CN1920951A - Speed transformation method and system - Google Patents
Speed transformation method and system Download PDFInfo
- Publication number
- CN1920951A CN1920951A CNA2006101520831A CN200610152083A CN1920951A CN 1920951 A CN1920951 A CN 1920951A CN A2006101520831 A CNA2006101520831 A CN A2006101520831A CN 200610152083 A CN200610152083 A CN 200610152083A CN 1920951 A CN1920951 A CN 1920951A
- Authority
- CN
- China
- Prior art keywords
- signal
- sampling
- module
- correlation value
- processing
- 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.)
- Granted
Links
- 238000011426 transformation method Methods 0.000 title 1
- 238000000034 method Methods 0.000 claims abstract description 26
- 238000005070 sampling Methods 0.000 claims description 109
- 238000012545 processing Methods 0.000 claims description 94
- 238000001914 filtration Methods 0.000 claims description 43
- 230000008859 change Effects 0.000 claims description 42
- 238000004364 calculation method Methods 0.000 claims description 19
- 238000010030 laminating Methods 0.000 claims description 15
- 238000010586 diagram Methods 0.000 description 12
- 230000008569 process Effects 0.000 description 11
- 239000000203 mixture Substances 0.000 description 2
- 230000000630 rising effect Effects 0.000 description 2
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000004904 shortening Methods 0.000 description 1
- 230000005236 sound signal Effects 0.000 description 1
- 238000001228 spectrum Methods 0.000 description 1
Images
Landscapes
- Compression, Expansion, Code Conversion, And Decoders (AREA)
- Cable Transmission Systems, Equalization Of Radio And Reduction Of Echo (AREA)
Abstract
The invention relates to a speed-changing method, which divides the cross-correlation search range into at least one sections based on the search point value from lower to higher and the cross-correlation weights are from large to small, wherein based on the speed-changing rate, calculating the cross-correlation value between the input and output signals of search pint in the search range, the maximum cross-correlation value of each section, combining the cross-correlation weights, to find the last maximum cross-correlation value, and adding the input and output signals relative to the laser maximum cross-correlation value, to obtain the signal with changed speed. The invention also provides a speed-changing system, which comprises input buffer module, output buffer module, cross-correlation calculating module, maximum cross-correlation value fixing module and superposition module. The invention improves the smooth at the connected point of speed-changing signal, improves the voice quality of changed speed and improves the voice quality of changed tone.
Description
Technical field
The present invention relates to the sound signal processing technical field, be specifically related to a kind of method for changing speed and system that modifies tone in handling that be applied in.
Background technology
Modified tone (Pitch Scaling) is meant under the total length that does not change voice data, changes the tone of sound, and tone is raise or reduction.
The process that modifies tone is actually expands the frequency spectrum of sound, and the relation between simultaneously original each harmonic components still keeps, and modifying tone needs to realize by means of certain algorithm.The modified tone algorithm mainly is divided into Time-Domain algorithm and frequency domain algorithm two big classes.Frequency domain algorithm generally adopts phase vocoder (Phase Vocoder) to realize, relatively be fit to the dodgoing of vast scale, but the naturalness of the sound after modifying tone is lower; The method that Time-Domain algorithm generally adopts speed change to add variable sampling rate realizes, relatively is fit to the dodgoing of small scale, and during the tone large percentage, the sound joint after the speed change algorithm process is second-rate.
Speed change (Time Scaling) is promptly carried out the processing of speed-variation sound to signal, just the duration of one section sound is elongated or shortening, and the tone of sound is constant.
The concrete steps of existing variable-speed processing as shown in Figure 1, its concrete steps are as follows:
Step 101: establishing the modified tone ratio is a, determine that according to the modified tone ratio frame length of input signal is Sa, the frame length of output signal is Ss, and Sa/Ss=a, and Sa/Ss is called the speed change ratio, if the length of input block is 2W, the length of output buffer is W, and initialization k=1 is provided with simple crosscorrelation search terminating point Kmax, stack length Wov=W-Ss is set, and the concrete implication of each parameter as shown in Figure 2.
The value of a requires to set according to modifying tone, if rising tune processing, then a＜1; If falling tone is handled, then a＞1; Kmax rule of thumb sets, common 1＜Kmax＜150, and unit is a sampled point.
Step 102: from the input block, take out the individual input signal of k～(W+k).
1 input signal here i.e. 1 input signal sampled point.
K input signal be for entering k input signal of input block at first, and then the individual input signal of k～(W+k) is the individual input signal of the k that enters the input block at first～(W+k).
