JPH03217898A - Envelope follower - Google Patents

Abstract

PURPOSE:To detect an amplitude envelope with high accuracy by removing the high frequency component of a digital signal and setting and controlling the cutoff frequency of a digital low-pass filter according to the increase/ decrease direction of the level. CONSTITUTION:An absolute value converting circuit 10 processes an input digital acoustic signal equivalently by the full-wave rectification of analog signal processing, a linear/logarithmic converting circuit 20 converts the signal into logarithmic table data using such a floating point that the maximum level is O dB, and the digital low-pass filter 30 is controlled by a filter control circuit 40 to detect and outputs the envelope of the input digital acoustic signal. Thus, the envelope detection can be realized by the simple circuit constitution in the form of digital data and the digital signal does not contain any error originating from the accuracy of an analog circuit element, etc., to perform the highly accurate envelope detection.

Description

[Detailed description of the invention] [Industrial application field]

The present invention is intended for use in level meters of various audio equipment, envelope control of musical tone signals in electronic or electric musical instruments (for example, guitar synthesizers, etc.), and is intended for use in envelope control of musical tone signals such as voice signals and musical tone signals. The present invention relates to an envelope follower that detects an envelope and outputs a detected envelope signal.

[Prior art]

Conventionally, this type of device rectifies an input analog acoustic signal by a half-wave or full-wave rectifier using a rectifier circuit using diodes, and outputs the AT analog signal through an integrating circuit composed of a capacitor, a resistor, etc. By this,
The amplitude envelope of the input analog audio signal was detected. (Problem to be Solved by the Invention) However, since the above-mentioned conventional device uses analog signal processing, it is difficult to improve the precision of each circuit element forming the rectifier circuit and the integrating circuit. It was not possible to accurately detect the amplitude envelope of the input analog audio signal.In addition, in recent years, many types of audio equipment use digital signal processing, and in this case, the amplitude envelope of the input analog audio signal cannot be detected with high accuracy. Since the signal is expressed digitally, it has been desired to use this digitally expressed acoustic signal to detect the amplitude envelope of the signal as it is digitally expressed. The purpose is to digitally detect the amplitude envelope of an acoustic signal and to provide an envelope follower that can perform this detection with high precision.

[Means to solve division problems]

In order to achieve the above object, the invention of @1 (above claim 1)
The invention relates to a conversion circuit that inputs a digital acoustic signal representing each instantaneous value of the acoustic signal in time series, converts the input signal into a digital signal that changes on one side of a reference level, and outputs the digital signal; A digital low-pass filter inputs the digital signal from the conversion circuit, removes the high frequency components of the input signal, and outputs it. It also detects the direction of increase or decrease in the level of the digital signal output from the conversion circuit. An envelope follower is configured with a filter control means that sets and controls the cutoff frequency of the digital low-pass filter to be high when the detection direction indicates an increasing direction, and sets and controls the cutoff frequency of the digital low-pass filter to be low when the detection direction indicates a decreasing direction. It's what I did. Further, the structural feature of the second invention (the invention according to claim 2) is that in the envelope follower of the first invention, the conversion circuit inputs a linearly displayed digital acoustic signal, and the conversion circuit also inputs a linearly displayed digital acoustic signal. The reason is that a linear/logarithmic conversion circuit is provided between the digital low-pass filter.

