CN106297814B - Speech signal processing device and audio signal processing method - Google Patents
Speech signal processing device and audio signal processing method Download PDFInfo
- Publication number
- CN106297814B CN106297814B CN201510294353.1A CN201510294353A CN106297814B CN 106297814 B CN106297814 B CN 106297814B CN 201510294353 A CN201510294353 A CN 201510294353A CN 106297814 B CN106297814 B CN 106297814B
- Authority
- CN
- China
- Prior art keywords
- sample point
- scaled signals
- value
- original
- frame
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Landscapes
- Compression, Expansion, Code Conversion, And Decoders (AREA)
- Transmission Systems Not Characterized By The Medium Used For Transmission (AREA)
Abstract
The present invention provides a kind of speech signal processing device and audio signal processing method.Number first sample point of corresponding sample point phase matched with phase reference sample point in m-th of original down-scaled signals frame according to phase reference sample point number decision in m-1 original down-scaled signals frames and corresponding to the intermediate sample point of the m-1 update down-scaled signals frame.This first is sampled the continuous q sample point lighted as the sample point of m-th of update down-scaled signals frame, phase mismatches caused distorted signals situation when doing further frequency reducing (such as frequency is reduced to a quarter) to sampled voice signal, can still effectively improve the overlapping of signal frame.
Description
Technical field
The invention relates to a kind of signal processing apparatus, and in particular to a kind of speech signal processing device and language
Signal processing method.
Background technique
For hearing-impaired people, the voice signal of higher-frequency, such as consonant letter often can not be clearly received
Number, but the signal of low frequency can clearly be heard.The general prior art is by dropping the voice signal of high frequency
Frequently, and by signal frame it is overlapped, to solve the problems, such as this.After carrying out frequency reducing to signal, since time span is elongated, continuously
Signal value between two sampled signals need to be acquired in the way of interpolation.Since the Property comparison of voice signal is close to string
Wave, if seeking the signal value of interpolation in a manner of general arithmetic mean, there is signal mistake in the signal after often making frequency reducing
Genuine situation.In addition, the prior art can't consider whether its phase matches, therefore when carrying out the overlap action of signal frame
The situation of a part of signal addition, a part of signal subtraction to occur in overlapping, and in turn result in the distortion of signal, and with drop
The situation of the bigger distortion of the amplitude of frequency will be more serious.
Summary of the invention
The present invention provides a kind of speech signal processing device and audio signal processing method, to sampled voice signal do into
When the frequency reducing of one step, phase mismatches caused distorted signals situation when can still effectively improve the overlapping of signal frame.
Speech signal processing device of the invention includes processing unit, to frequency reducing sampled voice signal, to generate packet
The down-scaled signals of the original down-scaled signals frame of a sequence are included, and generate corresponding update down-scaled signals according to original down-scaled signals frame
Frame, wherein each original down-scaled signals frame includes p sample point.Processing unit is also according in m-1 original down-scaled signals frames
It is original at m-th that the number decision of phase reference sample point corresponding to the intermediate sample point of down-scaled signals frame is updated with the m-1
First sample point of corresponding sample point phase matched is numbered in down-scaled signals frame with phase reference sample point, will from phase
Benchmark sample point numbers first of corresponding sample point phase matched and samples the continuous q sample point lighted as m-th more
The sample point of new down-scaled signals frame, and the update down-scaled signals frame that aliasing is adjacent, to generate overlapping voice signal, wherein phase base
Quasi- sample point number is to update the intermediate sample point institute of down-scaled signals frame with the m-1 in m-1 original down-scaled signals frames
The number of corresponding sample point, p, q are positive integer, m is the positive integer greater than 1.
In one embodiment of this invention, the frequency of above-mentioned down-scaled signals is a quarter of sampled voice signal, respectively more
The length of new down-scaled signals frame is equal to the half of the length of each original down-scaled signals frame.
In one embodiment of this invention, the above-mentioned two neighboring overlay region for updating down-scaled signals frame and being respectively provided with 50%
Section.
In one embodiment of this invention, above-mentioned processing unit is also according to sample point in m-th of original down-scaled signals frame
Sampling value adds up the first count value and the second count value, wherein when count up to sample point that corresponding sample value is 0 or with sampling
Value for 0 a sample point adjacent at least sample point when, first count value that is zeroed or the second count value.Processing unit is by m-th
First count value corresponding to phase reference sample point numbering corresponding sample point in original down-scaled signals frame second counts
Value be used as a reference value, and benchmark value determine it is corresponding with phase reference sample point number in m-th of original down-scaled signals frame
First sample point of sample point phase matched.
In one embodiment of this invention, above-mentioned processing unit also judge in m-1 original down-scaled signals frames with phase
Benchmark sample point numbers whether the first count value corresponding to corresponding sample point is less than or equal to m-1 original down-scaled signals frames
In with phase reference sample point number corresponding sample point corresponding to the second count value.If m-1 original down-scaled signals frames
In with phase reference sample point number corresponding sample point corresponding to the first count value be less than or equal to m-1 original frequency reducings letters
Second count value corresponding to phase reference sample point numbering corresponding sample point in number frame, m-1 original frequency reducings are believed
First count value corresponding to phase reference sample point numbering corresponding sample point in number frame is as a reference value, and by m-th
The sample point sampled earliest in corresponding sample point when the first count value is equal to a reference value in original down-scaled signals frame is as m
First sample point of corresponding sample point phase matched is numbered in a original down-scaled signals frame with phase reference sample point, if the
First count value corresponding to phase reference sample point numbering corresponding sample point in m-1 original down-scaled signals frames no less than
It is counted equal to corresponding to numbering corresponding sample point with phase reference sample point in m-1 original down-scaled signals frames second
Value, corresponding sample point will be numbered with phase reference sample point in m-1 original down-scaled signals frames corresponding to second count
Value is used as a reference value, and when the second count value in m-th of original down-scaled signals frame is equal to a reference value in corresponding sample point
The sample point sampled earliest is as sample point phase corresponding with phase reference sample point number in m-th of original down-scaled signals frame
Matched first sample point.
In one embodiment of this invention, above-mentioned processing unit is also by down-scaled signals multiplied by a Hamming window.
In one embodiment of this invention, above-mentioned processing unit band takes according to continuous three in each original down-scaled signals frame
Sample value calculates the value of interpolated parameter function corresponding with each original down-scaled signals frame, and corresponding according to each original down-scaled signals frame
The value of interpolated parameter function calculates the interpolated value in each original down-scaled signals frame between adjacent two sample point.
In one embodiment of this invention, above-mentioned processing unit also judges whether the value of interpolated parameter function is less than upper limit value
And it is more than or equal to lower limit value, if the value of interpolated parameter function is not more than or equal to lower limit value, modified interpolation ginseng no less than upper limit value or
The value of number function, wherein the value of interpolated parameter function is modified to the upper limit if the value of interpolated parameter function is more than or equal to upper limit value
The value of interpolated parameter function is modified to lower limit value if the value of interpolated parameter function is less than lower limit value by value.
In one embodiment of this invention, above-mentioned sampled voice signal is to be generated by sampling primary speech signal, on
Limit value and lower limit value are associated with the frequency of primary speech signal and sample the sampling frequency of the primary speech signal.