Step 103: get preceding Wov the input signal of individual input signal of this k～(W+k), calculate the cross correlation value of preceding Wov output signal of this Wov input signal and output buffer, obtain k cross correlation value.
Preceding Wov output signal of output buffer refers to Wov the output signal that enters output buffer at first.
Step 104: judge whether k=Kmax sets up, if, execution in step 106; Otherwise, execution in step 105.
Step 105: make k=k+1, go to step 102.
Step 106: in all cross correlation values, search maximum cross correlation value.
Step 107: preceding Wov input signal of W input signal of maximum cross-correlation value correspondence and preceding Wov output signal of output buffer are superposeed.
Step 108: replace preceding Wov output signal in the output buffer with stack result, with back Ss output signal in back Ss the input signal replacement output buffer of W input signal of maximum cross-correlation value correspondence.
Step 109: with the output signal of preceding Ss output signal in the output buffer after as speed change; Preceding Sa input signal of input block shifted out, a up-to-date Sa input signal is input to the input block.
The concrete processing of step 107～109 can be referring to Fig. 2.
To the variable rate signal that obtains through variable-speed processing at every turn, carry out variable sampling rate again and handle, to obtain variation signals.
Fig. 3 is the process flow diagram that existing variable sampling rate is handled, and as shown in Figure 3, its concrete steps are as follows:
Step 301: determine to increase sampling coefficient L and down-sampled coefficient M according to modified tone ratio a, and L/M=a, to the signal after the speed change carry out L doubly increase sampling processing.
In concrete the application, in order to reduce the complexity that increases sampling and down-sampled processing, should be provided with the value of L, M as much as possible little, usually: when L, M are set, as long as L, M do not have common divisor.
Through after the processing of this step, the length of the signal after the speed change is expanded to by Ss: Ss*L.
Step 302: the signal that increases after the sampling processing is carried out low-pass filtering treatment.
Low-pass filter be that (π/M, π/L), the purpose of low-pass filtering is that the filtering signals of frequency greater than f fallen to f=min by frequency.
Step 303: the signal after the Filtering Processing is carried out the down-sampled processing of 1/M, the signal after obtaining modifying tone.
After the processing of this step, the length of signal becomes Ss*L/M=Sa by Ss*L, and as can be seen: the length of signal does not change.
As can be seen from Figure 1, in variable-speed processing, most critical be find with output buffer in Wov the strongest input signal of preceding Wov output signal cross correlation.Prior art is searched in the 1～Kmax of simple crosscorrelation hunting zone and the strongest input signal of output signal cross correlation; though the input signal of maximum cross-correlation value correspondence is and the strongest input signal of output signal cross correlation in theory; but; when the value of the simple crosscorrelation of maximum cross-correlation value correspondence search point is big, usually can cause joint second-rate of the sound after adjacent twice variable-speed processing.
The shortcoming of variable sampling rate disposal route shown in Figure 3 is: when the modified tone ratio hour, the value of L, M is often more approaching, and value is all bigger, at this moment, bring difficulty can for the design of low-pass filter, operand is also bigger simultaneously.
Summary of the invention
The invention provides method for changing speed and system, to improve sound quality.
Technical scheme of the present invention is achieved in that
A kind of method for changing speed is divided into the more than one stage with the simple crosscorrelation hunting zone from low to high according to the search point value, according to the search point value that is comprised from low to high, for each stage is provided with the simple crosscorrelation weights from big to small, comprising:
According to the speed change ratio, calculate the input signal of each the search point correspondence in the simple crosscorrelation hunting zone and the cross correlation value of output signal, determine the maximum cross-correlation value in each stage, simple crosscorrelation weights in conjunction with each stage, determine final maximum cross-correlation value, the input signal of the maximum cross-correlation value correspondence finally determined is superposeed with output signal, obtain the signal after the speed change.
Described definite final maximum cross-correlation value comprises:
Maximum cross-correlation value in each stage be multiply by the simple crosscorrelation weights in this stage, with the maximum cross-correlation value in stage of max product correspondence as final maximum cross-correlation value.
The stack of the input signal of the described maximum cross-correlation value correspondence that will finally determine and output signal, the signal that obtains after the speed change comprises:
According to predetermined stack length Wov, with the signal of the preceding Wov length in the input signal of finally definite maximum cross-correlation value correspondence and the signal stack of the preceding Wov length in the output signal, the signal of the preceding Ss length in the signal that stack is obtained is as variable rate signal, wherein, Ss is the length according to the definite variable rate signal of speed change ratio.
