The system is configured to include a subtractor 1 that subtracts a feedback signal from an input signal X(n), a quantizer 2 that quantizes the output signal of the subtractor 1, a feedback delay 4 which is the output signal of the quantizer 2 and delays the output signal Y(n) of the entire coding system, a feedback inverse quantizer 4 which inversely quantizes the output signal of the feedback delay 4 to obtain the feedback signal Y1(n), a plurality of adaptive delays 5·5… that delay the output signal of the quantizer 2, and a scale value determination means 6 that determines the scale value of the quantization performed by the quantizer 2 based on the output signals of the plurality of adaptive delays 5·5….

Description

Technical Field

[0001] The present invention relates to an encoder for encoding an input signal and a decoder for decoding an encoded signal in a signal transmission means using digital modulation. Background Art

[0002] DSM (Delta-Sigma Modulation) is a type of pulse density modulation, which is a technology that enables transmission of high-quality audio signals at low bits by performing noise shaping to distribute quantization noise over a high frequency band. Applications thereof include 1-bit A / D converters and amplifiers. In a basic configuration, a signal obtained by delaying an output signal with a delay device is fed back and subtracted from an input signal. The output signal is a signal obtained by performing integration processing by an integrator and quantization processing by a quantizer on the signal after subtraction.

[0003] In DSM, a signal can be encoded and transmitted as a 1-bit digital signal, but can also be made into a multi-bit digital signal. Generally, transmission at low bits is performed by increasing the sampling frequency as much as possible (also referred to as oversampling). An encoded signal can be restored to the original signal by performing decoding with an inverse quantizer.

[0004] On the other hand, in the field of digital modulation technology, it is important to reduce deterioration in sound quality while reducing the data volume for efficient communication. As such a data compression technology, for example, the ADPCM (Adaptive Differential Pulse Code Modulation) system has been developed. ADPCM achieves both reduction of data volume and prevention of sound quality deterioration by quantizing the difference from data one sampling before, utilizing the continuity of audio signals.

[0005] Patent Document 1 discloses a technique developed for the purpose of reducing circuit scale in ADPCM. In this configuration, the encoded output signal is input to a quantization step adjuster. The quantization step adjuster adjusts the quantization step width based on the output signal. In other words, it applies the output signal to the quantization width of the signal after one sample. To put it another way, the quantization width is determined based on the output signal one sample prior.

[0006] In the technology described in Patent Document 1, in addition to this quantization step adjuster, the output signal is configured to obtain the difference between the output signal and the input signal via a predictor. This predictor is configured to either output the input signal from one sample prior as the next predicted value, or to output a value obtained by linearly extrapolating the signals from one and two samples prior as the predicted value, and then output this predicted value to a subtractor. [Prior art documents] [Patent Documents]

[0007] [Patent Document 1] Japanese Patent Application Publication No. 11-220405 [Overview of the project] [Problems that the invention aims to solve]

[0008] However, in the technology described in Patent Document 1, when the input signal changes rapidly, the quantization width determined based on the immediately preceding sampling data may become inappropriate, potentially leading to a decrease in sound quality. This problem is particularly pronounced with input signals that have a small dynamic range.

[0009] This invention has been made in view of the above-mentioned problems, and its objective is to provide an encoder and decoder that can encode and decode an input signal as a high-quality audio signal with sufficiently reduced noise, even when the amplitude fluctuation range of the input signal is large. [Means for solving the problem]

[0010] The means employed by the inventors to solve the above problems are described below. The encoder of the present invention is an encoder for digital modulation and demodulation, and comprises a subtractor that subtracts a feedback signal from an input signal, a quantizer that quantizes the output signal of the subtractor, a feedback delay that delays the output signal of the entire encoding system which is the output signal of the quantizer, a feedback inverse quantizer that inversely quantizes the output signal of the feedback delay to obtain the feedback signal, a plurality of adaptive delays that delay the output signal of the quantizer, and a scale value determination means that determines the scale value of the quantization performed by the quantizer based on the output signals of the plurality of adaptive delays.

[0011] In the encoder of the present invention, the output signal is delayed by a feedback delay unit and fed back to the input signal, thereby enabling quantization of the difference between the input and output signals and reducing noise. In addition to the feedback delay unit, the encoder is equipped with multiple adaptive delay units that delay the output signal. Then, the scale value of the quantizer is determined based on the output signals of these multiple adaptive delays. The scale value is a reference value used to determine the quantization range when performing quantization. When the amplitude of the input signal becomes small, the accuracy of quantization can be improved by decreasing the scale value.

