close

Вход

Забыли?

вход по аккаунту

?

DESCRIPTION JP2006303618

код для вставкиСкачать
Patent Translate
Powered by EPO and Google
Notice
This translation is machine-generated. It cannot be guaranteed that it is intelligible, accurate,
complete, reliable or fit for specific purposes. Critical decisions, such as commercially relevant or
financial decisions, should not be based on machine-translation output.
DESCRIPTION JP2006303618
PROBLEM TO BE SOLVED: To provide a speaker driving system for driving a speaker by a digital
sound signal, capable of reproducing a sound with good sound quality by reducing noise to a
reproduction sound generated on a small scale and due to an error accompanying quantization.
SOLUTION: This is a speaker drive system for driving a speaker 1 based on a digital acoustic
signal, and the gradation adjustment means 4 for adjusting the gradation of the digital acoustic
signal, and the gradation adjustment means 4 for maintaining gradation information. Serial
conversion means 5 for converting the output of the digital audio signal into serial data, the noise
caused by the quantization error during generation of the data of the digital acoustic signal and
the signal processing, and the distortion caused by the discontinuity of the data are discretized
And noise reduction means 6 for reduction by interpolation between the data. [Selected figure]
Figure 1
Speaker drive system
[0001]
The present invention relates to a speaker drive system for driving a speaker based on a digital
sound signal.
[0002]
The speaker vibrates the diaphragm with an electric signal to convert it into air vibration and
emits sound (sound and music).
10-04-2019
1
Thus, in general, this electrical signal is an analog signal having a waveform similar to sound. On
the other hand, in recent years, many devices that handle sound by digital signal processing, such
as portable telephones and portable music players, CD players, MD players, etc., are in
widespread use. In order to make the speaker installed or connected to these devices ring, these
devices may be equipped with a D / A converter (Digital to Analogue Converter), and drive the
speakers by the analog converted electrical signal. There are many.
[0003]
For example, in a mobile phone, a digital signal is used for signal processing and transmission,
and is converted into an analog signal by a D / A converter provided immediately before a
speaker. The analog signal has continuous amplitude and time changes, and can faithfully
reproduce sound, but is susceptible to external electromagnetic noise. On the other hand, digital
signals are discrete signals, and although the signal waveform itself does not faithfully represent
sound, it is not susceptible to external electromagnetic noise, and high-speed signal processing,
transmission, etc. are possible. Taking advantage of such features, digital signals are used in
portions that are susceptible to electromagnetic noise from circuits and transmission lines, and
are converted back to analog signals immediately before a speaker. That is, sound is reproduced
faithfully while reducing the influence of electromagnetic noise as much as possible.
[0004]
Here, although there are various methods for D / A conversion, a method using a capacitor is
generally used as an operation principle. That is, a digital signal which is a sharp signal
displacement is referred to as a high level (hereinafter referred to as “H” as appropriate). The
capacitor is charged at the time of), and it is called low level (hereinafter, referred to as "L" as
appropriate). The capacitor is discharged at the time of). In this way, abrupt changes in the digital
signal are smoothed and converted from discrete signals to continuous signals.
[0005]
By the way, as long as digital signals are also present in the natural world as electric signals, they
can be regarded as analog signals which are not physically separated and whose change is very
steep. Also, even if a steep change analog signal is given to the diaphragm of the speaker, the
10-04-2019
2
diaphragm can not follow this signal. In this respect, the diaphragm works in the same way as the
above-mentioned capacitor. Therefore, in principle, it is possible to emit sound by directly
inputting a digital signal to a speaker. Patent Document 1 (Japanese Patent Application LaidOpen No. 9-46787) shows a circuit for driving a speaker by inputting a digital signal as described
above. According to this, a digital sound signal represented by a plurality of digital signals is
converted into a single bit (serial) PWM (Pulse Width Modulation) waveform, and the PWM
waveform is input to a speaker to reproduce sound. There is.
