close

Вход

Забыли?

вход по аккаунту

?

JP2001292491

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 JP2001292491
[0001]
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an
equalizer apparatus for correcting the frequency characteristic of an audio signal.
[0002]
2. Description of the Related Art Generally, music software of various genres, such as classical
music, is used by collecting various sounds actually generated by playing various musical
instruments or singing by a singer with a microphone and recording the same. Produced. Usually,
in the process of this production, a number of sound quality corrections are performed. For
example, when classical music is considered, a certain maker performs correction to emphasize
high frequency components, and another maker performs correction to suppress low frequency
components. Also, even with the same maker, different correction processes are performed for
classical music and music for vocals.
[0003]
Since music software commercially available is subjected to such sound quality correction to a
greater or lesser extent, the user who purchases the music software and listens to the contents
may not meet his preference. In such a case, an apparatus for adjusting the frequency
characteristics to a certain extent is an equalizer apparatus, and in particular, the gain set for
08-05-2019
1
each of a plurality of frequency bands and the signal level for each frequency band of the audio
signal actually output What is visually represented using a light emitting diode or a liquid crystal
display device is a graphic equalizer.
[0004]
By the way, in the above-described conventional equalizer apparatus, the user can finely set the
gain value of each frequency band, and various representative setting patterns are registered. In
many cases, simple settings can be made by selecting an arbitrary setting pattern from among
them. However, in any case, it is necessary for the user to set each time according to his / her
preference, so when trying to listen to an audio sound that suits the user's preference for music
software that has been subjected to the various sound quality corrections described above, It is
necessary to change the setting contents of the equalizer device for each music software, and
there is a problem that the operation becomes complicated.
[0005]
For example, after setting an equalizer device that is optimal for music software produced by a
certain manufacturer and in which high frequency components are emphasized, and trying to
listen to music software produced by another manufacturer and whose low frequency
components are suppressed. In order to obtain an optimal (desired) audio sound, it is necessary
to set the equalizer device again. Therefore, when trying to listen to these music software
alternately, it is necessary to change the setting contents of the equalizer device each time. If the
number of music software to be listened to increases, the time and effort for such setting
becomes enormous.
[0006]
The present invention has been made in view of such a point, and its object is to provide an
equalizer device capable of preventing the operation from being complicated when performing
the sound quality correction according to the user's preference. It is to provide.
[0007]
SUMMARY OF THE INVENTION In order to solve the above-mentioned problems, the equalizer
08-05-2019
2
apparatus of the present invention analyzes the characteristics of the input audio signal by an
analysis means, and based on the analysis result by a filter control means. Then, the content of
the correction process by the correction filter is set.
When the music software purchased or distributed by broadcast etc. is changed, the
characteristic (processing content) of the correction filter is variably set according to the
characteristic of the audio signal input for each music software. Therefore, there is no need for
the listener (user) of this music software to change the characteristics of the control filter each
time, and it is possible to prevent the operation from becoming complicated when performing the
sound quality correction according to the preference. be able to.
[0008]
Also, while analyzing the signal level of each frequency of the audio signal by the analysis means
described above, the audio signal after passing through the correction filter is specified by the
filter control means described above based on the analysis result by the analysis means. It is
desirable to set the filter characteristics of the correction filter so as to achieve the target
characteristics. By analyzing the signal level for each frequency, it is possible to know the
tendency of the sound quality correction performed in advance at the music software creator
side etc., so how is the correction filter applied to the input audio signal? It is possible to know
whether an audio signal having a desired target characteristic can be obtained by performing a
proper correction process, so that the characteristic of the correction filter can be set
automatically according to the characteristic of the input audio signal. It will be possible.
[0009]
The analysis means described above analyzes the transfer characteristic in the process of
recording the audio signal based on the input audio signal, and the filter control means performs
the correction filter based on the analysis result by the analysis means. It is desirable to set the
filter characteristics of the correction filter so that the audio signal after passing has a
predetermined target characteristic. At the time of production of music software, various sound
quality correction processes are applied to the source signal obtained by collecting singing voice
and musical instrument sound with a microphone etc., and the transfer characteristics in the
process of recording as audio sound are analyzed. The tendency of sound quality correction
performed in advance for each music software can be determined more appropriately.