Further comprise after the described signal that obtains after the speed change:
According to the modified tone ratio, the signal after the speed change is carried out variable sampling rate handle the signal after obtaining modifying tone.
Described according to the modified tone ratio, the signal after the speed change is carried out the variable sampling rate processing comprise:
Determine to increase sampling coefficient and down-sampled coefficient according to the modified tone ratio, to increase the sampling coefficient is divided into more than one son and increases the sampling coefficient, increasing the signal of sampling coefficient after to speed change according to each son carries out classification and increases sampling and low-pass filtering treatment, according to down-sampled coefficient, the signal after the low-pass filtering treatment is carried out down-sampled processing then.
The first order processing that described classification increases in sampling and the low-pass filtering treatment comprises: increase the sampling coefficient according to the first order of determining, signal after the speed change is increased sampling processing, the signal that increases after the sampling processing is carried out low-pass filtering, filtered signal is increased the input signal of sampling processing as next stage;
The afterbody processing that described classification increases in sampling and the low-pass filtering treatment comprises: increase the sampling coefficient according to afterbody of determining, signal after the previous stage low-pass filtering is increased sampling processing, the signal that increases after the sampling processing is carried out low-pass filtering, with the input signal of filtered signal as down-sampled processing;
Described classification increases in sampling and the low-pass filtering treatment classification between the first order and afterbody to be handled and to comprise: increase the sampling coefficient according to the son of determining when prime, signal after the previous stage low-pass filtering is increased sampling processing, the signal that increases after the sampling processing is carried out low-pass filtering, filtered signal is increased the input signal of sampling processing as next stage.
Describedly the simple crosscorrelation weights are set are for each stage: the ratio of simple crosscorrelation weights that the simple crosscorrelation weights of stage A and adjacent phases B are set is for greater than 1 and less than 1.5 value,
The value of simple crosscorrelation that wherein, stage A comprised search point is lower than the value of the simple crosscorrelation search point that stage B comprises.
A kind of speed change system comprises: input buffer module, output buffer module, cross-correlation calculation module, maximum cross-correlation value determination module and laminating module, wherein:
Input buffer module according to request, outputs to cross-correlation calculation module, laminating module, output buffer module with self input signal;
Output buffer module, according to request, the output signal that self is participated in cross-correlation calculation outputs to the cross-correlation calculation module, replaces the output signal piece of self with the signal of laminating module and input buffer module output, according to the speed change ratio, the variable rate signal in self signal is outputed to the outside;
The cross-correlation calculation module is used to calculate the output signal obtained from output buffer module and the cross correlation value of the input signal that obtains from input buffer module, and the cross correlation value that obtains and corresponding simple crosscorrelation search point are outputed to the maximum cross-correlation value determination module;
The maximum cross-correlation value determination module, be used for the simple crosscorrelation hunting zone is divided into the more than one stage from low to high according to the search point value, and for each stage simple crosscorrelation weights are set respectively from big to small from low to high according to the search point value that is comprised, in the cross correlation value that the cross-correlation calculation module is sent, search the maximum cross-correlation value in each stage respectively, simple crosscorrelation weights according to each stage are determined final maximum cross-correlation value, send notice from input signal to laminating module that export to input buffer module according to this maximum cross-correlation value;
Laminating module is used for input signal and output signal stack that input buffer module and output buffer module are sent, and stack result is outputed to output buffer module.
Described system further comprises: the variable sampling rate processing module, be used for according to the modified tone ratio, and the variable rate signal that output buffer module is exported carries out the variable sampling rate processing, and the variation signals that obtains is outputed to the outside.
Described variable sampling rate module comprises: son increases the sampling coefficient determination module, classification increases sampling and Filtering Processing module and down-sampled processing module, wherein:
Son increases the sampling coefficient determination module, is used for according to the modified tone ratio, determines that more than one son increases the sampling coefficient, with each height increase the sampling coefficient output to classification increase the sampling and the Filtering Processing module;
Classification increases sampling and Filtering Processing module, be used for increasing each height of sending of sampling coefficient determination module and increase the sampling coefficient according to son, the variable rate signal that output buffer module is exported carries out classification and increases sampling processing and low-pass filtering treatment successively, and the signal after handling is sent to down-sampled processing module;
Down-sampled processing module is used for according to the modified tone ratio, and classification is increased sampling and the signal sent of Filtering Processing module carries out down-sampled processing, and the signal after the processing is outputed to the outside.