[0006]
The speaker drive circuit described in Patent Document 1 has a small circuit scale, and is very
suitable for mounting on a small device such as a mobile phone or a portable music player.
However, in order to apply this principle in practice, it is necessary to solve the problems
described below.
[0007]
For example, in a mobile phone, two types of sound signals, that is, speech of a conversation and
music such as a ringing melody, are handled. The sampling frequency and the number of
gradations at the time of digitizing (quantizing) these acoustic signals are 8 kHz for the audio
signal and 14 bits, and 44.1 kHz for the music signal and 16 bits. The 14 bits correspond to
16384 gradations, and the 16 bits correspond to 65536 gradations. However, for ease of
explanation, they will be expressed by the number of bits as appropriate. The difference between
the audio signal and the music signal is that the audio signal has a smaller amount of change
with respect to time than the music signal, and the frequency band to be reproduced is narrow.
[0008]
Here, when the audio signal thus digitized is serialized while maintaining the gradation, the
frequency of one gradation is about 131 MHz (= 8000 Hz * 16384 gradation) for the audio
signal and about 2. for the music signal. It will be 89 GHz (= 44100 Hz * 65536 gradations). This
serialization is performed by a processor such as a microcomputer or DSP (Digital Signal
Processor) incorporated in the device. Therefore, it is necessary to use a processor capable of
high-speed processing even for voice signals of relatively low frequency.
10-04-2019
3
[0009]
Therefore, if the number of gradations is reduced, the frequency per gradation is also reduced, so
that the load on the processor can be alleviated, but the sound quality is reduced according to the
reduced number of gradations. As mentioned above, the analog signal is continuous and the
digital signal is discrete. When an original acoustic signal that is an analog signal is digitized
(quantized), an error called quantization error occurs. Here, if the number of gradations of the
digitized signal is further reduced, that is, requantized, in addition to the quantization error, the
requantization error is also added. The quantization error and the requantization error manifest
as audible noise to the reproduced sound. Therefore, in audio equipment and the like that require
high sound quality, a quantization error and a requantization error are reduced by including a
circuit called a ΔΣ modulator (see, for example, Patent Document 2 shown below). ).
[0010]
JP-A-9-46787 (Paragraphs 6-14, FIG. 1, 3) JP-A-6-335082 (Paragraphs 1-8)
[0011]
However, the ΔΣ modulator is a circuit that is complicated in processing and requires
peripheral components.
For example, in the case of an audio signal such as an audio signal of a mobile phone which does
not require the sound quality like a music signal, the circuit scale can be increased more than
necessary while the sound quality improvement beyond the request can be expected. That is, in
order to obtain a speaker drive circuit small enough to be integrated with a speaker, if the
principle of Patent Document 1 driven directly by a digital signal is used without using a D / A
converter, the sound quality deterioration due to noise etc. There is a need to. However, for this
reason, it is not preferable to always adopt a ΔΣ modulator as in Patent Document 2.
[0012]
The present invention has been made in view of the above problems, and is a small-scale digital
acoustic signal that can reproduce noise with a good sound quality by reducing noise to the
reproduced noise caused by an error accompanying quantization. It aims at providing the
speaker drive system which drives a speaker.
10-04-2019
4
[0013]
The characteristic configuration of the speaker driving system according to the present invention
for achieving the above object is a speaker driving system for driving a speaker based on a digital
sound signal, and a gradation adjusting unit for adjusting the gradation of the digital sound
signal And serial conversion means for maintaining the gradation information and converting the
output of the gradation adjustment means into serial data; noise caused by an error in generation
of data of the digital audio signal and quantization in the signal processing; And noise reduction
means for reducing distortion caused by data discontinuity by interpolation between the data.
[0014]
According to this feature configuration, quantization errors caused by digitization and tone
adjustment are reduced by interpolation between data.
For example, by interpolating new data between two data adjacent in time, it is possible to
improve the continuity of a digital signal which is discrete data.