08-05-2019
3
[0010]
Also, the analysis means described above determines an autocorrelation function corresponding
to at least one of the right channel signal and the left channel signal included in the audio signal,
and the cross correlation function of the right channel signal and the left channel signal. It is
desirable to analyze the transfer characteristics by computing the impulse response of the
transfer characteristics based on the autocorrelation function and the cross correlation function
and converting the impulse response into a signal in the frequency domain. Since the
autocorrelation function and the cross correlation function can be determined relatively easily,
they can be used to calculate the impulse response of the transfer characteristic and easily
analyze the transfer characteristic by converting it into a signal in the frequency domain. it can.
[0011]
Further, while extracting a predetermined feature amount corresponding to the input audio
signal by the feature amount extraction means, the learning processing means sets a target
characteristic for each of the extracted feature amounts and newly input audio It is desirable to
update the content of the corresponding target characteristic in consideration of the analysis
result by the analysis means obtained corresponding to the signal. Based on the feature amount,
the audio signal is classified into a plurality of types (genres), and the contents of the target
characteristic set for each feature amount are updated, thereby reflecting the preference of the
user different for each type. Becomes easy.
[0012]
Moreover, it is preferable that at least one of the average value and the variance of the audio
signal input in a predetermined time is included in the above-mentioned feature amount. By
focusing on at least one of the average value and the variance, the audio signal can be easily
classified into a plurality of types.
[0013]
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An on-vehicle equalizer apparatus
08-05-2019
4
according to an embodiment of the present invention will be described below with reference to
the drawings. First Embodiment FIG. 1 is a diagram showing an entire configuration of an
equalizer device according to a first embodiment. The equalizer apparatus 100 shown in the
figure performs predetermined sound quality correction processing on the input audio signal,
and the correction filter 10, the FFT processing unit 12, the filter control unit 14, and the
average value / dispersion calculation unit 16 , And a learning processing unit 18 and an
operation unit 20.
[0014]
The correction filter 10 corrects the level (band level) for each frequency band of the input audio
signal based on the filter coefficient set by the filter control unit 14. The audio signal that has
been corrected by the correction filter 10 is amplified by the amplifier 30 and input to the
speaker 32, and an audio sound corresponding to the input audio signal is output from the
speaker 32.
[0015]
FIG. 2 is a view showing a specific configuration example of the correction filter 10. The
correction filter 10 shown in FIG. 2 is configured by cascading n band filters B-1 to B-n. Each of
the band filters B-1 to B-n includes four delay elements 51, five filter coefficient units 52, and a
mixer 53, and adjusting the gain of each filter coefficient unit 52 adjusts the frequency. The
signal level of each band can be changed.
[0016]
The FFT processing unit 12 performs FFT (Fast Fourier Transform) analysis on the input audio
signal to detect a signal level for each frequency band. The filter control unit 14 sets the filter
coefficients of the correction filter 10 in accordance with the analysis result output from the FFT
processing unit 12 and the instruction given from the learning processing unit 18. Details of the
filter coefficient setting operation by the filter control unit 14 will be described later. The average
value / dispersion calculation unit 16 calculates the average value and the dispersion as feature
amounts by performing statistical processing on the input audio signal.
08-05-2019
5
[0017]
The learning processing unit 18 outputs a band level (hereinafter referred to as a “target band
level”) which is a target when the filter control unit 14 sets the filter coefficient to the
correction filter 10, and When the audio sound output from the speaker 32 does not match the
user's preference based on the audio signal after passing through the filter 10, the learning
operation of updating the characteristics of the target band level according to the user's
preference is performed. Do. The target band levels described above are classified into
predetermined genres in consideration of the average value and the variance obtained by the
mean value / variance computing unit 16, and the contents thereof are stored in the learning
processing unit 18. Therefore, the learning processing unit 18 selects a genre based on the
average value and the variance obtained by the average value / dispersion computing unit 16,
selects a target band level corresponding to the selected genre, and selects the filter control unit
14. Output to Details of the operation of the learning processing unit 18 will be described later.
[0018]
The operation unit 20 is for the user to give various instructions to the equalizer device 100.