Compared with prior art, the present invention is by being divided into the more than one stage with the simple crosscorrelation hunting zone from low to high according to the value of searching for point, according to search point value that each stage comprised from low to high, being followed successively by each stage is provided with the simple crosscorrelation weights from big to small, according to the speed change ratio, calculate the input signal of each the search point correspondence in the simple crosscorrelation hunting zone and the cross correlation value of output signal, determine the maximum cross-correlation value in each stage, simple crosscorrelation weights according to each stage, determine final maximum cross-correlation value, input signal and output signal stack with finally definite maximum cross-correlation value correspondence, obtain the signal after the speed change, the present invention is the less corresponding bigger weights of input signal setting of search point of value, increased and be worth the probability that the corresponding input signal of less search point is used as the maximum cross correlation signal of output signal, thereby improved the flatness of the joint of adjacent variable rate signal, improved the voice quality after the speed change, thereby improved the voice quality after modifying tone.In addition, the present invention increases sampling processing by classification in variable sampling rate is handled, reduced computational complexity, and reduced signal aliasing, has further improved the voice quality after modifying tone.
Description of drawings
Fig. 1 is the process flow diagram of existing variable-speed processing;
Fig. 2 is the process synoptic diagram of existing variable-speed processing;
Fig. 3 is the process flow diagram that existing variable sampling rate is handled;
Fig. 4 is the process flow diagram of variable-speed processing provided by the invention;
Fig. 5 is the process flow diagram that variable sampling rate provided by the invention is handled;
Fig. 6 is that the system of variable-speed processing provided by the invention forms synoptic diagram;
Fig. 7 is the composition synoptic diagram of variable sampling rate module provided by the invention.
Embodiment
Cause the second-rate reason of the joint of the sound after the variable-speed processing to be: what the search point with the strongest input signal place of output signal cross correlation that calculates was corresponding is not real fundamental tone time-delay, this be because: voice signal has pitch period, therefore, when having undesired signal on the input signal, can cause the search point at the input signal place that searches with maximum cross-correlation value corresponding be not real fundamental tone time-delay, but the integral multiple of real fundamental tone time-delay, that is: in fact the input signal with maximum cross-correlation value that obtains this moment is not and the strongest input signal of output signal cross correlation.Therefore, core concept of the present invention is: the simple crosscorrelation hunting zone is divided into the more than one stage from low to high successively according to the search point value, according to the search point value that comprises from low to high, being followed successively by each stage is provided with simple crosscorrelation weights from big to small, find out the input signal in each stage and the maximum cross-correlation value of output signal respectively, simple crosscorrelation weights and the maximum cross-correlation value in each stage according to each stage, determine final maximum cross-correlation value, determine and the strongest input signal of output signal cross correlation according to this final maximum cross-correlation value, to be added on the output signal with the strongest input signal of output signal cross correlation, obtain the output signal after the speed change.
Among the present invention, for the lower stage of search point value that comprises is provided with bigger simple crosscorrelation weights, with increase the corresponding input signal of search point in this stage as with the probability of the strongest input signal of output signal cross correlation, the flatness of the signal after the raising speed change.
Below in conjunction with drawings and the specific embodiments variable-speed processing among the present invention and variable sampling rate processing are described in further detail respectively.
Fig. 4 is the process flow diagram of variable-speed processing provided by the invention, and as shown in Figure 4, its concrete steps are as follows:
Step 401: modified tone ratio a is set according to the modified tone demand, determine that according to modified tone ratio a the frame length of input signal is Sa, the frame length of output signal is Ss, and Sa/Ss=a is called the speed change ratio with Sa/Ss, and the length of establishing the input block is 2W, the length of output buffer is W, stack length is Wov=W-Ss, simple crosscorrelation search starting point K0, simple crosscorrelation search terminating point Kmax is set, initialization k=K0.
Common K0=20, K0＜Kmax＜150.
Step 402: with simple crosscorrelation hunting zone [K0, the Kmax] value according to the search point, be divided into three phases from low to high, the phase one: [K0,40], subordinate phase: [40,80], the phase III: [80, Kmax], the simple crosscorrelation weights that the phase one is set are m
2, the simple crosscorrelation weights of subordinate phase are m, the simple crosscorrelation weights of phase III are 1.
The span of m is: 1＜m＜1.5, get 1.2 usually.
In actual applications, can be as required, each stage of searching for for simple crosscorrelation is provided with the simple crosscorrelation weights, and the principle that should follow is: the simple crosscorrelation weights that are provided with for the low stage are higher than the simple crosscorrelation weights that are provided with for high-stage.