In order to interpolate this data, simple processing such as interpolation of the average value of
two data can be used without using complicated processing such as ΔΣ modulation.
Interpolating data is equivalent to apparently raising the sampling frequency, and can correct the
distortion caused by the discontinuity and reduce the quantization error. That is, distortion can
be reduced by improving the continuity of data by interpolation. Also, by becoming equivalent to
raising the sampling frequency, it is possible to move the noise generation area called so-called
aliasing noise to a higher frequency band. As a result, aliasing noise is out of the audio frequency
range, and noise can be reduced.
[0015]
In addition, it is preferable to include second noise reduction means for reducing periodic
background noise involved in the signal processing.
[0016]
10-04-2019
5
Digital signal processing circuits have periodic signal sources such as the frequency of a
sampling clock and the period of one data.
Then, due to this periodic signal source, background noise may be superimposed on the
reproduced sound output from the speaker. Therefore, for example, when noise reduction means
for adding a noise reduction signal generated based on one cycle of data of serial data for driving
a speaker and a signal for driving a speaker is provided, the background depending on the cycle
of data is provided. Noise can be suppressed to further improve the sound quality.
[0017]
Further, the noise reduction means interpolates by equally distributing the data according to the
difference between the data to reduce the noise and the distortion, and the second noise
reduction means interpolates the data It is preferable to reduce the background noise based on a
signal interpolated by repeating constant data at least between them.
[0018]
Interpolation between discretized data includes equidistant interpolation by equal ratio
distribution according to the difference between the data, and equivalence interpolation by
repeating any one of those data.
Interpolation of the discretized data can be performed smoothly by interpolating between the
discretized data by equal ratio distribution according to the difference between the data. As a
result, noise and distortion can be favorably reduced. On the other hand, since the same value
interpolation is used by repeating the data which exist surely, it can be said that it is an
interpolation which paid attention to, for example, "the sound afterglow" to the equal interval
interpolation which paid attention to continuity. Although the above-mentioned background
noise appears as a periodic signal, if this is caused by the "aftertone", the background noise can
be favorably reduced based on the interpolation signal in the case of performing equivalent
interpolation. . Also, even if the same value interpolation is not strictly performed, a pulse signal
with a duty of 50% (corresponding to a median value as an audio signal) corresponding to the
cycle of one gradation in the case where interpolation processing is added. The background noise
of a constant period can be reduced even by the method of interpolating as a constant value.
According to the experiment of the inventor of the present invention, distortion and noise
associated with quantization and background noise are favorably obtained by using a signal
obtained by interpolating data with equal ratio distribution and a signal obtained by interpolating
10-04-2019
6
data with a constant value. It is clear that it can be reduced.
[0019]
Preferably, the gradation adjustment means adjusts the gradation of the digital acoustic signal by
subtracting the gradation of the digital acoustic signal.
[0020]
As described above, the acoustic signal digitized based on the general sampling frequency and
the number of gradations has too much data amount as it is, and is difficult to process in a smallscale system.
Further, if the number of gradations represented by the converted serial signal is large, the
vibration width of the diaphragm of the speaker, which vibrates in accordance with the serial
signal, also has the same type as the number of gradations. The vibration width of the diaphragm
also relates to the magnitude of the sound, and as the number of types of vibration width
increases, the magnitude of the size of the sound emitted from the speaker also increases. As a
result, it may be possible to emit a sound that is so loud that the voice can not be heard.
Therefore, if this gradation number is reduced by the gradation adjustment means and adjusted
to an appropriate amount of data, processing in a small scale system can be enabled, and an
acoustic signal can be reproduced at an appropriate volume. .
[0021]
Preferably, the serial conversion means converts the digital audio signal into serial data by pulse
width modulation.
[0022]
If so-called serial-to-parallel conversion is performed by simply converting parallel data into
serial data, gradation information can not be obtained unless information is extracted by
decoding serial data or the like.