[Operations and effects of the invention l As mentioned above, 4il! In the first aspect of the invention, when a digital audio signal representing each instantaneous value of an audio signal in time series is input, this input signal is converted into a digital signal that changes on one side of a reference level by a conversion circuit. For example, the converted digital signal is converted into a digital signal that changes on the positive side and the negative side, and the converted digital signal is supplied to a digital low-pass filter, where the high frequency components are removed. This digital low-pass filter then outputs a smoothed digital acoustic signal. On the other hand, Filter I! The Il#I means detects the increasing/decreasing direction of the level of the digital signal output from the conversion circuit, and controls the cant-off frequency of the digital low-pass filter according to the detected direction. In this control, when the detection direction represents an increasing direction, the cutoff frequency of the digital low-pass filter is set high and controlled s1, so that the filter output quickly follows the direction in which the level of the digital acoustic signal increases. Further, when the detection direction represents a decreasing direction, the cutoff frequency of the digital low-pass filter is controlled to be set low, so that the filter output follows the decreasing direction of the digital audio signal level slowly. As a result, the peaks of the input digital acoustic signal are traced in order, and digital data representing the amplitude envelope of the input digital acoustic signal is output with good followability. As a result, according to the first invention, the amplitude envelope of the input digital acoustic signal can be easily detected as it is digitally expressed, and the amplitude envelope can be detected without including the error caused by analog processing as in the conventional method. can be detected with high accuracy. Further, the second invention operates almost in the same manner as the first invention, but in this second invention, the input digital signal is displayed linearly, and the linear/logarithmic conversion circuit converts the digital signal from the conversion circuit. Since it is output in logarithmic form, the digital low-pass filter outputs the pair 'i! i Displayed amplitude envelope data is obtained. In this way, according to the second invention, logarithmically expressed amplitude envelope data can be obtained, so in addition to the effects of the first invention, a wide dynamic range of a digital acoustic signal can be expressed with a small number of bits. As a result, amplitude envelope data corresponding to human auditory sense can be obtained, and the use of this data becomes easier. Embodiment An embodiment of the present invention will be described below with reference to the drawings. FIG. 1 shows a block diagram of an envelope follower according to the embodiment. This envelope follower is equipped with an absolute value conversion circuit 10 that inputs a digital acoustic signal that represents each instantaneous value of an acoustic signal such as a voice signal or a musical tone signal in time series. The digital audio signal is a linear representation of each instantaneous value using "two's complement", and is composed of, for example, 16 bits. The absolute value conversion circuit 10 includes 15 exclusive OR circuits E.
Consisting of XOR to EXOR15, each exclusive OR circuit EXOR+ to EXOR+s inputs the most significant bit MSB (sign bit) of the input digital audio signal to one input, and inputs the most significant bit MSB (sign bit) of the input digital audio signal to each other input. Each bit of the same audio signal except the most significant bit MSB is inputted to each of the respective bits. This allows a positive input digital acoustic signal, e.g.
00...Oi" (16 bits) is the most significant bit MS
Excluding B<roo-01J (15 bits) is output as a signal. In addition, a negative input digital audio signal, such as "111...IOJ (16 bits), is processed as a signal of "00...OIJ (15 bits)" with the most significant bit MSB removed and the bits other than the MSB inverted. As a result, this absolute value conversion circuit 1
From 0, the input acoustic signal is subjected to absolute value conversion (full-wave rectification), and a digital signal is output in which the absolute value is linearly expressed in "two's complement". A linear/logarithmic conversion circuit 20 including an encoder 21, a shifter 22, and an inverter circuit group 23 is connected to the absolute value conversion circuit 10. The encoder 21 converts the output from the absolute value conversion circuit 10 into M size {ffirll...L
IJ (15 bits) is regarded as 0 decibel, and the exponent part of the logarithmic conversion data, which is a floating point representation of the same output, is represented in 3 bits and output. That is, this encoder 21 converts the 15-bit data into r
It is configured with a logic circuit so as to convert it into 3-bit data representing IJ. Note that this 3-bit data is
In the order of rooOJ, ro・・rl 1 1J, O, 01J ro10J・−6,−12・・・・. This will convert it as follows. r111111111...1''~r10000000
0...0"r011111111...1"~r01
0000000...0"r001111111...
1” ~ “001000000...0” ro001 1
11 11・IJ ~r 000100000・-OJ
The data representing -42 decibels and the above 15 bits are as follows: "000""001"ro10"ro11"rooooooooxt...1"~r00000001
0.0" → "111" The shifter 22 displays and outputs the mantissa part of the logarithm conversion data expressed as a floating point number in 3 bits. The 15 bit data from the absolute value conversion circuit 10 is outputted as 3 bits from the encoder 21. By shifting the bit to the upper side by the data value of the bit, the most significant bit MSB of the output is always set to "1", and 3-bit data of the second to fourth upper bits is output. For example, the rljllllllll...1''~r1
00000000...O" is not bit shifted,
The 3-bit data of the upper 2nd to 4th bits rllll'' to rooO+ is output, and the r000111111 and r000111111.
...1" to roo0100000...0" are shifted by 3 bits, and the upper 2nd to 4th bits are "111" to "0".
3-bit data of ``00'' is output. Of course, in such a case, the 15-bit data from the absolute value conversion circuit 10 may be shifted to the upper side by a number of bits that is "1" larger than the output value from the encoder 21, and the upper 3 bits may be output. . The 3-bit mantissa data from the shifter 22 is combined with the 3-bit output data from the encoder 21 via the inverter circuit group 23, and is output as 6-bit level data. Inverter circuit #23 is inverter circuit I
NV+. It is composed of INVIINv3, and inverts each bit output from the shifter 22 and outputs it. The reason why the mantissa data is inverted by the inverter circuit group 23 is as follows. That is, as mentioned above, when creating the exponent part data, rllll...1'' to r1000...0''
The data value is 0 decihers, r0111・IJ − r0
100-OJ(i') data value -6 dB, r00
11...1"~roo10...0" data value -
12 decibels, etc., and the shifter 22 always outputs data values from "1l1" to "000" with "1" of the most significant bit MSH deleted as mantissa data. In such a case, displaying this mantissa data is equivalent to viewing the triangular hunting part of -6 decibels or less in FIG. 2 onto the triangular hunting part of O to -6 decibels. On the other hand, music record O~-6
As shown in FIG. 3, which shows the decibel part enlarged, regarding the mantissa output from the shifter 22, O decibel as the smallest absolute value (the maximum level) is "