In one embodiment of this invention, above-mentioned processing unit also takes according to continuous three in each original down-scaled signals frame
Trigonometric function relationship between sample value calculates the corresponding interpolated parameter function of each original down-scaled signals frame, and wherein interpolated parameter function is
Trigonometric function.
Audio signal processing method of the invention includes the following steps.Frequency reducing sampled voice signal includes a sequence to generate
The down-scaled signals of the original down-scaled signals frame of column, wherein each original down-scaled signals frame includes p sample point, wherein p is positive integer.
Phase corresponding to the intermediate sample point of down-scaled signals frame is updated with the m-1 according in m-1 original down-scaled signals frames
Benchmark sample point number determines the sample point phase corresponding with phase reference sample point number in m-th of original down-scaled signals frame
Matched first sample point, wherein m is the positive integer greater than 1, and phase reference sample point number is in m-1 original frequency reducings
The number of sample point corresponding to the intermediate sample point of down-scaled signals frame is updated in signal frame with the m-1.Will from phase base
The continuous q sample point that first sampling that quasi- sample point numbers corresponding sample point phase matched is lighted is updated as m-th
The sample point of down-scaled signals frame, wherein q is positive integer.The adjacent update down-scaled signals frame of aliasing, to generate overlapping voice signal.
In one embodiment of this invention, the frequency of above-mentioned down-scaled signals is a quarter of sampled voice signal, respectively more
The length of new down-scaled signals frame is equal to the half of the length of each original down-scaled signals frame.
In one embodiment of this invention, the above-mentioned two neighboring overlay region for updating down-scaled signals frame and being respectively provided with 50%
Section.
In one embodiment of this invention, above-mentioned foundation drops in m-1 original down-scaled signals frames with the m-1 update
Phase reference sample point corresponding to the intermediate sample point of frequency signal frame number determine in m-th of original down-scaled signals frame with phase
The step of position benchmark sample point numbers first sample point of corresponding sample point phase matched includes the following steps: according to m
The sampling value of sample point adds up the first count value and the second count value in a original down-scaled signals frame, wherein when counting up to correspondence
When sample point or an at least sample point adjacent with the sample point that sampling value is 0 that sampling value is 0, be zeroed its corresponding first meter
Numerical value or the second count value.Sample point institute corresponding with phase reference sample point number in m-th of original down-scaled signals frame is right
The first count value or the second count value answered are as a reference value.Benchmark value determine in m-th of original down-scaled signals frame with phase
Position benchmark sample point numbers first sample point of the phase matched of corresponding sample point.
In one embodiment of this invention, above-mentioned to be numbered in m-th of original down-scaled signals frame with phase reference sample point
First count value or the second count value corresponding to corresponding sample point include the following steps: judgement as the step of a reference value
Whether the first count value corresponding to phase reference sample point numbering corresponding sample point in m-1 original down-scaled signals frames is small
In be equal in m-1 original down-scaled signals frames number corresponding sample point with phase reference sample point corresponding to second count
Value.If first counts corresponding to phase reference sample point numbering corresponding sample point in m-1 original down-scaled signals frames
Value be less than or equal in m-1 original down-scaled signals frames to number corresponding sample point with phase reference sample point corresponding to second
Count value, corresponding sample point will be numbered with phase reference sample point in m-1 original down-scaled signals frames corresponding to first
Count value is as a reference value.If sample point corresponding with phase reference sample point number institute in m-1 original down-scaled signals frames
It is corresponding with phase reference sample point number in no less than equal to m-1 original down-scaled signals frames of corresponding first count value to take
Second count value corresponding to sampling point takes corresponding with phase reference sample point number in m-1 original down-scaled signals frames
The second count value is as a reference value corresponding to sampling point.
In one embodiment of this invention, if in m-1 original down-scaled signals frames with phase reference sample point number pair
First count value corresponding to the sample point answered is less than or equal to compile in m-1 original down-scaled signals frames with phase reference sample point
Second count value corresponding to number corresponding sample point, above-mentioned audio signal processing method includes: by m-th of original down-scaled signals
The sample point sampled earliest in corresponding sample point when the first count value is equal to a reference value in frame is as m-th of original frequency reducing letter
First sample point of corresponding sample point phase matched is numbered in number frame with phase reference sample point.
In one embodiment of this invention, if it is corresponding with phase reference sample point number in m-th of original down-scaled signals frame
Sample point corresponding to the first count value no less than being equal in m-th original down-scaled signals frame and phase reference sample point is numbered
Second count value corresponding to corresponding sample point, above-mentioned audio signal processing method include: by m-th of original down-scaled signals frame
In the sample point that is sampled earliest in corresponding sample point when being equal to a reference value of the second count value as m-th of original down-scaled signals
First sample point of corresponding sample point phase matched is numbered in frame with phase reference sample point.
In one embodiment of this invention, above-mentioned audio signal processing method includes, by down-scaled signals multiplied by Hamming window.
In one embodiment of this invention, above-mentioned audio signal processing method includes the following steps: according to each original frequency reducing
Continuous three sampling values calculate the value of interpolated parameter function corresponding with each original down-scaled signals frame in signal frame；Judge interpolation
Whether the value of parametric function is less than upper limit value and is more than or equal to lower limit value, if the value of interpolated parameter function is no less than upper limit value or not
More than or equal to lower limit value, the value of modified interpolation parametric function；Value meter according to the corresponding interpolated parameter function of each down-scaled signals frame
Calculate the interpolated value in each down-scaled signals frame between adjacent two sample point.
In one embodiment of this invention, wherein if the value of interpolated parameter function is more than or equal to upper limit value, by interpolated parameter
The value of function is modified to upper limit value, if the value of interpolated parameter function is less than lower limit value, the value of interpolated parameter function is modified to down
Limit value, wherein sampled voice signal is to be generated by sampling primary speech signal, and upper limit value and lower limit value are associated with original language
The frequency of sound signal and the sampling frequency of sampling primary speech signal.
In one embodiment of this invention, above-mentioned audio signal processing method includes, according in each original down-scaled signals frame
Trigonometric function relationship between continuous three sampling values calculates the corresponding interpolated parameter function of each original down-scaled signals frame, wherein in
Inserting parametric function is trigonometric function.
Based on above-mentioned, the embodiment of the present invention according in m-1 original down-scaled signals frames with the m-1 update frequency reducing
Phase reference sample point corresponding to the intermediate sample point of signal frame number determine in m-th of original down-scaled signals frame with phase
Benchmark sample point numbers first sample point of corresponding sample point phase matched, will be certainly corresponding with phase reference sample point number
First of sample point phase matched sample sampling of the continuous q sample point lighted as m-th of update down-scaled signals frame
Point, can still effectively improve when doing further frequency reducing (such as frequency is reduced to a quarter) to sampled voice signal
Phase mismatches caused distorted signals situation when signal frame is overlapped.