In other words, it is necessary to grasp the beginning and end of the data, to find one data block
10-04-2019
7
based on this, and to extract the gradation information from there. However, by performing pulse
width modulation simultaneously at the time of serialization, serial data itself comes to have
analog gradation information. As a result, it becomes possible to directly input this serial data to
the speaker, and if this serial data is corrected, distortion and noise can be eliminated and the
speaker drive can be mounted on a small device with good sound quality and good sound quality.
System can be provided.
[0023]
Hereinafter, embodiments of the present invention will be described based on the drawings. FIG.
1 is a block diagram showing an example of a speaker drive system according to the present
invention. This speaker drive system is used, for example, to output sound in a mobile phone.
Then, preferably, the speaker drive system shown in FIG. 1 is integrated and integrated with the
speaker 1 so as to be housed in a mobile phone which is required to save space. In the circuit
including the above-described ΔΣ modulator, the scale is too large, and integration with a small
speaker 1 can not be achieved. However, with the loudspeaker drive system according to the
present invention described below, the scale is small and can be sufficiently integrated. The
mobile phone is also provided with a speaker for outputting music like a ringing tone, but in the
present embodiment, an example in the case of handling an audio signal will be described.
[0024]
An audio signal used for a mobile phone generally has 14-bit gradation. However, the voice
signal transmitted when making a so-called call between mobile phones is compressed to 8 bits
in consideration of communication efficiency and the like. As this compression method, different
systems are adopted depending on the region. For example, in Japan and the United States, a
compression format called μ-law is used, and in Europe, a compression format called A-low is
used. The audio signal has a narrow frequency band compared to the music signal, and the audio
quality is not required to be so high, so the compression is non-linear in any compression format,
and the audio information has distortion. . Therefore, in order to reproduce the audio signal S0
received by the mobile phone through the speaker 1, it is necessary to decompress the
compressed audio signal S0 first. The decompression unit 2 is a decoder circuit that
decompresses the audio signal S0 compressed into 8-bit data into 14-bit data.
[0025]
10-04-2019
8
The audio signal S1 expanded to 14 bits is coded as shown in FIG. 2 (b) corresponding to the sine
wave shown in FIG. 2 (a). The most significant bit S1 [13] of the audio signal S1 represents a sign,
where 1 is the positive direction side (upper side in FIG. 2A) from the amplitude center, and 0 is
the negative direction side from the amplitude center (FIG. 2 (a ) Lower side is shown. The lower
13 bits S1 [12: 0] of the audio signal S1 represent the amplitude, which is the largest value at the
center of the amplitude (point A in FIG. 2A), and the maximum amplitude (points B and C in FIG.
2A). Is the smallest number. In the voice signal S1 when viewed in 14 bits, the minimum value C
point of the sine wave shown in FIG. 2A becomes 0 in decimal notation, and thereafter the
numerical value increases up to 8192 at the point A as the waveform rises. Then, the value
greatly jumps from 8192 to 16383 in decimal notation with the point A in between, and
thereafter the value decreases to 8192 at the point B as the waveform rises. Therefore, even if
the code of the audio signal S1 is a code suitable for digital signal processing, it can not be said
that it is not a code representing a physical quantity suitable for driving the speaker 1.
[0026]
Therefore, the code converter 3 shown in FIG. 1 converts the code of the audio signal S1 into a
code suitable for driving the speaker 1 to obtain a new audio signal S2. The code after conversion
is as shown in FIG. 2 (c). With the rise from point C of the waveform shown in FIG. 2A to point C
via point A, it rises continuously as 0, 8191, 8192 and 16383. That is, the converted audio signal
S2 has a code representing a physical quantity suitable for driving the speaker 1. This conversion
can be easily performed by leaving the most significant bit of the audio signal S1 as it is and
inverting the lower 13 bits when the most significant bit is 1. As an example, as shown in FIG. 3,
it can be realized by a logic circuit using an Ex-OR gate.