111'', and the largest absolute value (-6 decibels, the smallest) is expressed as roOOJ, and the output of the shifter 22 shows an opposite change tendency as a negative decibel display. Therefore, the inverter circuit group 23 converts the value of 0 to -6 decibels from roOOJ to "1".
A linear approximation is performed so that the value changes sequentially up to 11'' to ensure consistency with the exponent data. As can be understood from this explanation, the 6-bit level data consisting of the exponent and mantissa parts is expressed as a negative decibel value whose absolute value increases as the absolute level of the input digital audio signal decreases. has been done. To this 6-bit level data, 4-bit digital data consisting of "roooo" is added on the lower side, that is, the 6-bit level data. The bit level data is expanded to 10 bits and supplied to the digital filter 30. The digital filter 3o includes a subtracter 31, a multiplier 32, and an adder 3.
It consists of a delay circuit 34 consisting of a shift register of 3 and 1 stage and 10 bits. The subtracter 31 subtracts the level data from the delay circuit 34 from the supplied level data, and supplies the subtraction result to the adder 33 via the multiplier 32. An adder 33 adds the output from the multiplier 32 and the output from the delay circuit 34,
The addition result is supplied to the delay circuit 34 and output as detection envelope data. Note that this digital filter 30 makes the output of the adder 33 a low-pass output, and at the same time makes the output of the subtracter 31 a high-pass output. Moreover, in the same filter 30,
The cutoff frequency characteristic of the low bass output with respect to the input signal is changed and controlled by the gain coefficient g supplied to the multiplier 32. That is, as the gain coefficient g becomes larger, the cutoff frequency in the low-pass characteristic of the digital filter 30 becomes larger. A filter control circuit 40 consisting of a sign/negative discrimination circuit 41 and a gain memory 42 is connected to the output of the subtracter 31, and the control circuit 40 controls the gain of the multiplier 32. The positive/negative discrimination circuit 41 is constituted by a comparator, and determines whether the result of subtraction by the subtracter 31 is positive or negative and outputs the result. The gain memory 42 supplies a gain coefficient g to the multiplier 32 under the control of the positive/negative discrimination circuit 41, and supplies a predetermined large gain coefficient g+ to the multiplier 32 when the determination result is negative, and When the result is positive, a predetermined small gain coefficient g2 (g+>g2) is supplied to the multiplier 32. As a result, when the absolute level of the input digital audio signal increases, the cutoff frequency in the low-pass characteristic of the digital filter 30 becomes higher as shown by the solid line in FIG. 4, and the envelope detection follows faster. Further, when the absolute level of the acoustic signal decreases, the cutoff frequency decreases as shown by the two-dot chain line in FIG. 4, and the tracking of the envelope detection becomes slow. This is because although the level data logarithmically expressed by the linear/logarithmic conversion circuit 20 represents a negative decibel value, the sign bit representing the negative value of the data is omitted. As a result, the peaks of the input digital acoustic signal are sequentially traced from the digital filter 30.
Digital data representing the amplitude envelope of the acoustic signal is output. According to the embodiment configured as described above, when a digital acoustic signal is input, the absolute value conversion circuit 20 performs processing equivalent to full-wave rectification in analog signal processing on the input digital acoustic signal, and then The linear/logarithmic conversion circuit 20 converts the digital acoustic signal into logarithmic display data using a floating point whose maximum level is O decibels, and the digital filter 30 is Ig-controlled by the filter control circuit 40 to convert the input digital acoustic signal. Detect and output the envelope. As a result, it is possible to detect the envelope of a digital acoustic signal using a simple circuit configuration without changing the digital data, and the error caused by the accuracy of analog circuit elements is not included in the digital signal, allowing highly accurate envelope detection. becomes possible. Furthermore, the linear/logarithmic conversion circuit 20 displays the input digital audio signal logarithmically (decibels) using a floating point, and the envelope detection value is also displayed in decibels. The processing after the circuit 20 can be done with a small number of bits, and the detected envelope value can be expressed in a form that corresponds to human auditory sense. In the above embodiment, the absolute value conversion circuit 10 converts the exclusive circuit EXO R + to EX○R
+5 to perform digital processing equivalent to full-wave rectification in analog processing to the input digital acoustic signal, but digital processing equivalent to half-wave rectification in analog processing may also be performed. In such a case, AND circuits AND+ to AND1s are used in place of the exclusive circuits EXOR+ to EX○RI6, and one input of each of the AND circuits AND+ to AND+s is used. is the most significant bit M of the input digital audio signal.
A signal obtained by inverting the SB (sign bit) may be commonly supplied, and the lower 15 bits of the input digital acoustic signal may be respectively supplied to the other input. As a result, if the most significant bit MSB (sign bit) is "0", that is, the input digital acoustic signal is positive,
The signal is output as is, and the most significant bit MS
If B (sign bit) is "1", that is, the input digital acoustic signal is negative, a digital signal representing "0" is output, and the half-wave rectification is realized. Furthermore, instead of using "O" as the reference level as in the full-wave and half-wave rectification, other values can be used as the reference level, and digital acoustic signals having a level higher or lower than that can be output in the form of slices. You can do it like this. In such a case, a comparator and a gate circuit are used, the comparator compares the input digital acoustic signal with a predetermined reference level value, controls the conduction/non-conduction of the gate circuit according to the comparison result, and inputs the signal. The supply of the digital acoustic signal to the linear/logarithmic conversion circuit 20 may be controlled. Furthermore, when the linear/logarithmic conversion circuit 20 of the above embodiment converts the linear display data regarding the mantissa part into a logarithm (decibel) display, the inverter circuit group 23 is used for linear approximation, but as shown in FIG. As shown, the same circuit group 23
By using the linear/logarithmic conversion table 24 instead, data related to the mantissa may also be converted from linear display to logarithm (decibel) display. This eliminates the error caused by the linear approximation in the above embodiment, and improves the envelope detection accuracy of the envelope follower. Furthermore, in the filter control circuit 40 of the above embodiment, the positive/negative discrimination circuit 41 connected to the output (high-pass output) of the subtracter 31 detects the rise and fall of the input digital acoustic signal. As shown, in place of the discrimination circuit 41, a comparator 44 compares the magnitude of input data to the digital filter 30 and the output of a delay circuit 43 that delays the same data by one bit. and a fall may be detected.