Step 403: from the input block, take out the individual input signal of k～(W+k).
Step 404: get preceding Wov the input signal of individual input signal of this k～(W+k), calculate the cross correlation value of preceding Wov output signal of this Wov input signal and output buffer.
Step 405: judge whether k=Kmax sets up, if, execution in step 407; Otherwise, execution in step 406.
Step 406: make k=k+1, go to step 403.
Step 407: search maximum cross-correlation value in this stage in the search phase in each simple crosscorrelation, the maximum cross-correlation value of establishing three phases is respectively: Cof1, Cof2 and Cof3.
Step 408: judge Cof3＞m*Cof2 and Cof3＞m
2* whether Cof1 sets up simultaneously, if, execution in step 409; Otherwise, execution in step 410.
Step 409: Cof3 as maximum cross-correlation value, is gone to step 413.
Step 410: judge whether Cof2＞m*Cof1 sets up, if, execution in step 411; Otherwise, execution in step 412.
Step 411: Cof2 as maximum cross-correlation value, is gone to step 413.
Step 412: with Cof1 as maximum cross-correlation value.
Step 413: preceding Wov input signal of W input signal of maximum cross-correlation value correspondence and preceding Wov output signal of output buffer are superposeed.
Step 414: replace preceding Wov output signal in the output buffer with stack result, with back Ss output signal in back Ss the input signal replacement output buffer of W input signal of maximum cross-correlation value correspondence.
Step 415: with the output signal output of preceding Ss output signal in the output buffer after as speed change; The 1st～Sa input signal of input block shifted out, a up-to-date Sa input signal is input to the input block.
Ss the output signal that speed change is obtained just can obtain the variation signals that the modified tone ratio is L/M after carrying out the L down-sampled processing that increases sampling processing and 1/M doubly.
It is pointed out that in actual applications, can as required the simple crosscorrelation hunting zone be divided into the more than one stage.
Fig. 5 is the process flow diagram that variable sampling rate provided by the invention is handled, and as shown in Figure 5, its concrete steps are as follows:
Step 501: determine to increase sampling coefficient L and down-sampled coefficient M according to modified tone ratio a, and L/M=a, will increase sampling coefficient L and convert two or more products to, each integer is called son increases the sampling coefficient greater than 1 integer.
When definite modified tone ratio a, should consider that a can be expressed as two as far as possible little integer L, form that M is divided by, and should be taken into account that L should be expressed as two or more products greater than 1 integer.
Step 502: increase the sampling coefficient according to each son that obtains, the signal after the speed change is carried out classification increase sampling and low-pass filtering treatment.
For example: L=L1*L2, then earlier the signal after the speed change is carried out L1 times the sampling processing that increases, the length of signal becomes Ss*L1, then signal is carried out the low-pass filtering treatment that cutoff frequency is π/L1, to filtered signal carry out L2 doubly increase sampling processing, the length of signal becomes Ss*L1*L2, again signal is carried out the low-pass filtering place that cutoff frequency is π/L2.
Step 503: the signal after the Filtering Processing is carried out the down-sampled processing of 1/M, obtain variation signals.
Fig. 6 is that speed change system provided by the invention is formed synoptic diagram, and as shown in Figure 6, it mainly comprises: input buffer module 61, output buffer module 62, cross-correlation calculation module 63, maximum cross-correlation value determination module 64 and laminating module 65, wherein:
Input buffer module 61: be used for receiving and preserve input signal,, Wov input signal of correspondence outputed to cross-correlation calculation module 63 according to the request of cross-correlation calculation module 63; According to the request of maximum cross-correlation value determination module 64, Wov input signal of correspondence outputed to laminating module 65, simultaneously Ss input signal after this Wov input signal outputed to output buffer module 62.
Output buffer module 62: be used to preserve output signal, request according to cross-correlation calculation module 63, a preceding Wov output signal is outputed to cross-correlation calculation module 63, replace preceding Wov the output signal of self with Wov signal of laminating module 64 outputs, with back Ss output signal of Ss the input signal replacement self of input buffer module 61 output, self preceding Ss output signal outputed to the variable sampling rate processing module.
Cross-correlation calculation module 63: be used for Wov output signal before output buffer module 62 is obtained, obtain (the individual input signal of k=K0～Kmax) of k～(k+Wov) to input buffer module 61, the input signal that calculating gets access to and the cross correlation value of output signal output to maximum cross-correlation value determination module 64 with the cross correlation value that obtains and a corresponding simple crosscorrelation search point k value.