[0027]
The above-described two expanders 2 and code converter 3 are input signal processing means
for converting digital audio signals input in various formats into digital audio signals suitable for
driving the speaker 1. As described above, the compression method includes the μ-low method
and the A-low method, and if the input signal processing means can be switched according to the
destination, one system can cope with a plurality of methods. . Further, the input signal
processing means may be provided with determination means for determining these methods,
and switching may be performed based on the result of the automatic determination.
10-04-2019
9
[0028]
When the code conversion unit 3 performs conversion to a digital sound signal (audio signal S2)
suitable for driving the speaker 1, next, the gradation adjustment unit (gradation adjustment
unit) 4 performs gradation adjustment. As described above, for example, in a mobile phone, the
sampling frequency at the time of digitizing (quantizing) an audio signal is 8 kHz. When
serialization is performed while maintaining the current gradation number of 14 bits, the
frequency of one gradation becomes about 131 MHz (= 8000 Hz * 16384 gradations) for an
audio signal. This serialization is performed by a processor such as a microcomputer or DSP
(Digital Signal Processor) incorporated in the device. Therefore, when the above frequency is
required, it is necessary to use a processor capable of high speed processing. Therefore, the
number of gradations is reduced to slow down the frequency necessary for processing.
[0029]
In this embodiment, the case of converting into 9 bits that can obtain the clearest speech
experimentally when using a processor with a system clock of 33 MHz will be described as an
example. As will be described later, a pulse width modulation unit (serial conversion means) 5 is
provided at the subsequent stage of this gradation adjustment unit 4, and here it is converted to
a serial sound signal (audio signal S4) pulse width modulated. Ru. The converted serial signal
vibrates the diaphragm of the speaker 1 according to the pulse width. At this time, when the
variation range of the pulse width is large, that is, the number of gradations is large, the
difference between the minimum pulse width and the maximum pulse width becomes large.
Along with this, the magnitude of the magnitude of the sound emitted from the speaker 1 also
increases, and in a portion where the pulse width is wide, the sound may be louder as the sound
can not be heard. Therefore, also for this reason, the gray level is adjusted by the gray level
adjusting unit 4 before input to the pulse width modulating unit 5. In this example, based on the
experimental results, conversion into 9 bits that can most clearly hear speech is performed.
However, the present invention is not limited to this, and as shown in FIG. 1, the output of the
gradation adjustment unit 4 is N bits, and may be appropriately selected according to the system.
[0030]
The gradation adjustment unit 4 adjusts the number of gradations by reducing the lower 5 bits of
the 14-bit audio signal S2. That is, as shown in FIG. 2C, the audio signal S2 [13: 5] is set as the
10-04-2019
10
audio signal S3 [8: 0] after gradation adjustment. The tone adjustment unit 4 can also be
integrated with the code conversion unit 3 shown in FIG. 3 if only such simple reduction is
performed. That is, in the circuit of FIG. 3, the Ex-OR gate for converting the audio signal S1 into
the audio signal S2 may not be provided for the lower 5 bits. In this case, the effect of reducing
the size of the circuit can also be obtained.
[0031]
Further, instead of the simple reduction as described above, the gradation adjustment unit 4 may
perform rounding such as rounding off. For example, as shown in FIG. 4, adding the 10th most
significant bit (S2 [4]) of the audio signal S2 to the upper 9 bits (S2 [13: 5]) of the audio signal S2
before gradation adjustment. Then, rounding off can be realized. Since an overflow may occur at
this time, the overflow signal CY from the adder is used to mask all bits of the audio signal S3 to
1 and maximize the value at the overflow. The adder is a 9-bit + 1-bit one-increment adder, and
can be realized by a small-scale circuit even if an overflow circuit using an OR gate is added.