[0012] In this invention, the outputs of these multiple adaptive delay devices are used to adjust the noise level. For example, the multiple adaptive delay devices can be set to slightly different delay times, and the output signal with the largest amplitude among them can be used, or the average amplitude can be used. By determining the quantization scale value based not only on the immediately preceding sampled data but also on the output signals of multiple adaptive delay devices, tracking performance can be improved even in the event of rapid amplitude changes.

[0013] Another means employed by the present invention to solve the aforementioned problems is to configure the plurality of adaptive delay devices so that their respective delay times differ by one sample. By using different delay times for each sample, the historical data used to determine the scale value can be made continuous. Therefore, when there is a sudden change in amplitude, the data used to determine the scale value will not be lost, further improving tracking performance.

[0014] To solve the aforementioned problems, the present invention also employs a configuration in which the scale value determination means determines the scale value using the signal with the largest amplitude among the signals output by the plurality of adaptive delay devices. In adaptive systems, for signals whose amplitude repeatedly increases and decreases, if a large amplitude signal is input immediately after determining the quantization scale value for a small amplitude, the scale value adjustment may not be completed in time, and the data with the large amplitude may not be quantized properly, resulting in reduced accuracy.

[0015] In the above configuration, for signals where the amplitude repeatedly increases and decreases, the scale value is determined based on the largest amplitude value among the amplitudes at multiple delay times. Therefore, a threshold margin can be secured, and quantization can be performed appropriately to prevent degradation of sound quality.

[0016] In the present invention, a decoder having the same characteristics as the above-described configuration can also be provided. The decoder of the present invention is a decoder for digital modulation and demodulation, and comprises: a decoding inverse quantizer that dequantizes an encoded signal; an adder that adds a delayed output signal to the output signal of the decoding inverse quantizer; a decoding delay unit that delays the output signal of the entire decoding system from the output signal of the adder to obtain the delayed output signal; a plurality of adaptive delay units that delay the decoding signal; and a scale value determination means that determines the scale value of the dequantization performed by the decoding inverse quantizer based on the output signals of the plurality of adaptive delay units.

[0017] The decoder of the present invention comprises a plurality of adaptive delay units for delaying an encoded signal, separately from the configuration as a general decoder that inversely quantizes an encoded signal. Then, the scale value of the inverse quantizer is determined based on the outputs of the plurality of adaptive delay units. Even with this configuration of the decoder, decoding can be performed with favorable amplitude followability.

[0018] Further, in the above configuration, the plurality of adaptive delay units may be configured such that their respective delay times differ by one sampling period each. By setting delay times that differ by one sampling period each, past data for determining the scale value can be made continuous in time series. Therefore, when there is a sudden change in amplitude, there is no loss of data for determining the scale value, and followability can be further improved.

[0019] Furthermore, in the present invention, the scale value determining means may be configured to determine the scale value using the signal with the largest amplitude among the signals output from the plurality of adaptive delay units. For a signal in which amplitude increases and decreases repeatedly, by determining the scale value based on the largest amplitude value among the amplitudes at a plurality of delay times, a margin for quantization width can be secured, and appropriate inverse quantization can be performed to prevent sound quality degradation. Effects of the Invention

[0020] The encoder and decoder of the present invention determine the scale value based on amplitude values at a plurality of delay times when performing quantization or inverse quantization. This provides the effect that even when the amplitude fluctuation range of an input signal is large, a high-quality audio signal with sufficiently reduced noise can be transmitted at a low bit rate. Brief Description of Drawings

[0021] [Figure 1]Fig. 1 is a block diagram showing an encoder according to the present invention. [Figure 2] Fig. 2 is an explanatory diagram of scale value determining means according to the present invention. [Figure 3] Fig. 3 is a block diagram showing a decoder according to the present invention. [Figure 4] Fig. 4 is a graph showing spectra of audio signals transmitted by the encoder and decoder according to the present invention and a conventional encoder and decoder. DETAILED DESCRIPTION OF EMBODIMENTS

[0022] Embodiments for carrying out the present invention will be described below with reference to Figs. 1 to 4. Each drawing is schematically described for the sake of explanation, and for example, graphs are described with a part thereof emphasized or omitted.

[0023] The encoder 100 according to the present invention is a component for encoding a signal in the field of digital modulation / demodulation technology. The encoder 100 may be configured as a single independent device, or may be incorporated as a program by a DSP or an FPGA. In the following description, an input signal is explained as an audio signal, but other signals may also be used.

[0024] As shown in Fig. 1, the encoder 100 comprises: a subtractor 1 for subtracting a feedback signal Y1(n) from an input signal X(n); a quantizer 2 for quantizing a subtracted signal X1(n) which is an output signal of the subtractor 1; a feedback delay unit 3 for delaying an output signal Y(n) of the entire encoding system, which is an output signal of the quantizer 2; a feedback inverse quantizer 4 that inversely quantizes an output of the feedback delay unit 3 to obtain the feedback signal Y1(n); a plurality of adaptive delay units 5, 5,... provided separately from the feedback delay unit 3 and configured to delay the output signal Y(n); and scale value determining means 6 for determining a quantization scale value to be used by the quantizer 2 based on outputs of the plurality of adaptive delay units 5, 5,....