Detailed description of the invention
Fig. 1 is shown as the structural schematic diagram of the speech signal processing device of one embodiment of the invention；
Fig. 2 shows the signal processing schematic diagrames of the sampled voice signal for one embodiment of the invention；
Fig. 3 is shown as the schematic diagram of the down-scaled signals of one embodiment of the invention；
Fig. 4 is shown as the schematic diagram of the original down-scaled signals frame WL3 of one embodiment of the invention；
Fig. 5 is shown as the flow diagram of the audio signal processing method of one embodiment of the invention；
Fig. 6 is shown as the flow diagram of the audio signal processing method of another embodiment of the present invention；
Fig. 7 is shown as the flow diagram of the audio signal processing method of another embodiment of the present invention.
Description of symbols:
102: processing unit；
104: sampling unit；
S1: primary speech signal；
S2: sampled voice signal；
W1~W4: sampled signal frame；
WL1~WL4: original down-scaled signals frame；
WL1 '~WL4 ': down-scaled signals frame is updated；
WH1~WH4: multiplied by the update down-scaled signals frame after Hamming window；
SL, SL ', SH: down-scaled signals；
sm(4n)、sm(4n+4)、sm(4n+8): sample point；
sm(4n+1)、sm(4n+2)、sm(4n+3)、sm(4n+5)、sm(4n+6)、sm(4n+7): interpolated point；
SO: overlapping voice signal；
First count value；
N: sample point number；
Wm: m-th original down-scaled signals frame；
S502~S510, S602~S608, S702~S712: the step of audio signal processing method.
Specific embodiment
Fig. 1 is shown as the structural schematic diagram of the speech signal processing device of one embodiment of the invention, please refers to Fig. 1.Voice
Signal processing apparatus includes processing unit 102 and sampling unit 104, and processing unit 102 couples sampling unit 104, wherein locating
Reason unit 102 can for example be implemented with central processing unit, and sampling unit 104 then can for example be implemented with logic circuit, so
It is not limited.104 sampling primary speech signal S1 of sampling unit, to generate sampled voice signal S2.Processing unit 102 can
Frequency reducing sampled voice signal S2 to generate the down-scaled signals of the down-scaled signals frame including a sequence, Fig. 2 shows for the present invention one it is real
The signal processing schematic diagram of the sampled voice signal of example is applied, as shown in Fig. 2, sampled voice signal S2 may include the sampling of a sequence
Signal frame only shows 4 sampled signal frame W1~W4 in Fig. 2 embodiment, is so not limited thereto to simplify explanation.Frequency reducing
Signal SL includes multiple original down-scaled signals frame WL1~WL4, since down-scaled signals SL is obtained by frequency reducing sampled voice signal S2
It arrives, therefore the length of sampled signal frame of the length of original down-scaled signals frame greater than sampled voice signal S2, in the present embodiment
A quarter that the frequency of down-scaled signals SL is sampled voice signal S2 (therefore the length of each original down-scaled signals frame is right for its
4 times of the length for the sampled signal frame answered), so it is not limited.
Processing unit 102 can obtain updating down-scaled signals frame from the sample point of selected part in original down-scaled signals frame
(such as update down-scaled signals frame WL1 '~WL4 ' of Fig. 2, in the present embodiment, each length for updating down-scaled signals frame is equal to each
The half of the length of original down-scaled signals frame), and make it is each update down-scaled signals frame intermediate sample point with it is next more
The phase matched of the initial sample point of new down-scaled signals frame, and then improve phase when signal frame is overlapped and mismatch caused signal
It is distorted situation.
Specifically, the sample point of part can be obtained for example by executing interpolative operation in original down-scaled signals frame.Place
It manages unit 102 and can first be calculated according to continuous three sampling values known in each original down-scaled signals frame and believed with each original frequency reducing
The value of number corresponding interpolated parameter function of frame, and come according to the value of interpolated parameter function corresponding to each original down-scaled signals frame
The interpolated value in each original down-scaled signals frame between sample point known to adjacent two is calculated, wherein interpolated parameter function is triangle letter
Number, such as SIN function or cosine function, are so not limited.
For example, Fig. 3 is shown as the schematic diagram of the down-scaled signals of one embodiment of the invention, referring to figure 3..In Fig. 3,
The part of black circle is known sample point in original down-scaled signals frame, and the part of soft dot is 102 foundation of processing unit
Known sample point carries out the calculated interpolated point of interpolative operation institute, and the part of square dot is processing unit 102 according to known
Sample point and calculated interpolated point carry out the calculated interpolated point of interpolative operation institute again.Processing unit 102 can according to
The sampling value of continuous three known sample points calculates corresponding with each original down-scaled signals frame in each original down-scaled signals frame
Interpolated parameter function, for example, interpolated parameter function C corresponding to m-th of original down-scaled signals frame WmmIt (g) can be according in original
Three sample point s of continuous sampling in beginning down-scaled signals framem(4n)、sm(4n+4) and smTrigonometric function between (4n+8) closes
To acquire, the interpolated parameter function in the time range of original down-scaled signals frame Wm corresponding to it can be shown below for system:
Wherein g is 0 or positive integer, CmIt (g) is functional value of the interpolated parameter function in time point g, interpolated parameter function Cm
It (g) is a trigonometric function.
It generates, and causes calculated since speech signal processing device might have noise during signal processing
The value of interpolated parameter function includes the ingredient of noise, so will affect the accuracy that processing unit 102 seeks interpolated value.Processing is single
Member 102 whether can be fallen in a preset range by judging the value of interpolated parameter function monitor interpolated parameter function value whether
By noise jamming, such as it can determine whether the value of interpolated parameter function is less than upper limit value and is more than or equal to lower limit value, if interpolation
The value of parametric function is not more than or equal to lower limit value no less than upper limit value or, then the value of representation parameter function is by noise jamming, place
Manage unit 102 can modified interpolation parametric function value, to remove noise contribution included in the value of interpolated parameter function.Example
Such as, if the value of interpolated parameter function is more than or equal to upper limit value, the value of interpolated parameter function can be modified to the upper limit by processing unit 102
Value, if the value of interpolated parameter function is less than lower limit value, the value of interpolated parameter function can be modified to lower limit value by processing unit 102,
And if the value of interpolated parameter function is less than upper limit value and is more than or equal to lower limit value, is not necessary to repair the value of interpolation parametric function
Just.For example, in the fig. 3 embodiment, interpolated parameter function Cm(g) correcting mode of value can be with following formula subrepresentation:
Namely above-mentioned upper limit value and lower limit value is respectively 1 and 0.5 in the embodiments of figure 3, if Speech processing fills
Influence during signal processing by noise is set, and makes interpolated parameter function Cm(g) value is more than or equal to 1, then locates
Unit 102 is managed by interpolated parameter function Cm(g) value is modified to 1, and if interpolated parameter function Cm(g) value is then located less than 0.5
Unit 102 is managed by interpolated parameter function Cm(g) value is modified to 0.5.It is worth noting that, the upper limit value and lower limit value of formula (3)
Only exemplary embodiment is not limited thereto.Wherein upper limit value and the situation of lower limit value visual actual noise interference are adjusted
It is whole, such as upper limit value and lower limit value can be adjusted according to the frequency of primary speech signal and the sampling frequency of sampling unit.