Maximum cross-correlation value determination module 64: be used for simple crosscorrelation hunting zone being divided into from low to high the more than one stage according to the search point value, according to search point value that each stage comprised from low to high, for each stage is provided with simple crosscorrelation weights from big to small respectively, a cross correlation value that reception cross-correlation calculation module 63 is sent and a corresponding simple crosscorrelation search point k, stage according to the simple crosscorrelation hunting zone of preserving in advance divides, search the maximum cross-correlation value in each stage respectively, simple crosscorrelation weights and the maximum cross-correlation value in each stage according to each stage, determine final maximum cross-correlation value, send notice from Wov the input signal corresponding to laminating module 65 that export to input buffer module 61 with the final maximum cross-correlation value of determining.
Laminating module 65: be used for after receiving Wov the input signal that input buffer module 61 is sent, Wov output signal superposes input signal and output signal before output buffer module 62 is obtained, and stack result is outputed to output buffer module 62.
Further, variable-speed processing system shown in Figure 6 can comprise: variable sampling rate processing module 66, and be used for that Ss output signal of output buffer module 62 outputs carried out variable sampling rate and handle, will handle the variation signals that obtains through variable sampling rate and output to the outside.
Fig. 7 is the composition synoptic diagram of variable sampling rate processing module provided by the invention, and as shown in Figure 7, it mainly comprises: son increases sampling coefficient determination module 661, classification increases sampling and Filtering Processing module 662 and down-sampled processing module 663, wherein:
Son increase the sampling coefficient determination module 661: be used for according to the modified tone ratio, determine that each son increases the sampling coefficient, Jiang Gezi increase the sampling coefficient output to classification increase the sampling and Filtering Processing module 662.
Classification increases sampling and Filtering Processing module 662: be used to receive the variable rate signal that output buffer module 62 is sent, increase each son of sending of sampling coefficient determination module 661 according to son and increase the sampling coefficient, variable rate signal is carried out classification increase sampling processing and low-pass filtering treatment, the signal after handling is sent to down-sampled processing module 663.
Down-sampled processing module 663: be used to receive classification and increase the signal that sampling and Filtering Processing module 662 are sent,, this signal is carried out down-sampled processing, the signal after the processing is outputed to the outside according to the modified tone ratio.
Experiment shows, the method that the present invention proposes is that 1～2 times of rising tune in the scope is handled for the modified tone ratio, and the modified tone ratio is that falling tone in 1～0.5 is handled and can both be obtained the good treatment effect, do not introduce noise, kept the naturalness of sound simultaneously.
The above only is process of the present invention and method embodiment, in order to restriction the present invention, all any modifications of being made within the spirit and principles in the present invention, is not equal to replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (10)
1, a kind of method for changing speed is characterized in that, the simple crosscorrelation hunting zone is divided into the more than one stage from low to high according to the search point value, according to the search point value that is comprised from low to high, for each stage is provided with the simple crosscorrelation weights from big to small, comprising:
According to the speed change ratio, calculate the input signal of each the search point correspondence in the simple crosscorrelation hunting zone and the cross correlation value of output signal, determine the maximum cross-correlation value in each stage, simple crosscorrelation weights in conjunction with each stage, determine final maximum cross-correlation value, the input signal of the maximum cross-correlation value correspondence finally determined is superposeed with output signal, obtain the signal after the speed change.
2, the method for claim 1 is characterized in that, described definite final maximum cross-correlation value comprises:
Maximum cross-correlation value in each stage be multiply by the simple crosscorrelation weights in this stage, with the maximum cross-correlation value in stage of max product correspondence as final maximum cross-correlation value.
3, the method for claim 1 is characterized in that, the stack of the input signal of the described maximum cross-correlation value correspondence that will finally determine and output signal, and the signal that obtains after the speed change comprises:
According to predetermined stack length Wov, with the signal of the preceding Wov length in the input signal of finally definite maximum cross-correlation value correspondence and the signal stack of the preceding Wov length in the output signal, the signal of the preceding Ss length in the signal that stack is obtained is as variable rate signal, wherein, Ss is the length according to the definite variable rate signal of speed change ratio.
4, the method for claim 1 is characterized in that, further comprises after the described signal that obtains after the speed change:
According to the modified tone ratio, the signal after the speed change is carried out variable sampling rate handle the signal after obtaining modifying tone.