[0032]
As described above, when the tone adjustment unit 4 sets the audio signal S3 to 9 bits, the
number of tones is 512. The pulse width modulation (PWM) unit 5 as serial conversion means
converts the parallel data of 512 gradations into serial data. When the tone of the audio signal S3
is maintained and serialized, the frequency of one tone is approximately 4.1 MHz (= 8000 Hz *
512 tones). Then, as shown in FIG. 5, pulse width modulation can be performed by continuing
pulses of up to about 240 ns (= 1 / 4.1 MHz) per gradation according to the gradation. For
example, when the gray level indicated by the audio signal S3 [8: 0] is 1, it is a pulse for one
pulse width P, and when the gray level is 2, it is a pulse for two pulse widths P. Thereafter, the
number of pulse widths P is increased according to the gradation, and when the gradation is the
maximum of 511, the pulse width of 511 pulses of the pulse width P is set. In this case, although
the pulse width is 0 to 511 corresponding to the gradation 0 to 511, the design change such as
the pulse width 1 to 512 corresponding to the gradation 0 to 511 is It may be performed as
appropriate.
[0033]
By the way, as described above, in the present embodiment, control of the speaker drive system
10-04-2019
11
is performed using a processor with a system clock of 33 MHz. Therefore, the frequency per
gradation can be increased to 33 MHz in calculation. Alternatively, one audio signal S3 can be
processed multiple times. In the case of the present embodiment, eight times (= 33 / 4.1) of
processing are possible, and interpolation processing can be performed in seven of these. Here,
the interpolation processing is to fill in between the data of the audio signal which is a discrete
numerical value and the data with data of an appropriate value. For example, when seven
interpolation processes are performed in addition to one normal process, eight processes in total
are performed, which is equivalent to raising the sampling frequency of 8 kHz to approximately
64 times 64 kHz. This is called oversampling. The oversampling frequency is higher than 44.1
kHz, which is the sampling frequency of the music signal described above, and has an effect of
improving the sound quality. That is, coupled with interpolation processing to be described later,
quantization noise resulting from quantization and requantization can be reduced. In addition,
aliasing noise generated when a digital signal is converted into an analog signal can also be
moved to a high frequency band, which can also improve the sound quality.
[0034]
FIG. 6 is a diagram for explaining aliasing noise. FIG. 6A is an explanatory view schematically
showing the appearance of aliasing noise in the case where oversampling is not performed. The
original audio signal has a sampling frequency of 8 kHz and shows a frequency spectrum
distributed at ± 4 kHz from the center as shown in FIG. Then, when oversampling is not
performed, the spectrum of the original audio signal appears around 8 kHz with 4 kHz as the
folding point, and this becomes aliasing noise. The same aliasing noise appears every 8 kHz
thereafter. The upper limit of the human audible range is approximately 20 kHz, but as shown in
FIG. 6A, aliasing noise is present in the audible range. Therefore, it actually sounds as noise.
[0035]
FIG. 6B is an explanatory view schematically showing the appearance of aliasing noise when
oversampling is performed. For example, in the case of performing eight times of processing
including seven times of oversampling, the turning point moves to 32 kHz, which is a frequency
eight times higher than that in the case of no oversampling. Then, the center of the aliasing noise
folded at this folding point also moves to 64 kHz, which is a frequency eight times higher. Since
the aliasing noise having a width of 8 kHz centering on 64 kHz does not reach within the audible
range of 20 kHz or less, it does not affect the output sound from the speaker 1. Thus, by
performing oversampling processing, it is possible to reduce quantization noise due to
quantization and requantization.
10-04-2019
12
[0036]
Oversampling is achieved by interpolating the data that is actually lost due to quantization. There
are various interpolation methods, and an example thereof is shown in FIG. FIG. 7A shows an
analog signal which is a source signal. FIG. 7B shows a digital signal obtained by quantizing this
analog signal based on the sampling frequency. FIG. 7C is an example of the same value
interpolation in which each data is interpolated with one or more (here, three) data by using each
quantized data a plurality of times (here, four times). is there. FIG. 7 (d) is an example of
equidistant interpolation interpolated by data proportionally distributed among the quantized
data. As apparent from the comparison between FIGS. 7 (c) and 7 (d), the equidistant
interpolation has a form closer to the analog signal which is the source signal. Therefore,
distortion due to quantization can be favorably reduced by using this equidistant interpolation.