[0025] In Figure 1, the input signal X(n) is a sampled input signal, where n is the sampling number. In other words, the discrete input signal X(n) is quantized and the encoded signal is output. However, in this invention, the input signal can also be a continuous analog signal. When an analog signal is used as input, it is sampled before being input to subtractor 1.

[0026] The feedback signal X(n) is delayed by one sample relative to the quantized output signal Y(n) and subtracted from the input signal X(n) by subtractor 1. At this point, since the output signal Y(n) is quantized, it is returned to its pre-quantization state by the feedback inverse quantizer 4 before being added to the input signal X(n). This method is used in differential PCM, and by quantizing only the difference, digital data can be compressed.

[0027] In the configuration shown in Figure 1, this configuration is further enhanced by a plurality of adaptive delays 5·5… and a scale value determination means 6. The adaptive delay units 5·5… in the configuration shown in Figure 1 consist of three delay units, ranging from one that delays the output signal Y(n) by one sample to one that delays it by three samples. These output signals are then input to the scale value determination means 6.

[0028] As shown in Figure 2(a), the scale value determination means 6 determines the scale value for quantization by the quantizer 2 based on the signals input from the three adaptive delay units 5·5…. For example, if the amplitude A2 of a signal delayed by two samples is twice the amplitude A1 of a signal delayed by one sample, and the amplitude A3 of a signal delayed by three samples is half the amplitude, the scale value determination means 6 determines the scale value based on the maximum amplitude A2 of the signal delayed by two samples.

[0029] If the quantizer 2 performs 1-bit quantization, the scale value determination means 6 determines the quantization width based on the maximum amplitude of the signal A2 after a 2-sampling delay, and then determines whether to output a pulse of 1 or 0 for the data of the subtracted signal X1(n) based on that quantization width. The scale value may be the width of the maximum amplitude, or it may be the width obtained by multiplying the maximum amplitude by a predetermined coefficient. In this way, by determining the scale value based on the output of the three delay units with the largest amplitude, signals with smaller amplitudes, such as signals delayed by 1 pulse or signals delayed by 3 pulses, can also be appropriately encoded.

[0030] In another example shown in Figure 2(b), the amplitude A2 of a signal delayed by two samples is half the amplitude A1 of a signal delayed by one sample, and the amplitude A3 of a signal delayed by three samples is twice the amplitude. In this case, the scale value determination means 6 determines the scale value based on the maximum amplitude A3 of the signal delayed by three samples.

[0031] Now, let's consider the case where the amplitude decreases and then increases sharply, as shown in Figure 2(b), and we assume that the scale value is determined only by the amplitude A1 of the signal with a 1-sampling delay. If the scale value is determined based on the amplitude of the previous signal for each sample, and the tracking performance is poor due to processing delays, etc., the system will try to quantize the next large amplitude signal with the scale value determined for the small amplitude. In this case, a small scale value will output either 1 or 0 for the large amplitude, so quantization cannot be performed properly, and noise will increase.

[0032] However, in the encoder 100 of the present invention, the scale value is determined using the largest amplitude value based on the amplitude of the signal from the past three samples. Therefore, even if there is a rapid increase or decrease in amplitude, the fluctuation of the scale value becomes gradual, and proper encoding can be performed.

[0033] In the configuration shown in Figure 1, there are three adaptive delay units 5·5…, each configured to have a delay time that differs by one sample. However, the number of adaptive delay units 5·5… and their delay times can be changed as needed. For example, it is possible to delay by two samples each, or to use three or more adaptive delay units 5·5…, or even just two.

[0034] Next, the decoder 200 for decoding the encoded signal will be explained based on Figure 3. The decoder 200 of the present invention comprises a decoding inverse quantizer 7 that inversely quantizes the encoded signal Y'(n), an adder 8 that adds a delayed output signal X1'(n) to the output signal of the decoding inverse quantizer 7, a decoding delay 7 which is the output signal of the adder 8 and delays the output signal X'(n) of the entire decoding system to obtain a delayed output signal X1'(n), a plurality of adaptive delays 5·5… that delay the encoded signal Y'(n), and a scale value determination means 6 that determines the scale value of the inverse quantization performed by the decoding inverse quantizer 7 based on the output signals of the plurality of adaptive delays 5·5….