5, method as claimed in claim 4 is characterized in that, and is described according to the modified tone ratio, the signal after the speed change carried out the variable sampling rate processing comprise:
Determine to increase sampling coefficient and down-sampled coefficient according to the modified tone ratio, to increase the sampling coefficient is divided into more than one son and increases the sampling coefficient, increasing the signal of sampling coefficient after to speed change according to each son carries out classification and increases sampling and low-pass filtering treatment, according to down-sampled coefficient, the signal after the low-pass filtering treatment is carried out down-sampled processing then.
6, method as claimed in claim 5, it is characterized in that, the first order processing that described classification increases in sampling and the low-pass filtering treatment comprises: increase the sampling coefficient according to the first order of determining, signal after the speed change is increased sampling processing, the signal that increases after the sampling processing is carried out low-pass filtering, filtered signal is increased the input signal of sampling processing as next stage;
The afterbody processing that described classification increases in sampling and the low-pass filtering treatment comprises: increase the sampling coefficient according to afterbody of determining, signal after the previous stage low-pass filtering is increased sampling processing, the signal that increases after the sampling processing is carried out low-pass filtering, with the input signal of filtered signal as down-sampled processing;
Described classification increases in sampling and the low-pass filtering treatment classification between the first order and afterbody to be handled and to comprise: increase the sampling coefficient according to the son of determining when prime, signal after the previous stage low-pass filtering is increased sampling processing, the signal that increases after the sampling processing is carried out low-pass filtering, filtered signal is increased the input signal of sampling processing as next stage.
7, the method for claim 1 is characterized in that, describedly the simple crosscorrelation weights are set are for each stage: the ratio of simple crosscorrelation weights that the simple crosscorrelation weights of stage A and adjacent phases B are set is for greater than 1 and less than 1.5 value,
The value of simple crosscorrelation that wherein, stage A comprised search point is lower than the value of the simple crosscorrelation search point that stage B comprises.
8, a kind of speed change system is characterized in that, comprising: input buffer module, output buffer module, cross-correlation calculation module, maximum cross-correlation value determination module and laminating module, wherein:
Input buffer module according to request, outputs to cross-correlation calculation module, laminating module, output buffer module with self input signal;
Output buffer module, according to request, the output signal that self is participated in cross-correlation calculation outputs to the cross-correlation calculation module, replaces the output signal piece of self with the signal of laminating module and input buffer module output, according to the speed change ratio, the variable rate signal in self signal is outputed to the outside;
The cross-correlation calculation module is used to calculate the output signal obtained from output buffer module and the cross correlation value of the input signal that obtains from input buffer module, and the cross correlation value that obtains and corresponding simple crosscorrelation search point are outputed to the maximum cross-correlation value determination module;
The maximum cross-correlation value determination module, be used for the simple crosscorrelation hunting zone is divided into the more than one stage from low to high according to the search point value, and for each stage simple crosscorrelation weights are set respectively from big to small from low to high according to the search point value that is comprised, in the cross correlation value that the cross-correlation calculation module is sent, search the maximum cross-correlation value in each stage respectively, simple crosscorrelation weights according to each stage are determined final maximum cross-correlation value, send notice from input signal to laminating module that export to input buffer module according to this maximum cross-correlation value;
Laminating module is used for input signal and output signal stack that input buffer module and output buffer module are sent, and stack result is outputed to output buffer module.
9, system as claimed in claim 8, it is characterized in that described system further comprises: the variable sampling rate processing module is used for according to the modified tone ratio, variable rate signal to output buffer module output carries out the variable sampling rate processing, and the variation signals that obtains is outputed to the outside.