After the value for obtaining interpolated parameter function, processing unit 102 can calculate original drop according to interpolated parameter function
Interpolated value in frequency signal frame between adjacent two sample point.By taking the embodiment of Fig. 3 as an example, between taking in original down-scaled signals frame Wm
Sampling point sm(4n)、smInterpolated point s between (4n+4)m(4n+2) and between sample point sm(4n+4)、smIt is interior between (4n+8)
Insert point sm(4n+6) can distinguish shown in following formula:
N is 0 or positive even numbers in formula (3), formula (4).
Similarly, in Fig. 3 square dot part can also the interpolative operation mode of soft dot obtain.For example, processing is single
Member 102 can be according to the sample point s in original down-scaled signals framem(4n), interpolated point sm(4n+2) and sample point sm(4n+4) it
Between trigonometric function relationship acquire interpolated parameter function C'm(n), its institute is right in the time range of original down-scaled signals frame Wm
The interpolated parameter function C' answeredm(n) it can be shown below:
Wherein n is 0 or positive even numbers, interpolated parameter function C'm(n) correcting mode of value can be with following formula subrepresentation:
Between sample point s in original down-scaled signals frame Wmm(4n) and interpolated point smInterpolated point s between (4n+2)m(4n+
1) and between with interpolated point sm(4n+2) and sample point smInterpolated point s between (4n+4)m(4n+3) can distinguish following formula institute
Show:
In addition, processing unit 102 can be according to the sample point s in original down-scaled signals framem(4n+4), interpolated point sm(4n+
And sample point s 6)mTrigonometric function relationship between (4n+8) acquires interpolated parameter function C "m(n), in original down-scaled signals
Interpolated parameter function C " in the time range of frame Wm corresponding to itm(n) it can be shown below:
Wherein n is 0 or positive even numbers, in addition, interpolated parameter function C "m(n) correcting mode of value can be with following formula sublist
Show:
Between sample point s in original down-scaled signals frame Wmm(4n+4), interpolated point smInterpolated point s between (4n+6)m(4n
+ 5) and between interpolated point sm(4n+6), sample point smInterpolated point s between (4n+8)m(4n+7) can distinguish following formula institute
Show:
The rest may be inferred, the interpolated value in other original down-scaled signals frames between sample point or the interpolation between sample point and interpolated point
Value can also acquire in an identical manner, and those skilled in the art should can push away to obtain its embodiment party according to the teaching of above-described embodiment
Formula, thus details are not described herein.
As described above, the present embodiment is interpolated value (or sample point and the interpolation estimated between sample point using trigonometric function
Interpolated value between point), according to interpolated parameter function come calculate in original down-scaled signals frame between adjacent two sample point interpolated value (or
The interpolated value of adjacent sample point and interpolated point), using the sampling as the new sample point between sample point known in down-scaled signals
Value.Since the characteristic of trigonometric function and the characteristic of voice signal are more similar, arithmetic is merely utilized compared to the prior art
Average seeks interpolated value, and the calculation of the present embodiment can get more accurate interpolated value, and can be effectively prevented from frequency reducing
There is the situation of distorted signals in voice signal afterwards.
In addition, above-mentioned each original down-scaled signals frame may include that (wherein p is positive integer to p sample point, in the present embodiment
P can be equal to 4N-3, and N is the positive integer greater than 1), processing unit 102 can will in m-1 original down-scaled signals frames with m-1
Sample point number corresponding to a intermediate sample point for updating down-scaled signals frame is numbered as phase reference sample point, and according to phase
Position benchmark sample point number determines the sample point corresponding with this phase reference sample point number in m-th of original down-scaled signals frame
First sample point of phase matched, and the continuous q sample point lighted is sampled as m-th of update frequency reducing using from this first
The sample point (wherein q is positive integer, and q can be equal to 2N-1 in the present embodiment, and N is the positive integer greater than 1) of signal frame, so that the
The phase matched of the initial sample point of the m-1 intermediate sample point for updating down-scaled signals frame and m-th of update down-scaled signals frame,
Middle m is the positive integer greater than 1.In this way, which the m-1 update down-scaled signals frame and m-th of update down-scaled signals frame carry out
When 50% signal frame aliasing (namely the m-1 update down-scaled signals frame has respectively with m-th of update down-scaled signals frame and has
50% overlap section), the unmatched situation of phase can be reduced significantly, and improve the situation of distorted signals.
Specifically, processing unit 102 can be according to the sampling value of sample point in m-th of original down-scaled signals frame accumulative first
Count value and the second count value, wherein when processing unit 102 counts up to the sample point that corresponding sample value is 0 or with sampling value is
When the adjacent at least sample point of 0 sample point (such as adjacent forward and backward sample point, be so not limited), zero pair
The first count value or the second count value answered.Specifically, the bulk billing system of above-mentioned count value can according to following formula (13)~
(16) it indicates:
Wherein m is the positive integer greater than 1, and n=0,1,2 ..., 4N-4, N is the positive integer greater than 1, sm(n) former for m-th
The sampling value of the sample point of number n, PN in beginning down-scaled signals framemIt (n) is by sampling value sm(n) switch to " 10 ", " 3 " or " 0 " table
The value shown, wherein PNm(- 1)=PNm(0)。Corresponding to sample point for number n in m-th of original down-scaled signals frame
The first count value, andSecond counts corresponding to sample point for number n in m-th of original down-scaled signals frame
Value, whereinAndFrom formula (1), (2)For correspondence
Stored count value of the down-scaled signals in positive half cycle, andFor stored count of the corresponding down-scaled signals in negative half period
Value.As shown in formula (1)~(4), in the present embodiment, by sampling value sm(n) it is greater than 0, sm(n) it is equal to 0 and sm(n) less than 0 when
Sampling value be set to 10,3,0, count the first count valueWhen handleIt is corresponding when equal to 10 or 7
In addition first counting value returns and are counting the second count valueWhen handleIt is corresponding when equal to -10 or -3
The second counting value returns.Due to sampling value sm(n) sampling value when being equal to 0 is set as 3, thereforeEqual to 10,7 ,-
The position of 10 or -3 equal numerical value will appear in and sampling value sm(n) it is equal to the adjacent sample point position of sample point corresponding when 0.
Processing unit 102 can by m-th of original down-scaled signals frame in m-1 original down-scaled signals frames obtained by
Phase reference sample point number the first count value or the second count value corresponding to corresponding sample point (it be processing unit
102 are counted to get in m-1 original down-scaled signals frames, and counting mode and above-mentioned processing unit 102 are original at m-th
The mode counted in down-scaled signals frame is identical) it is used as a reference value, and determined in m-th of original down-scaled signals frame according to this reference value
First sample point of corresponding sample point phase matched is numbered with phase reference sample point.Such as processing unit 102 can determine whether
Whether the first count value corresponding to phase reference sample point numbering corresponding sample point in m-1 original down-scaled signals frames
Less than or equal to the second meter corresponding to numbering corresponding sample point with phase reference sample point in m-1 original down-scaled signals frames
Numerical value can be indicated with following formula (17):
WhereinFor sample point corresponding with phase reference sample point number in m-1 original down-scaled signals frames
The first corresponding count value, andIt is corresponding with phase reference sample point number in m-1 original down-scaled signals frames
Sample point corresponding to the second count value.