Further, in the present embodiment, the user inputs whether or not the audio quality of the audio
sound output from the speaker 32 as a result is automatically corrected for the audio signal by
the equalizer device 100 according to his or her preference. The operation for the operation is
also performed using the operation unit 20.
[0019]
The above-described FFT processing unit 12 corresponds to an analysis unit, the filter control
unit 14 corresponds to a filter control unit, the average value / dispersion calculation unit 16
corresponds to a feature quantity extraction unit, and the learning processing unit 18
corresponds to a learning processing unit. The equalizer apparatus 100 of the present
embodiment has such a configuration, and its operation will be described next. FIG. 3 is a flow
chart showing the operation procedure of the equalizer device 100 of the present embodiment.
When an audio signal is input to the equalizer device 100, the learning processing unit 18
classifies the audio signal into a genre based on the average value and the dispersion output from
the average value / dispersion calculating unit 16 (step 100).
08-05-2019
6
[0020]
FIG. 4 is a diagram showing an example of the case of dividing the audio signal into genres based
on the average value and the variance. As shown in the figure, for example, the genre 1 is the
case of the average value A and the variance a, and the target band level in this case is set to M1.
The genre 2 is the case of the average value A and the variance b, and the target band level is set
to M2. Similarly, the genre 3 is the case of the average value A and the variance c, and the target
band level is set to M3. The same applies to genre 4 and later.
[0021]
In addition, each average value (A, B, C) and each dispersion (a, b, c) are actually set as values
having a predetermined range. Further, these categories of genres 1, 2,... Correspond to genres
such as “classic” or “pops”, for example, but they do not necessarily have to be one to one.
In addition, the characteristics of the target band levels M1 to M9 corresponding to the
respective genres are respectively provided with standard characteristics (characteristics that
may be preferred by a large number of listeners) previously determined as initial characteristics,
The content is updated by the subsequent learning.
[0022]
Next, the learning processing unit 18 outputs a target band level corresponding to the genre to
which the input audio signal belongs to the filter control unit 14 (step 101). For example, if the
genre to which the audio signal belongs is determined to be "genre 1", the target band level M1 is
output to the filter control unit 14 based on the table shown in FIG. 4 described above.
[0023]
The filter control unit 14 to which the target band level is input detects the detected band level
detected by the FFT processing unit 12 (the band level obtained by performing the FFT operation
on the input audio signal) and the target band level. The filter coefficient of the correction filter
10 is set so that the characteristic (band level) of the audio signal output from the correction
filter 10 approaches the target band level based on the difference (step 102). ).
[0024]
08-05-2019
7
Next, the user who has listened to the audio sound output from the speaker 32 based on the
corrected audio signal inputs an instruction to the effect that the sound quality of the audio
sound is "not favorable". It is determined whether or not (step 103).
Specifically, the operation unit 20 is provided with an operation key for instructing the equalizer
device 100 of the present embodiment that the current sound quality is "unfavorable", and the
user can use this operation key. By pressing the key, it is possible to indicate that the current
sound quality is "unfavorable."
[0025]
When an instruction to the effect of "unfavorable" is issued, an affirmative determination is made
in step 103, and the learning processing unit 18 instructs the filter control unit 14 to change the
filter coefficient of the correction filter 10 by a predetermined amount. Do. The filter control unit
14 having received the instruction changes the filter coefficient of the correction filter 10 by a
predetermined amount (step 104).
[0026]
In the present embodiment, when changing the filter coefficients of the correction filter 10, the
filter control unit 14 cyclically selects a plurality of preset change patterns and changes the filter
coefficients along the change patterns. Do. For example, consider five patterns of “1: no
change”, “2: low range emphasis”, “3: low range suppression”, “4: high range emphasis”,
and “5: high range suppression” as change patterns It shall be. However, the change pattern is
not limited to this.