10, system as claimed in claim 9 is characterized in that, described variable sampling rate module comprises: son increases the sampling coefficient determination module, classification increases sampling and Filtering Processing module and down-sampled processing module, wherein:
Son increases the sampling coefficient determination module, is used for according to the modified tone ratio, determines that more than one son increases the sampling coefficient, with each height increase the sampling coefficient output to classification increase the sampling and the Filtering Processing module;
Classification increases sampling and Filtering Processing module, be used for increasing each height of sending of sampling coefficient determination module and increase the sampling coefficient according to son, the variable rate signal that output buffer module is exported carries out classification and increases sampling processing and low-pass filtering treatment successively, and the signal after handling is sent to down-sampled processing module;
Down-sampled processing module is used for according to the modified tone ratio, and classification is increased sampling and the signal sent of Filtering Processing module carries out down-sampled processing, and the signal after the processing is outputed to the outside.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006101520831A CN100561577C (en) | 2006-09-11 | 2006-09-11 | The method for changing speed of voice signal and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006101520831A CN100561577C (en) | 2006-09-11 | 2006-09-11 | The method for changing speed of voice signal and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1920951A true CN1920951A (en) | 2007-02-28 |
| CN100561577C CN100561577C (en) | 2009-11-18 |
Family
ID=37778659
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2006101520831A Active CN100561577C (en) | 2006-09-11 | 2006-09-11 | The method for changing speed of voice signal and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100561577C (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101290775B (en) * | 2008-06-25 | 2011-09-14 | 无锡中星微电子有限公司 | Method for rapidly realizing speed shifting of audio signal |
| CN101354889B (en) * | 2008-09-18 | 2012-01-11 | 北京中星微电子有限公司 | Method and apparatus for tonal modification of voice |
| CN101399036B (en) * | 2007-09-30 | 2013-05-29 | 三星电子株式会社 | Device and method for conversing voice to be rap music |
| CN106469559A (en) * | 2015-08-19 | 2017-03-01 | 中兴通讯股份有限公司 | The method of adjustment of speech data and device |
| CN107068160A (en) * | 2017-03-28 | 2017-08-18 | 大连理工大学 | A kind of regular system and method for voice duration |
-
2006
- 2006-09-11 CN CNB2006101520831A patent/CN100561577C/en active Active
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101399036B (en) * | 2007-09-30 | 2013-05-29 | 三星电子株式会社 | Device and method for conversing voice to be rap music |
| CN101290775B (en) * | 2008-06-25 | 2011-09-14 | 无锡中星微电子有限公司 | Method for rapidly realizing speed shifting of audio signal |
| CN101354889B (en) * | 2008-09-18 | 2012-01-11 | 北京中星微电子有限公司 | Method and apparatus for tonal modification of voice |
| CN106469559A (en) * | 2015-08-19 | 2017-03-01 | 中兴通讯股份有限公司 | The method of adjustment of speech data and device |
| CN106469559B (en) * | 2015-08-19 | 2020-10-16 | 中兴通讯股份有限公司 | Voice data adjusting method and device |
| CN107068160A (en) * | 2017-03-28 | 2017-08-18 | 大连理工大学 | A kind of regular system and method for voice duration |
| CN107068160B (en) * | 2017-03-28 | 2020-04-28 | 大连理工大学 | Voice time length regulating system and method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100561577C (en) | 2009-11-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN1920951A (en) | Speed transformation method and system | |
| CN1763846A (en) | Voice gain factor estimating device and method | |
| CN1241170C (en) | Method and system for line spectral frequency vector quantization in speech codec | |
| CN1494353A (en) | Sounder signal processor and its method | |
| CN1666456A (en) | Clock data recovery circuit | |
| CN1866739A (en) | Delay circuit and semiconductor device including same | |
| CN1909383A (en) | Narrow-band interference rejection method and its device | |
| CN1538650A (en) | Multicarrier signal wave removing device and method | |
| CN101034878A (en) | Gain adjusting method and gain adjusting device | |
| CN101055467A (en) | IDP based industrial process dynamic optimization system and method | |
| CN1554147A (en) | Digital filter and its designing method | |
| CN1149534C (en) | Sound decoding device and sound decoding method | |
| CN1885216A (en) | Chaos generation method based on direct delay feedback | |
| CN101042691A (en) | System, apparatus, method, and computer program product for processing an integer transform | |
| CN103440870A (en) | Method and device for voice frequency noise reduction | |
| CN1237719C (en) | Phase-locked loop circuit of eliminating self-shaking in signals received by control circuit | |
| CN1149739C (en) | Phase-Locking oscillation circuit | |
| CN1492359A (en) | Automatic state machine searching and matching method of multiple key words | |
| CN1472726A (en) | Device and method for determining coretative coefficient between signals and signal sectional distance | |
| CN1870131A (en) | Character string retrieving circuit and character string retrieving method | |
| CN1913320A (en) | Digital controlled inverter and its control method | |
| CN100336370C (en) | SDH frame header detection and data rearrangement circuit | |
| CN1905713A (en) | Method and system for estimating and regulating mobile terminal frequency deviation | |
| CN1229919C (en) | Adaptive line enhancer | |
| CN1805279A (en) | Method for handling adaptive filter in transform domain |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C41 | Transfer of patent application or patent right or utility model | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20151222 Address after: 519031 Guangdong city of Zhuhai province Hengqin Baohua Road No. 6, room 105 -478 Patentee after: GUANGDONG ZHONGXING ELECTRONICS CO., LTD. Address before: 100083, Haidian District, Xueyuan Road, Beijing No. 35, Nanjing Ning building, 15 Floor Patentee before: Beijing Vimicro Corporation |