[Brief explanation of drawings]

FIG. 1 is an overall block diagram of an envelope follower according to an embodiment of the present invention, FIGS. 2 and 3 are graphs for explaining the conversion operation of the linear/logarithmic conversion circuit of FIG. 1, and FIG. A graph showing the low-pass characteristics of the digital filter shown in FIG. 1, and FIG. 5 shows the logarithmic/linear conversion circuit 20 shown in FIG.
FIG. 6 is a block diagram showing a modification of the filter control circuit of FIG. 1. FIG. Explanation of symbols 10... Absolute value conversion circuit, 20... Linear/logarithmic conversion circuit, 21... Encoder, 22... Shifter, 2
3... Inverter circuit details, 24... Linear/logarithmic conversion table, 30... Digital filter, 31.
... Subtractor, 32... Multiplier, 33... Adder, 34... Delay circuit, 40... Filter control circuit,
41... Positive/negative discrimination circuit, 42... Gain memory, 43
...delay circuit, 44... comparator.

Claims (2)

[Claims] (1) A conversion circuit that inputs a digital audio signal that represents each instantaneous value of the audio signal in time series, converts the input signal into a digital signal that changes on one side of a reference level, and outputs the digital signal, and the conversion circuit a digital low-pass filter that inputs a digital signal from the input signal, removes high-frequency components of the input signal, and outputs the filter; filter control means for controlling the cutoff frequency of the digital low-pass filter to be set high when the detection direction indicates a direction, and for controlling the cutoff frequency of the digital low-pass filter to be set low when the detection direction indicates a decreasing direction; Features an envelope follower. (2) The conversion circuit inputs a linearly displayed digital audio signal, and a linear/logarithmic conversion circuit is provided between the conversion circuit and the digital low-pass filter. envelope follower.