If the first meter corresponding to phase reference sample point numbering corresponding sample point in m-th of original down-scaled signals frame
Numerical value be less than or equal in m-th original down-scaled signals frame to number corresponding sample point with phase reference sample point corresponding to second
Count value, processing unit 102 is by sample point corresponding with phase reference sample point number institute in m-1 original down-scaled signals frames
Corresponding first count value as a reference value, and by the first count value in m-th of original down-scaled signals frame be equal to this reference value when
The sample point sampled at first in corresponding sample point can be indicated as first sample point with following formula (18), (19):
As formula (18), (19) it is found that working as the first meter corresponding to the sample point of number n in m-th of original down-scaled signals frame
Numerical value be equal in m-1 original down-scaled signals frames number corresponding sample point with phase reference sample point corresponding to the first meter
When numerical value,Equal to number n corresponding to sample point, otherwiseEqual to 4N-4.AndThen in institute
HaveIn minimum value, represent it is corresponding with phase reference sample point number in m-th of original down-scaled signals frame
The number of first sample point of the phase matched of sample point, this sample point is to update the first of down-scaled signals frame as m-th
Beginning sample point.
On the contrary, if sample point institute corresponding with phase reference sample point number is right in m-1 original down-scaled signals frames
Sampling corresponding with phase reference sample point number in no less than equal to m-1 original down-scaled signals frames of the first count value answered
Corresponding the second count value (i.e. formula (17) is invalid) of point, then processing unit 102 by m-1 original down-scaled signals frames with
Phase reference sample point numbers the second count value corresponding to corresponding sample point as a reference value, and by m-th of original frequency reducing
The sample point sampled at first in corresponding sample point when the second count value is equal to a reference value in signal frame is sampled as first
Point can be indicated with following formula (20), (21):
As formula (20), (21) it is found that working as the second meter corresponding to the sample point of number n in m-th of original down-scaled signals frame
Numerical value be equal in m-1 original down-scaled signals frames number corresponding sample point with phase reference sample point corresponding to the second meter
When numerical value,Equal to number n corresponding to sample point, otherwiseEqual to 4N-4.AndThen it is
AllIn minimum value, represent in m-th of original down-scaled signals frame with phase reference sample point number pair
The number of first sample point of the phase matched for the sample point answered, this sample point is to as m-th of update down-scaled signals frame
Initial sample point.
As an example it is assumed that each original down-scaled signals frame WL1~WL4 in Fig. 2 respectively includes 401 sample points,
0,1,2 is respectively included in i.e. each original down-scaled signals frame WL1~WL4 ..., 400 equal 401 sample points.Believe in original frequency reducing
In number frame WL2 with update down-scaled signals frame WL2 ' intermediate sample point corresponding to phase reference sample point number (it is 188)
The first corresponding count valueLess than or equal in original down-scaled signals frame WL2 with update down-scaled signals frame WL2 '
Intermediate sample point corresponding to corresponding second count value of phase reference sample point number sample pointAnd in original
The intermediate sample point (namely sample point that number is 188 in original down-scaled signals frame WL2) of beginning down-scaled signals frame WL2 is corresponding
The first count valueIt is 18.
To find out the initial sample point for updating down-scaled signals frame WL3 ', processing unit 102 is countable in original down-scaled signals
First count value in frame WL3The number of corresponding sample point is (due in original down-scaled signals frame WL2 when equal to 18
First count value corresponding to the sample point that middle number is 188Less than corresponding second count valueTherefore with the first count valueAs a reference value).Fig. 4 is shown as the original of one embodiment of the invention
The schematic diagram of beginning down-scaled signals frame WL3, as shown in figure 4, in the fig. 4 embodiment, first counts in original down-scaled signals frame WL3
ValueThe number of corresponding sample point is (namely not equal to 0 when equal to 18Value) include number 20,
40,63,79 ..., 300,325,342,363,392 etc. sample point, wherein the sample point of number 20 is in original down-scaled signals
First count value in frame WL3It is corresponding when equal to a reference value (its value be 18) in original down-scaled signals frame WL2
Sample point in the sample point that samples earliest, thereforeEqual to 20, processing unit 102 is as update down-scaled signals frame
The initial sample point of WL3 ', and continuous 201 sample points lighted from the sampling of number 20 in original down-scaled signals frame WL3 are made
For the sample point for updating down-scaled signals frame WL3 '.As shown in Fig. 2, updating down-scaled signals frame WL3 ' includes original down-scaled signals frame
The sample point of number 20~220 in WL3, wherein (it is to update the intermediate sample point of down-scaled signals frame WL3 ' original to number 120
Corresponding sample point number in down-scaled signals frame WL3) it can be used as phase reference sample point number, to be updated as searching
The foundation of the initial sample point of down-scaled signals frame WL4 '.Similarly, the initial sample point for updating down-scaled signals frame WL4 ' can also be with
Similar mode obtains, therefore details are not described herein.
Significantly, since original down-scaled signals frame WL1 is first original down-scaled signals frame, therefore update frequency reducing
The sample point of signal frame WL1 ' can be 201 continuous sampling points arbitrarily selecting from original down-scaled signals frame WL1 (in this implementation
It is the sample point of number 100~300 in example), and according in original down-scaled signals frame WL1 and in update down-scaled signals frame WL1 '
Between number corresponding to sample point as phase reference sample point number (being in the present embodiment the sample point of number 200).?
In the present embodiment, with the first of the phase matched of the intermediate sample point of original down-scaled signals frame WL1 in original down-scaled signals frame WL2
Number corresponding to a sample point is 188, wherein first sample point (sample point of number 188) seeks mode and above-mentioned reality
It is similar to apply example, those skilled in the art should can push away to obtain embodiments thereof according to above content, therefore details are not described herein.
After obtaining updating down-scaled signals frame, processing unit 102 can carry out 50% to adjacent update down-scaled signals frame
Aliasing, to generate overlapping voice signal, due at this time it is each update down-scaled signals frame intermediate sample point and next update
The phase matched of the initial sample point of down-scaled signals frame, therefore phase mismatches caused distorted signals feelings when the overlapping of signal frame
Shape will be significantly enhanced.In addition, in some embodiments, the corresponding update drop of each original down-scaled signals frame can also be being obtained
After frequency signal frame, by down-scaled signals multiplied by Hamming window (Hamming Window), down-scaled signals frame or so end is updated to increase
Continuity.As shown in Fig. 2, after it will include updating down-scaled signals frame WL1 '~WL4 ' down-scaled signals SL ' multiplied by Hamming window, it can
It obtains updating the down-scaled signals SH including down-scaled signals frame WH1~WH4, then will update down-scaled signals frame WH1~WH4 again and carry out
Overlapping voice signal SO can be obtained in aliasing.