[0027]
In the initial state, “1: no change” is set as the change pattern, and when the user performs a
predetermined operation indicating “not desirable”, the change pattern of “2: low range
emphasis” is next It is selected. In addition, when the user performs a predetermined operation
indicating "unfavorable" again within a predetermined time (for example, several seconds), a
08-05-2019
8
change pattern of "3: low range suppression" is selected next. In this manner, when the user
performs a predetermined operation of instructing "unfavorable" many times within a
predetermined time, the above-described five change patterns are cyclically selected. Further,
when the predetermined operation described above is not performed within the predetermined
time, the selection state of the change pattern set at that time is determined, and the change
operation of the filter coefficient in step 104 described above is performed.
[0028]
Next, the learning processing unit 18 acquires the change pattern selected by the filter control
unit 14 at that time, and updates the characteristics of the target band level based on the
contents (step 105). For example, as described above, when the genre to which the current audio
signal belongs is determined as "genre 1" and the corresponding target band level M1 is used as
the target band level, the filter control unit 14 selects When the change pattern is "2: low band
suppression", in order to reflect this content, the characteristic of the target band level M1 is
updated to the target band level M1 'having the characteristic in which the low band is
emphasized. .
[0029]
Therefore, when the genre to which the audio signal belongs is determined to be "genre 1" again
based on the calculation result (average value and variance) by the average value / dispersion
calculation unit 16 in the subsequent processing, the filter control unit 14 The filter coefficient of
the correction filter 10 is set such that the band level of the audio signal output from the
correction filter 10 approaches the updated target band level M1 '. After the target band level is
changed in this way, the operations of step 100 and subsequent steps are repeated.
[0030]
In addition, when a predetermined operation has not been made by the user and the
predetermined time has elapsed, or when the user has repeatedly made the predetermined
operation a plurality of times. When the selected state is determined in the state of "1: no
change", a negative determination is made in the determination of step 103 described above, and
the operations after step 100 are repeated.
[0031]
08-05-2019
9
As described above, the equalizer apparatus 100 according to the present embodiment performs
a predetermined FFT operation on the input audio signal to analyze the frequency characteristic,
and based on the analysis result, the target whose frequency characteristic of the audio signal is
desired Since the filter coefficients of the correction filter 10 are set to have a band level, the
user of the music software controls the control filter 10 each time according to the
characteristics of the audio signal input for each music software. There is no need for an
instruction to change the characteristics, and it is possible to prevent the operation from being
complicated when performing the sound quality correction to suit the preference.
In particular, by performing a predetermined FFT operation on the input audio signal and
analyzing the frequency characteristics, it is possible to perform desired correction processing on
the input audio signal by the correction filter 10. Since it can be known whether an audio signal
having characteristics can be obtained, it becomes possible to automatically set the
characteristics of the correction filter according to the characteristics of the input audio signal.
[0032]
Further, in the present embodiment, the tendency of the sound quality preferred by the user is
learned by the learning processing unit 18 and the characteristics of the corresponding target
band level are updated as needed. Therefore, the sound quality correction corresponding to the
user's preference is properly performed. It also has the advantage that it can be done. In
particular, since the average value and the variance are calculated as feature amounts, the input
audio signal is classified, and the target band level is set for each classification, the user who is
different for each audio signal having a different type (genre) You can set the characteristics of
your choice.
[0033]
Second Embodiment By the way, in the above-described first embodiment, a music software
creator side or the like performs FFT analysis on an input audio signal to detect a signal level for
each frequency band. We have sought the tendency of sound quality correction that has been
carried out in advance, but perform predetermined arithmetic processing based on the audio
signal, and at the time of producing music software, collect sounds such as musical instrument
sounds and human singing voices using microphones etc. Various sound quality correction
08-05-2019
10
processes are performed on the source signal obtained by sounding, and the transfer
characteristic in the process of being recorded as an audio signal (hereinafter referred to as
“transfer characteristic of the recording system”) is estimated. Alternatively, the tendency of
sound quality correction performed in advance for music software may be determined.
Hereinafter, the principle of estimating the transfer characteristic of the recording system at the
time of production of music software will be described in detail.