[0035] The encoded signal Y'(n) input to the decoder 200 is sampled and quantized digital data, and the decoder 200 decodes it by performing inverse quantization. In the decoding 200 of the present invention, the output signal X'(n) of the entire decoding system is delayed by a decoding delay unit 7, multiplied by a constant a, and added to the output of the decoding inverse quantizer 7 by an adder 8. In addition to this configuration, the encoded signal Y'(n) is delayed by a plurality of adaptive delay units 5·5…, and based on their output signals, the scale value determination means 6 determines the scale value of the inverse quantization in the decoding inverse quantizer 7.

[0036] Similar to the encoder 100 described above, the decoder 200 of the present invention determines the scale value using the largest amplitude value based on the amplitude of the signal from the past three samples. Therefore, even if there is a rapid increase or decrease in amplitude, the fluctuation of the scale value becomes gradual, and decoding can be performed appropriately. Furthermore, in the decoder 200, the number of adaptive delay units 5·5… is not limited to three; the number of adaptive delay units 5·5… and the delay time settings can be changed as appropriate.

[0037] The spectrum of the encoded signal encoded by the encoder 100 of the present invention and the decoded signal by the decoder 200 will be explained with reference to Figure 4. Figure 4(a) shows the spectrum of the input signal. Figure 4(b) shows the spectrum of the transmitted signal when using a conventional encoder and decoder. In this example, the scale value is determined based only on the data from one sample prior, and then quantized. Figure 4(c) shows the spectrum of the signal after transmission when using the encoder 100 and decoder 200 of the present invention. In this example, to determine the scale value, data from 1 to 3 samples prior was used, and the scale value was determined based on the data with the largest amplitude, and then quantized.

[0038] Comparing Figure 4(b) and Figure 4(c), it can be seen that floor noise, particularly at low frequencies, has been reduced. Furthermore, noise components present in the frequency bands immediately before and after the frequency components contained in the input signal have also been reduced.

[0039] The present invention is not limited to the embodiments described above, and can be modified as appropriate within the scope of the claims of the present invention. For example, the number of encoders and their delay times may be varied as needed depending on the nature of the input signal. Furthermore, the number and delay times of the adaptive delay units in the encoder may differ from those in the decoder.

[0040] As described above, in the present invention, when performing quantization or dequantization, the scale value is determined based on the amplitude values ​​at multiple delay times, making it possible to transmit a high-quality audio signal with sufficiently reduced noise at a low bitrate, even when the amplitude fluctuation range of the input signal is large. [Explanation of symbols]

[0041] 100 encoder 200 Decoders 1. Subtractor 2 Quantizer 3. Feedback delay unit 4. Inverse quantizer for feedback 5 Adaptive delays 6. Scale value determination means 7. Inverse quantizer for decoding 8 Adder 9 Decoding Delay A1 Amplitude of the signal with a 1-sampling delay A2 Amplitude of the signal with a 2-sampling delay A3 Amplitude of the signal with a 3-sampling delay a constant X(n) Input signal X1(n) Subtraction signal Output signal of the entire Y(n) coding system Y1(n) Feedback signal X'(n) Output signal of the entire decoding system X1'(n) Delayed output signal Y'(n) encoded signal

Claims

1. An encoder in digital modulation / demodulation, A subtractor that subtracts a feedback signal from an input signal, A quantizer that quantizes the output signal of the subtractor, A feedback delay unit that delays the output signal of the quantizer and the output signal of the entire coding system, A feedback inverse quantizer for inversely quantizing the output signal of the feedback delay device to obtain the feedback signal, Multiple adaptive delays for delaying the output signal of the quantizer, An encoder characterized by comprising a scale value determination means for determining the scale value of the quantization performed by the quantizer based on the output signals of the plurality of adaptive delays.

2. The encoder according to claim 1, characterized in that the plurality of adaptive delays each have a delay time that differs by one sample.

3. The encoder according to claim 1 or 2, characterized in that the scale value determination means determines the scale value using the signal with the largest amplitude among the signals output by the plurality of adaptive delay devices.

4. A decoder in digital modulation / demodulation, A decryption inverse quantizer that dequantizes the encoded signal, An adder that adds a delayed output signal to the output signal of the aforementioned decryption inverse quantizer, A decoding delay unit, which delays the output signal of the entire decoding system, which is the output signal of the adder, to obtain the delayed output signal, Multiple adaptive delay units for delaying the encoded signal, A decoder characterized by comprising a scale value determination means for determining the scale value of the inverse quantization performed by the decoding inverse quantizer based on the output signals of the plurality of adaptive delays.

5. The decoder according to claim 4, characterized in that the plurality of adaptive delays each have a delay time that differs by one sample.

6. The decoder according to claim 4 or 5, characterized in that the scale value determination means determines the scale value using the signal with the largest amplitude among the signals output by the plurality of adaptive delay devices.

Images