Fig. 5 is shown as the flow diagram of the audio signal processing method of one embodiment of the invention, referring to figure 5..By upper
Embodiment is stated it is found that the audio signal processing method of speech signal processing device may include the following steps.Firstly, sampling original language
Sound signal, to generate sampled voice signal (step S502).Then, frequency reducing sampled voice signal includes a sequence to generate
The down-scaled signals (step S504) of original down-scaled signals frame, wherein the frequency of down-scaled signals may be, for example, the four of sampled voice signal
/ mono-.Wherein, the sample point of part can be obtained by interpolative operation in down-scaled signals.Fig. 6 is shown as another implementation of the present invention
The flow diagram of the audio signal processing method of example, as shown in fig. 6, as can be seen from the above embodiments, speech signal processing device
The method for calculating interpolated point may include the following steps.Firstly, according to continuous three sampling value meters in each original down-scaled signals frame
The value (step S602) of the corresponding interpolated parameter function of each original down-scaled signals frame is calculated, wherein interpolated parameter function can be according to each original
Trigonometric function relationship in beginning down-scaled signals frame between continuous three sampling values is calculated and is obtained, and interpolated parameter function can be triangle letter
Number.Later, it can then judge whether the value of interpolated parameter function is less than upper limit value and is more than or equal to lower limit value (step S604), if
The value of interpolated parameter function is not more than or equal to lower limit value no less than upper limit value or, then value (the step of modified interpolation parametric function
S606), to remove unnecessary noise.Wherein upper limit value and the situation of lower limit value visual actual noise interference adjust, such as
Upper limit value and lower limit value can be adjusted according to the frequency of primary speech signal and the sampling frequency of sampling unit, and interpolated parameter letter
The correcting mode of several values may be, for example, when the value of interpolated parameter function is more than or equal to upper limit value, by interpolated parameter function
Value is modified to upper limit value, and when the value of interpolated parameter function is less than lower limit value, the value of interpolated parameter function is modified to lower limit value.
After the value for having corrected interpolated parameter function, value meter that can then according to the corresponding interpolated parameter function of each original down-scaled signals frame
Calculate the interpolated value (step S608) in each original down-scaled signals frame between adjacent two sample point.On the contrary, if interpolated parameter function
Value is less than upper limit value and is more than or equal to lower limit value, then is directly entered step S608, calculates adjacent two in each original down-scaled signals frame
Interpolated value between sample point.
It, can be according to a more with m-1 in m-1 original down-scaled signals frames after step S504 referring once again to Fig. 5
Phase reference sample point number corresponding to the intermediate sample point of new down-scaled signals frame determines in m-th of original down-scaled signals frame
First sample point (step S506) of corresponding sample point phase matched is numbered with phase reference sample point, wherein each update drop
The length of frequency signal frame is equal to the half of the length of each original down-scaled signals frame, and phase reference sample point number is in m-
The number of sample point corresponding to the intermediate sample point of down-scaled signals frame, m are updated in 1 original down-scaled signals frame with the m-1
For the positive integer greater than 1.Later, first of corresponding sample point phase matched sampling will be numbered from phase reference sample point
The continuous q sample point lighted updates the sample point (step S508) of down-scaled signals frame as m-th, and wherein q is positive integer.Most
Afterwards, then the adjacent update down-scaled signals frame of aliasing, to generate overlapping voice signal (step S510), wherein two neighboring updates drop
Frequency signal frame can for example be respectively provided with 50% overlap section.
Fig. 7 is shown as the flow diagram of the audio signal processing method of another embodiment of the present invention, please refers to Fig. 7.In detail
For thin, the step S506 of Fig. 5 embodiment may include step S702~S706 in the present embodiment, namely first former according to m-th
The sampling value of sample point adds up the first count value and the second count value in beginning down-scaled signals frame, wherein when counting up to corresponding sample
When sample point or an at least sample point adjacent with the sample point that sampling value is 0 that value is 0, be zeroed its corresponding first count value
Or second count value (step S702), it then will be corresponding with phase reference sample point number in m-th of original down-scaled signals frame
First count value or the second count value corresponding to sample point are as a reference value (step S704), and benchmark value determines again later
First sample point of the phase matched of corresponding sample point is numbered in m-th of original down-scaled signals frame with phase reference sample point
(step S706).More specifically, step S704 may include first judge in m-1 original down-scaled signals frames with phase base
Quasi- sample point numbers whether the first count value corresponding to corresponding sample point is less than or equal in m-1 original down-scaled signals frames
Second count value (step S708) corresponding to numbering corresponding sample point with phase reference sample point.If m-1 original drops
First count value corresponding to phase reference sample point numbering corresponding sample point in frequency signal frame is less than or equal to the m-1 original
Second count value corresponding to phase reference sample point numbering corresponding sample point in beginning down-scaled signals frame, by the m-1 original
First count value corresponding to phase reference sample point numbering corresponding sample point in beginning down-scaled signals frame is as a reference value (step
Rapid S710), in this case, when the first count value in m-th of original down-scaled signals frame can be equal to a reference value by step S706
The sample point sampled earliest in corresponding sample point is numbered as in m-th of original down-scaled signals frame with phase reference sample point
First sample point of corresponding sample point phase matched.On the contrary, if in m-1 original down-scaled signals frames with phase reference
Sample point is numbered in no less than equal to m-1 original down-scaled signals frames of the first count value corresponding to corresponding sample point and phase
Position benchmark sample point numbers the second count value corresponding to corresponding sample point, by m-1 original down-scaled signals frames with phase
Position benchmark sample point numbers corresponding to corresponding sample point that the second count value is as a reference value (step S712), in this respect
Under, when the second count value in m-th of original down-scaled signals frame can be equal to a reference value by step S706 in corresponding sample point
The sample point sampled earliest is as sample point phase corresponding with phase reference sample point number in m-th of original down-scaled signals frame
Matched first sample point.
In conclusion the embodiment of the present invention according in m-1 original down-scaled signals frames with the m-1 update frequency reducing
Phase reference sample point corresponding to the intermediate sample point of signal frame number determine in m-th of original down-scaled signals frame with phase
Benchmark sample point numbers first sample point of corresponding sample point phase matched, will be from corresponding with phase reference sample point number
First of sample point phase matched sample sampling of the continuous q sample point lighted as m-th of update down-scaled signals frame
Point, can still effectively improve when doing further frequency reducing (such as frequency is reduced to a quarter) to sampled voice signal
Phase mismatches caused distorted signals situation when signal frame is overlapped.
Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention., rather than its limitations；To the greatest extent
Pipe present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that: its according to
So be possible to modify the technical solutions described in the foregoing embodiments, or to some or all of the technical features into
Row equivalent replacement；And these are modified or replaceed, various embodiments of the present invention technology that it does not separate the essence of the corresponding technical solution
The range of scheme.
Claims (17)
1. a kind of speech signal processing device characterized by comprising
One processing unit, one sampled voice signal of frequency reducing, to generate the down-scaled signals of the original down-scaled signals frame including a sequence,
And corresponding update down-scaled signals frame is generated according to those original down-scaled signals frames, wherein each original down-scaled signals frame includes
P sample point, the processing unit update down-scaled signals frame with the m-1 according in m-1 original down-scaled signals frames
A phase reference sample point corresponding to intermediate sample point number determine in m-th of original down-scaled signals frame with the phase base
Quasi- sample point numbers first sample point of corresponding sample point phase matched, will from the phase reference sample point number pair
Described first of the sample point phase matched answered samples the continuous q sample point lighted as m-th of update down-scaled signals frame
Sample point, the adjacent update down-scaled signals frame of aliasing, to generate an overlapping voice signal, wherein the phase reference sample point
Number is to update corresponding to the intermediate sample point of down-scaled signals frame in the m-1 original down-scaled signals frames with the m-1
Sample point number, p, q are positive integer, m is positive integer greater than 1, wherein the sampled voice signal is to pass through sampling one
Primary speech signal and generate.