[0037]
In the block diagram shown in FIG. 1, the output of the pulse width modulation unit 5 which is a
serial conversion unit is oversampled, and the equal interval interpolation unit 6a as a noise
reduction unit 6 performs equal interval interpolation. The noise reduction means 6 outputs the
speech signal S5 with reduced quantization noise and distortion.
[0038]
When this audio signal S5 is input to the speaker 1, although the cause is not clear, a sine wave
of several kHz appears and this becomes background noise. According to the experimental
observation, the frequency of this background noise had a frequency characteristic substantially
matching the frequency of the system clock of 33 MHz. Therefore, when the system clock is
inverted and added to the audio signal S5 which is the output of the noise reduction means 6,
background noise can be reduced. In addition, background noise can also be added by adding
together a pulse signal of a 50% duty (equivalent to a PWM waveform indicating a gradation at
the center of amplitude) corresponding to the period of one gradation when adding oversampling
processing to the audio signal S5. It could be reduced. Further, in the experiment, background
noise was able to be reduced more satisfactorily by inverting the signal obtained by performing
oversampling by equivalent interpolation and adding it to the audio signal S5. Therefore, as
shown in FIG. 1, the second noise reduction means 7 is provided, in which the same value
10-04-2019
13
interpolation is performed to obtain the audio signal S6, and the inverted signal of the audio
signal S6 and the audio signal S5 after the equal interval interpolation are obtained. It was set as
the composition to add. Thereby, background noise can be reduced well.
[0039]
The flow of the above-described series of audio signal processing will be described based on FIG.
In the present embodiment, in order to simplify the illustration, it is assumed that the tone
adjustment unit 4 reduces the audio signal to 3 bits without using rounding and the like.
[0040]
The 8-bit audio signal S0 transmitted between the mobile phones is parallel data of F2 (h) in
hexadecimal notation and 11110010 (b) in binary notation as shown in the figure. In the
following, for hexadecimal notation, add (h) after the numerical value, for binary notation, add (b)
after the numerical value, and for decimal notation, add (d) after the numerical value. To
distinguish. The 8-bit audio signal S0 is expanded by the expansion unit 2 to become a 14-bit
audio signal S1. The audio signal S1 is parallel data of 3280 (h) and 1100101000000 (b). The
speech signal S1 is code-converted by the code converter 3. Since the most significant bit of the
audio signal S1 of this example is 1, all the lower 13 bits except the most significant bit are
inverted to become an audio signal S2 of parallel data of 2D7F (h) and 1011010111111 (b).
[0041]
The tone adjustment unit 4 converts the 14-bit code-converted audio signal S2 into an audio
signal S3 which is parallel data of N bits. In this example, as described above, in order to simplify
the illustration, N = 3, and the upper 3 bits of the audio signal S2 are used to generate the audio
signal S3. The audio signal S3 is parallel data of 101 (b) and 5 (d).
[0042]
The audio signal S3 is converted into an audio signal S4 which is serial data in the pulse width
modulation unit 5. The 3-bit parallel data has eight gradations of 0 (d) to 7 (d) from 000 (b) to
10-04-2019
14
111 (b). As shown in FIG. 8, the audio signal S4 serially converted by pulse width modulation is
converted to H level in five out of eight divided periods in the time axis direction.