[0034]
FIG. 5 is a diagram for explaining the relationship between input / output signals and transfer
characteristics at the time of production of music software. In the following description, source
signals are obtained by collecting voices such as musical instrument sounds and human singing
voices by using two microphones disposed on the left and right at predetermined intervals, and a
source signal is obtained. Let us consider the process of performing the sound quality correction
process of and recording the audio signal.
[0035]
In FIG. 5, x (n) represents a source signal obtained by collecting musical instrument sounds,
human singing voices and the like with a microphone, and h1 (n) and h2 (n) represent various
types of source signals. It shows the impulse response of the transfer characteristic (transfer
characteristic of the recording system) in the process of being subjected to the sound quality
correction processing and being recorded as an audio signal. Further, y1 (n) represents an audio
signal of the right channel, and y2 (n) represents an audio signal of the left channel.
[0036]
The relationship between the source signal x (n), the impulse responses h1 (n) and h2 (n), and
the audio signals y1 (n) and y2 (n) can be expressed by the following equation:
[0039]
【0039】となる。 Here, K is the number of taps of h1 (n), and J is the number of taps of h2
(n). By Fourier transforming (1) and (2),
08-05-2019
11
[0042] 【0042】となる。 Here, X (ω), H1 (ω), H2 (ω), Y1 (ω), Y2 (ω) are x (n), h1 (n), h2
(n), y1 (n), y2 (N) is each obtained by Fourier transform, and ω is an angular frequency. If the
cross spectrum of Y1 (ω) and Y2 (ω) and the power spectrum of Y1 (ω) are calculated,
[0045] 【0045】となる。 Here, PY1Y2 (ω) is a cross spectrum of Y1 (ω) and Y2 (ω),
PY1Y1 (ω) is a power spectrum of Y1 (ω), and PXX (ω) is a power spectrum of X (ω). Also, H1
* (ω) is a complex conjugate of H1 (ω). When H1 * (ω) PXX (ω) is eliminated from the
equations (5) and (6),
[0047] Is obtained. Inverse Fourier transform of equation (7)
[0049] 【0049】となる。 Here, ry1y2 (τ) is a cross correlation function of y1 and y2, ry1y1
(τ) is an autocorrelation function of y1, and τ is a time lag,
[0052] 【0052】である。 When the equation (8) mentioned above is transformed,
[0054] By assuming that h 1 (0) = 1,
[0056] 【0056】となる。 Note that assuming h1 (0) = 1, the values of h1 (n) and h2 (n)
calculated later are obtained as a ratio to h1 (0), but this is the overall level Since it only changes
uniformly, there is no problem in estimating the tendency of the frequency characteristic or the
phase characteristic of h1 (n) or h2 (n).
[0057] By rewriting the above equation (12) with matrix expression,
[0059] 【0059】となる。 ここで、
[0065] 【0065】である。 From equation (13) mentioned above,
[0067] Is obtained. Thereby, the impulse response of the transfer characteristic of the recording
08-05-2019
12
system can be estimated from the audio signal. Then, the transfer characteristic of the recording
system can be obtained by performing Fourier transform processing on the estimated impulse
response.
[0068] Next, an equalizer apparatus according to a second embodiment configured in accordance
with the above principle will be described. FIG. 6 is a diagram showing the overall configuration
of the equalizer device of the second embodiment. The equalizer device 100a shown in the figure
performs predetermined sound quality correction processing on an input audio signal, and the
correction filter 10, the filter control unit 14, the operation unit 20, the transfer characteristic
calculation unit 34, and the target band A level setting unit 42 is included. In the equalizer device
100 a of the second embodiment, the same reference numerals are given to components
overlapping with the equalizer device 100 of the first embodiment described above, and the
detailed description thereof will be omitted.
[0069] The transfer characteristic calculation unit 34 calculates the transfer characteristic of the
recording system, and includes an autocorrelation function / cross correlation function
calculation unit 36, an impulse response calculation unit 38, and an FFT processing unit 40. The
autocorrelation function / crosscorrelation function calculator 36 calculates the autocorrelation
function shown in the above equation (10) and the crosscorrelation function shown in the
equation (9) based on the input audio signal.
[0070] The impulse response calculation unit 38 calculates an impulse response based on the
above-mentioned equation (19) using the autocorrelation function and the cross correlation
function obtained by the autocorrelation function / cross correlation function calculation unit 36.
The FFT processing unit 40 performs FFT analysis on the impulse response obtained by the
impulse response calculating unit 38 to obtain the transfer characteristic of the recording
system.
[0071] The target band level setting unit 42 sets a target band level based on an operation
instruction given by the user using the operation unit 20 and outputs the target band level to the
filter control unit 14. The filter control unit 14 sets the filter coefficient of the correction filter 10
according to the calculation result (the transfer characteristic of the recording system) output
from the transfer characteristic calculation unit 34 and the target band level given by the target
band level setting unit 42.
[0072] The transfer characteristic calculation unit 34 described above corresponds to the
08-05-2019
13
analysis means, and the filter control unit 14 corresponds to the filter control means. As
described above, the equalizer device 100a of the second embodiment performs predetermined
arithmetic processing by the transfer characteristic calculation unit 34 based on the audio signal,
and estimates and obtains the transfer characteristic of the recording system at the time of
production of music software. Since the filter coefficient of the correction filter 10 is set based on
the determined transfer characteristic so that the frequency characteristic of the audio signal has
a desired target band level, the sound quality correction performed in advance for each music
software is performed. The tendency can be determined more appropriately, and sound quality
correction can be performed according to the user's preference.
[0073] The present invention is not limited to the above embodiment, and various modifications
can be made within the scope of the present invention. For example, in the first embodiment
described above, the FFT processing unit 12 performs the FFT operation to convert the audio
signal into a signal in the frequency domain, but instead of the FFT processing unit 12, m filters
are used The following filter bank may be used.
[0074] Also, although both average values and variances are used as feature quantities to be
extracted from the input audio signal, only one of them is used, or in combination with other
feature quantities, or instead of these feature quantities. The feature amount of may be used.
Further, in the equalizer apparatus 100 of the first embodiment shown in FIG. 1 described above,
the average value / dispersion calculation unit 16 is omitted and the learning processing unit 18
is set to the target band level setting unit as in the second embodiment. It may be replaced with
42.
[0075] Further, in the equalizer device 100a in the second embodiment described above, the
user has set a desired target band level using the operation unit 20, but as in the equalizer device
100 in the first embodiment, the average The audio signal is classified based on the value and the
variance, the target band level is set for each classification, the tendency of the sound quality
preferred by the user is learned, and the characteristic of the corresponding target band level is
updated as needed. Good.
[0076] FIG. 7 is a diagram showing an entire configuration of an equalizer apparatus of a
modification in which the tendency of the sound quality preferred by the user is learned and the
characteristic of the corresponding target band level is updated as needed. The equalizer device
100b shown in the figure includes a correction filter 10, a filter control unit 14, an average value
/ dispersion calculation unit 16, a learning processing unit 18, an operation unit 20, and a
transfer characteristic calculation unit 34. The equalizer device 100b of the modification has a
08-05-2019
14
configuration in which the portion of the FFT processing unit 12 in the equalizer device 100 of
the first embodiment described above is replaced with the transfer characteristic calculation unit
34 in the second embodiment, The operation content of each component is as described in the
first and second embodiments described above, and thus detailed description will be omitted.
[0077] In each of the above-described embodiments, although the on-vehicle equalizer device to
which the present invention is applied is described in detail, the present invention is not limited
to this, and it is also possible to other audio systems such as home audio systems. It can apply.
[0078] As described above, according to the present invention, the characteristics of the input
audio signal are analyzed, and the contents of the correction processing by the correction filter
are set based on the analysis result. Even when is changed, there is no need for the user of this
music software to change the characteristics of the control filter each time in accordance with
the characteristics of the audio signal input for each music software, and it was possible to meet
the preference It is possible to prevent the operation from becoming complicated when the
sound quality correction is performed.
08-05-2019
15
Документ
Категория
Без категории
Просмотров
0
Размер файла
27 Кб
Теги
jp2001292491
1/--страниц
Пожаловаться на содержимое документа