2. speech signal processing device according to claim 1, which is characterized in that the frequency of the down-scaled signals is described
The a quarter of sampled voice signal, each length for updating down-scaled signals frame are equal to the length of each original down-scaled signals frame
The half of degree.
3. speech signal processing device according to claim 1, which is characterized in that the two neighboring update down-scaled signals
Frame is respectively provided with 50% overlap section.
4. speech signal processing device according to claim 3, which is characterized in that the processing unit is also according to described
The sampling value of sample point adds up one first count value and one second count value in m original down-scaled signals frames, wherein working as counting
To corresponding sample value be 0 sample point or an at least sample point adjacent with the sample point that sampling value is 0 when, being zeroed, its is corresponding
First count value or second count value, by described m-th original down-scaled signals frame with the phase reference sample point
First count value or second count value corresponding to corresponding sample point are numbered as an a reference value, and according to described
A reference value determines sample point phase corresponding with the phase reference sample point number in m-th of original down-scaled signals frame
First sample point matched.
5. speech signal processing device according to claim 4, which is characterized in that the processing unit also judges described
First counting corresponding to corresponding sample point is numbered in m-1 original down-scaled signals frames with the phase reference sample point
Whether value is less than or equal to sample point corresponding with the phase reference sample point number in the m-1 original down-scaled signals frames
Corresponding second count value, if being numbered in the m-1 original down-scaled signals frames with the phase reference sample point
First count value corresponding to corresponding sample point is less than or equal in the m-1 original down-scaled signals frames and the phase
Position benchmark sample point numbers second count value corresponding to corresponding sample point, by the m-1 original down-scaled signals
First count value corresponding to corresponding sample point is numbered with the phase reference sample point in frame as a reference value,
And when the first count value described in described m-th original down-scaled signals frame is equal to a reference value in corresponding sample point most
The sample point early sampled is as sampling corresponding with the phase reference sample point number in described m-th original down-scaled signals frame
First sample point of point phase matched, if being sampled in the m-1 original down-scaled signals frames with the phase reference
Point is numbered in no less than equal to the m-1 original down-scaled signals frames of first count value corresponding to corresponding sample point
Second count value corresponding to corresponding sample point is numbered with the phase reference sample point, it is original by the m-1
It is numbered described in the conduct of second count value corresponding to corresponding sample point in down-scaled signals frame with the phase reference sample point
A reference value, and when the second count value described in described m-th original down-scaled signals frame is equal to a reference value, corresponding takes
The sample point sampled earliest in sampling point as in described m-th original down-scaled signals frame with the phase reference sample point number pair
First sample point for the sample point phase matched answered.
6. speech signal processing device according to claim 1, which is characterized in that the processing unit is also by the frequency reducing
Signal is multiplied by a Hamming window.
7. speech signal processing device according to claim 1, which is characterized in that the processing unit is also according to each described
Continuous three sampling values calculate interpolated parameter function corresponding with each original down-scaled signals frame in original down-scaled signals frame
Value, and the value according to the corresponding interpolated parameter function of each original down-scaled signals frame calculates each original down-scaled signals frame
In interpolated value between adjacent two sample point.
8. speech signal processing device according to claim 7, which is characterized in that the processing unit also judges in described
Insert parametric function value whether less than a upper limit value and be more than or equal to a lower limit value, if the value of the interpolated parameter function no less than
The upper limit value is not more than or equal to the lower limit value, the value of the interpolated parameter function is corrected, wherein if the interpolated parameter
The value of function is more than or equal to the upper limit value, the value of the interpolated parameter function is modified to the upper limit value, if the interpolation
The value of parametric function, which is less than, states lower limit value, and the value of the interpolated parameter function is modified to the lower limit value.
9. speech signal processing device according to claim 8, which is characterized in that the upper limit value and the lower limit value close
It is coupled to the frequency of the primary speech signal and the sampling frequency of the sampling primary speech signal.
10. speech signal processing device according to claim 7, which is characterized in that the processing unit is also according to each institute
It states the trigonometric function relationship in original down-scaled signals frame between continuous three sampling values and calculates each original down-scaled signals frame pair
The interpolated parameter function answered, wherein the interpolated parameter function is trigonometric function.
11. a kind of audio signal processing method characterized by comprising
One sampled voice signal of frequency reducing, to generate the down-scaled signals of the original down-scaled signals frame including a sequence, wherein each described
Original down-scaled signals frame includes p sample point, and wherein p is positive integer, wherein the sampled voice signal is former by sampling one
Beginning voice signal and generate；
It is updated corresponding to the intermediate sample point of down-scaled signals frame according in m-1 original down-scaled signals frames with the m-1
One phase reference sample point number determines corresponding with the phase reference sample point number in m-th of original down-scaled signals frame
First sample point of sample point phase matched, wherein m is positive integer greater than 1, the phase reference sample point number be
It updates in the m-1 original down-scaled signals frames with described the m-1 and is taken corresponding to the intermediate sample point of down-scaled signals frame
The number of sampling point；And
The company that will be lighted from first sampling for numbering corresponding sample point phase matched with the phase reference sample point
Continuous q sample point updates the sample point of down-scaled signals frame as m-th, and wherein q is positive integer；And
The adjacent update down-scaled signals frame of aliasing, to generate an overlapping voice signal.
12. audio signal processing method according to claim 11, which is characterized in that the foundation is original at m-1
The number of a phase reference sample point corresponding to the intermediate sample point of down-scaled signals frame is updated with the m-1 in down-scaled signals frame to determine
It is scheduled in described m-th original down-scaled signals frame and numbers the of corresponding sample point phase matched with the phase reference sample point
The step of one sample point includes:
Sampling value according to sample point in described m-th original down-scaled signals frame adds up one first count value and one second counting
Value, wherein when counting up to the sample point or an at least sample point adjacent with the sample point that sampling value is 0 that corresponding sample value is 0
When, be zeroed its corresponding first count value or second count value；
Institute corresponding to corresponding sample point will be numbered with the phase reference sample point in described m-th original down-scaled signals frame
The first count value or second count value are stated as an a reference value；And
It is determined according to a reference value corresponding with the phase reference sample point number in m-th of original down-scaled signals frame
First sample point of the phase matched of sample point.
13. audio signal processing method according to claim 12, which is characterized in that described by described m-th original drop
First count value corresponding to corresponding sample point or described are numbered in frequency signal frame with the phase reference sample point
Two count values include: as the step of a reference value
Judge in the m-1 original down-scaled signals frames corresponding to sample point corresponding with the phase reference sample point number
First count value whether be less than or equal in the m-1 original down-scaled signals frames and the phase reference sample point is compiled
Second count value corresponding to number corresponding sample point；
If being numbered corresponding to corresponding sample point in the m-1 original down-scaled signals frames with the phase reference sample point
First count value is less than or equal to corresponding with the phase reference sample point number in the m-1 original down-scaled signals frames
Second count value corresponding to sample point will sample in the m-1 original down-scaled signals frames with the phase reference
Point numbers first count value corresponding to corresponding sample point as a reference value；And
If being numbered corresponding to corresponding sample point in the m-1 original down-scaled signals frames with the phase reference sample point
First count value is no less than being equal in the m-1 original down-scaled signals frames and phase reference sample point number pair
Second count value corresponding to the sample point answered, by the m-1 original down-scaled signals frames with the phase reference
Sample point numbers second count value corresponding to corresponding sample point as a reference value.