[0043]
The voice signal S4 subjected to the serial conversion is equally interpolated between the
immediately preceding data and the current data in the equally-spaced interpolation unit 6a as
the noise reduction means 6. The number of interpolation data to be provided between the
previous data and the current data, that is, the number of oversampling is calculated from the
frequency of the system, the number of gradations of the audio signal, etc. In this example, it is
assumed that the oversampling number is calculated to be 3 for the sake of simplicity. Here, the
immediately preceding data is 1 (d), and three data are interpolated in the same interval
interpolation. Since the current data is 5 (d), three data are to be interpolated between 1 and 5 in
equal proportions. When three data are interpolated, one previous data and the present data are
combined to form five data, and four intervals occur between the one previous data and the
present data. That is, (current data-1 previous data) / (interpolation number + 1) = (5-1) / 4 = 1
and three data of 1 + 1 = 2, 1 + 1 + 1 = 3, 1 + 1 + 1 + 1 = 4 are interpolated. The Rukoto. The
serial data interpolated at equal intervals is output as an audio signal S5.
[0044]
On the other hand, in order to reduce the background noise described above, interpolation
processing is also performed in the same value interpolation unit 7a of the second noise
reduction means 7. The oversampling number is the same as that of the equal interval
interpolation. In the equivalence interpolation, since one previous data is repeatedly used, three
data of 1, 1, and 1 are interpolated as shown in FIG. The serial data subjected to the equivalence
interpolation is output from the equivalence interpolation unit 7a as the audio signal S6. Then,
the addition unit 7b adds the signal obtained by inverting the H level and L level of the audio
signal S6 to the audio signal S5 interpolated at equal intervals according to the signal level, and
outputs an audio signal S7 for driving the speaker 1 . The speaker 1 reproduces the sound
satisfactorily by the sound signal S7.
[0045]
As described above, according to the present invention, a loudspeaker is driven using a digital
10-04-2019
15
signal processed by a small-scale logic circuit without requiring a D / A converter or a peripheral
component such as a capacitor corresponding thereto. Can. The logic circuit is approximately
10000 or less when converting a 2-input NAND gate as one gate using a semiconductor such as
an application specific integrated circuit (ASIC), a field programmable gate array (FPGA), or a
programmable logic device (PLD). Can be configured on a scale of Therefore, it is also possible to
incorporate these semiconductor logic circuits into a small speaker. As a result, by directly
inputting a digital signal directly to the speaker, it is possible to reproduce sound (voice, music)
well.
[0046]
Also, not limited to ASICs, FPGAs, and PLDs, dedicated small microcomputers or small DSP chips
may be used. In addition, even when using any element, each part demonstrated above shows
assignment as a function, It is not necessarily limited to what shows a physically independent
part. For example, in FIG. 1, the noise reduction means 6 and the second noise reduction means 7
are respectively provided with the same interval interpolation part 6a and the same value
interpolation part 7a. However, if the interpolation unit which integrates these is physically
provided, the noise reduction means 6 and the second noise reduction means 7 are partially used
in common. Further, the code conversion unit 3, the gradation adjustment unit 4, the pulse width
modulation unit 5, etc. may be implemented using software as well as hardware according to the
ease of configuration. That is, whether hardware or software, the above-described units and units
described above do not depend on the embodiment disclosed in the present application as long
as they share the corresponding functions.
[0047]
FIG. 2 is a block diagram showing an example of a speaker drive system according to the present
invention an explanatory diagram showing codes before and after conversion by the code
conversion unit of FIG. 1 a circuit diagram showing an example of a logic circuit performing code
conversion of FIG. 1 is a circuit diagram showing an example of a waveform diagram showing an
example of pulse width modulation by the pulse width modulation unit of FIG. 1 an explanatory
diagram showing the generation principle of aliasing noise and a reduction effect of aliasing
noise by oversampling A schematic diagram for explaining a series of signal processing using the
speaker drive system of 1.
Explanation of sign
10-04-2019
16
[0048]
DESCRIPTION OF SYMBOLS 1 Speaker 4 gradation adjustment part (gradation adjustment
means) 5 pulse width modulation part (serial conversion means) 6 noise reduction means 6a
equally spaced interpolation part 7 second noise reduction means 7a equivalent interpolation
part 7b addition part
10-04-2019
17
Документ
Категория
Без категории
Просмотров
0
Размер файла
32 Кб
Теги
description, jp2006303618
1/--страниц
Пожаловаться на содержимое документа