14. audio signal processing method according to claim 13, which is characterized in that if the m-1 original
First count value corresponding to corresponding sample point is numbered with the phase reference sample point in down-scaled signals frame to be less than etc.
Described in numbering corresponding to corresponding sample point in the m-1 original down-scaled signals frames with the phase reference sample point
Second count value, the audio signal processing method include:
When by the first count value described in described m-th original down-scaled signals frame equal to a reference value in corresponding sample point
The sample point sampled earliest takes as corresponding with the phase reference sample point number in described m-th original down-scaled signals frame
First sample point of sampling point phase matched.
15. audio signal processing method according to claim 13, which is characterized in that if m-th of original drop
Number in frequency signal frame with the phase reference sample point first count value corresponding to corresponding sample point no less than etc.
It is numbered described corresponding to corresponding sample point in described m-th original down-scaled signals frame with the phase reference sample point
Two count values, the audio signal processing method include:
When by the second count value described in described m-th original down-scaled signals frame equal to a reference value in corresponding sample point
The sample point sampled earliest takes as corresponding with the phase reference sample point number in described m-th original down-scaled signals frame
First sample point of sampling point phase matched.
16. audio signal processing method according to claim 11 characterized by comprising
It is calculated according to three sampling values continuous in each original down-scaled signals frame corresponding with each original down-scaled signals frame
Interpolated parameter function value；
The value of the interpolated parameter function is judged whether less than a upper limit value and more than or equal to a lower limit value, if the interpolated parameter
The value of function is not more than or equal to the lower limit value no less than the upper limit value or, corrects the value of the interpolated parameter function；And
Value according to the corresponding interpolated parameter function of each down-scaled signals frame calculates in each down-scaled signals frame adjacent two and takes
Interpolated value between sampling point.
17. audio signal processing method according to claim 16, which is characterized in that if the interpolated parameter function
Value be more than or equal to the upper limit value, the value of the interpolated parameter function is modified to the upper limit value, if the interpolated parameter
The value of function is less than the lower limit value, the value of the interpolated parameter function is modified to the lower limit value, wherein the upper limit value
The frequency of the primary speech signal and the sampling frequency of the sampling primary speech signal are associated with the lower limit value.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510294353.1A CN106297814B (en) | 2015-06-02 | 2015-06-02 | Speech signal processing device and audio signal processing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201510294353.1A CN106297814B (en) | 2015-06-02 | 2015-06-02 | Speech signal processing device and audio signal processing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN106297814A CN106297814A (en) | 2017-01-04 |
| CN106297814B true CN106297814B (en) | 2019-08-06 |
Family
ID=57655557
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201510294353.1A Active CN106297814B (en) | 2015-06-02 | 2015-06-02 | Speech signal processing device and audio signal processing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN106297814B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2006133431A3 (en) * | 2005-06-08 | 2009-04-23 | Univ California | Methods, devices and systems using signal processing algorithms to improve speech intelligibility and listening comfort |
| CN101944364A (en) * | 2009-07-09 | 2011-01-12 | 展讯通信（上海）有限公司 | Voice frequency processing method and voice frequency system |
| CN102638755A (en) * | 2012-04-25 | 2012-08-15 | 南京邮电大学 | Digital hearing aid loudness compensation method based on frequency compression and movement |
| CN103503061A (en) * | 2011-02-14 | 2014-01-08 | 弗兰霍菲尔运输应用研究公司 | Apparatus and method for processing a decoded audio signal in a spectral domain |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| ATE449512T1 (en) * | 2005-06-27 | 2009-12-15 | Widex As | HEARING APPARATUS WITH IMPROVED HIGH FREQUENCY REPRODUCTION AND METHOD FOR PROCESSING A SOUND SIGNAL |
| SG191025A1 (en) * | 2010-12-08 | 2013-07-31 | Widex As | Hearing aid and a method of improved audio reproduction |
-
2015
- 2015-06-02 CN CN201510294353.1A patent/CN106297814B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2006133431A3 (en) * | 2005-06-08 | 2009-04-23 | Univ California | Methods, devices and systems using signal processing algorithms to improve speech intelligibility and listening comfort |
| CN101944364A (en) * | 2009-07-09 | 2011-01-12 | 展讯通信（上海）有限公司 | Voice frequency processing method and voice frequency system |
| CN103503061A (en) * | 2011-02-14 | 2014-01-08 | 弗兰霍菲尔运输应用研究公司 | Apparatus and method for processing a decoded audio signal in a spectral domain |
| CN102638755A (en) * | 2012-04-25 | 2012-08-15 | 南京邮电大学 | Digital hearing aid loudness compensation method based on frequency compression and movement |
Also Published As
| Publication number | Publication date |
|---|---|
| CN106297814A (en) | 2017-01-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Macleod | Fast nearly ML estimation of the parameters of real or complex single tones or resolved multiple tones | |
| CN112069235B (en) | Method, apparatus and storage medium for presenting target area demographic data | |
| CN104143324B (en) | A kind of musical tone recognition method | |
| CN106157976A (en) | A kind of singing evaluating method and system | |
| EP2392003B1 (en) | Audio signal quality prediction | |
| Li et al. | An improved EMD method with modified envelope algorithm based on C2 piecewise rational cubic spline interpolation for EMI signal decomposition | |
| US9431024B1 (en) | Method and apparatus for detecting noise of audio signals | |
| US20160217787A1 (en) | Speech recognition apparatus and speech recognition method | |
| US10068558B2 (en) | Method and installation for processing a sequence of signals for polyphonic note recognition | |
| Aunsri | A Bayesian Filtering Approach with Time-Frequency Representation for Corrupted Dual Tone Multi Frequency Identification. | |
| CN106297814B (en) | Speech signal processing device and audio signal processing method | |
| US10666490B1 (en) | Method and apparatus for estimation of a bit rate of a digital signal | |
| CN110186533A (en) | A kind of short-term tide prediction method in high-precision river mouth | |
| KR20120072661A (en) | Apparatus for real time clock frequency offset detection and method thereof | |
| JP4847466B2 (en) | Apparatus and method for determining arrival time of a reception sequence | |
| JP5319788B2 (en) | Audio signal alignment method | |
| EP2507791A2 (en) | Complex acoustic resonance speech analysis system | |
| JP2016500847A (en) | Digital processor based complex acoustic resonance digital speech analysis system | |
| TWI557729B (en) | Voice signal processing apparatus and voice signal processing method | |
| CN106328158B (en) | Speech signal processing device and audio signal processing method | |
| JP6891736B2 (en) | Speech processing program, speech processing method and speech processor | |
| CN107247819B (en) | Filtering method and filter for sensor | |
| KR20020028226A (en) | Method of calculating line spectral frequencies | |
| TWI583205B (en) | Voice signal processing apparatus and voice signal processing method | |
| KR102344223B1 (en) | Apparatus and method for detecting resistive leakage current of surge arrester for computational load reduction |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |