TW200412570A - Method for changing voice tone - Google Patents

Abstract

The present invention provides a method for changing voice tone, which includes the following steps: providing a time frame; sequentially storing digital data into the frame with a constant sampling frequency; and using the sampling frequency to convert a synthesized value into an analog signal output, wherein the synthesized value is obtained by performing a weighting operation on the data taken from the frame by a first pointer and a second pointer, and the first pointer and second pointer are changed corresponding to an offset.

Description

200412570 V. Description of the invention (1) [Technical field to which the invention belongs] This case is a method for changing the tone of a sound, which is characterized by the ability to change the tone of the sound in real time without the need for complicated calculations and a large amount of memory. [Previous Technology] A voice changer is a device that is widely used. It can change the characteristics of the speaker's tone. If it is used in consumer products such as karaoke or children's toys, it can achieve excellent entertainment effects; and in some communication equipment , Such as mobile phones or telephones, you can also use a voice changer to hide the identity of the caller. To change the tone of a voice changer, it is mainly accomplished through the algorithm of the tone change. Since the voice after the tone change must be output immediately, the calculation of the tone change is immediate. The conventional technique used to change the tone is to first convert the speech signal to the frequency domain (frequencyd 0mai η) for processing, and then convert it back to the time domain (timedomai η) output. However, this method not only has high computational complexity, It also requires a large amount of memory. If the real-time computing needs are to be achieved, the hardware cost is relatively high, so this technology cannot be adopted in low- and medium-priced consumer products. In addition, by using the principle that the rotation speed of the turntable can increase or decrease the pitch, we can increase the frequency of the playback (the pitch increases) or slow down (the pitch decreases) during the digital audio processing to achieve the purpose of changing the pitch. . However, this method will cause the playback time to be different from the original speaker's time after processing the sound data, so it cannot be directly applied to situations where the tone needs to be changed immediately.

200412570 I. Description of the invention (2) [Objective of the case] In order to meet the above requirements, this case is to conceive a magic law. The processed voice data can not only achieve the effect of tone rise and fall under the hardness of the square-storage rotation of the sound only. And, it only needs to be the same, and the time can be completed with a long inventory. In addition, the calculation and sampling frequency of the ADC and digital analog conversion (DAC) conceived in this case are more convenient to implement analog digital. To achieve the real-time (real, I., the same, to make the tone of the tone. U me) change the sound [Abstract] In order to achieve the above purpose, this case proposes a method of changing the tone of the sound, which includes the following steps: provide a time frame; take a sample Frequency, sequentially storing a number of data into the time frame; and using the sampling frequency, a composite value $ is: an analog signal output, where the composite value is determined by a first indicator and a second indicator since then The information extracted from the frame is obtained after a weighting (weighting # w), and the first index and the second index are changed according to an offset #. The person changes the sound as described Method, wherein the time axis is a frame of sound data. 'The method of changing the tone of a sound as described, wherein the sampling frequency is a frequency of sampling from analog to digital. Changing the tone of a sound as described The method, wherein the sampling frequency is a digital to analog sampling frequency.

Page 5 200412570 V. Description of the invention (3) The method of changing the tone of a sound as described, wherein the digital data is sequentially and repeatedly stored in the time frame. As described in the method for changing the tone of a sound, the pitch of the sound can be adjusted by changing the offset 5, where 5 > 0 means that the tone of the output sound is increased, 5 < 0 means that the tone is decreased, and 5 = 0. The tone does not change. As described in the method for changing the tone of a sound, in response to a change of the offset 5, each time data is fetched from the time frame, the first index p 1 =, and then the second index p 2 = pl + (N / 2), where N is the length of the time frame.

As described in the method for changing the tone of a sound, in response to the first index p 1 > (N-1), let pl = pl-N, where N is the length of the time frame. As described in the method for changing the tone of a sound, in response to the first index p 1 < 0, let pl = pl + N, where N is the length of the time frame. As described in the method for changing the tone of a sound, in response to the second index p2> (N-1), let N be the length of the time frame. As described in the method for changing the tone of a sound, in response to the second index ρ2 < 0, let ρ2 = ρ2 + Ν, where N is the length of the time frame. The method of changing the tone of a sound as described, wherein the weight operation is a Triangular Window conversion.

[Embodiment] Please refer to the first figure, which is a system block diagram of a preferred embodiment of the method for changing the tone of a sound. As shown in the figure, I set the sampling frequency of the analog to digital converter 11 (ADC, Analog to Digital Converter) as

Page 6 200412570 V. Description of the invention (4) SR, then the sampled data is processed by the tone change algorithm 1 2 and then sent through the digital to analog converter 13 (DAC, Digital to Analog Converter) at the same sampling frequency SR . The second picture is the frame of sound data, whose length is N. The data in the frame is represented by x [0], ..., x [N-1]. The third picture is the weighting corresponding to the frame data in the second picture. For the convenience of explanation, this case uses the Triangu 1 ar W indow conversion as an example. Of course, other functions such as Hamming window or Hanning window. The following are the detailed steps of the tone changing algorithm: Step 1. Preset the data in the frame X [0],…, X [N-1] in the second picture to 0. Step 2. Set the first indicator in the third picture p 1 is preset to 0. Step 3: Determine the offset 5, 5 > 0 means the tone of the output sound is increased, 5 < 0 means the tone is decreased, and 5 = 0 the tone is not changed. Step 4. Obtain a new sampling value from the ADC in the first figure, and send the sampling value to the box at the time of the second picture, and update the data in the box at the time of updating, that is: Φ] = Φ + 1] for ι = 0Α (N-2) for γ-1] = new sampling. Step 5. Calculate the first index p 1 and the second index p 2:

200412570 V. Description of the invention (5) ρί = pi-l · δ if (pl > N) ρ \ = ρ \ -N if (pi < 0) = + ^ ρ2 = ρ \ Λ- ^) if (ρ2 & gt Ν) ρ2 = ρ2-Ν if {ρ2 < 0) ρ2 = p2-l · Ν Step 6. Perform a weight calculation, and use the weights w [ρ 1] and w [ρ2] to calculate a new composite value:

The new compound 値 = 4 households 1] X pair of households 1] + ^ [;? 2] χ pair of households 2] The above formula can be simplified into a new compound 値 = W | > 1] X 4 households 1] + (1 -Η > 1]) X ζ | > 2] = 4 ^ 2] + (pipl)-κ [ρ2]) χ ^ ip \]

Step 7. Send the new synthesized value to the DAC, and put it at the same sampling frequency SR as the ADC. Step 8. Repeat step 4 to achieve the function of changing the tone. In the above steps, an offset of 5 < 0 indicates that the tone of the output sound is increased, (5 > 0 indicates that the tone is decreased, and the calculation of the first index ρ 1 in step 5 is changed.

Page 8 200412570

This case is technical in terms of: θ has at least the following characteristics and improvements1. Only one time frame is used to temporarily store voice data, which can save the use of memorization); and the weight w [n] 'in the third figure is due to symmetry Relationship; as long as w [n], n = 0, ..., (N / 2) —i, these (N / 2) weights can be stored in read-only memory (ROM) ), And then take 杳… ten =… cost ⑽—benyuan M, • Calculate #new 50% value phen, just calculate the first index M and the second (as in step A), and then re-evaluate the entire time frame :: r: rr Hours. (real ―) The purpose of changing the tone of the sound ... (·; I. The characteristics of the weight function fUnCtl〇n) such as the solid (TriangUlar Window) transformation, the synthesized voice data car ^; It will be generated due to discontinuities in data, differences ^ L., Smoothing. Explosion 3, so no further smoothing is required. The technology disclosed in this case was unprecedented the next day; hot learning * technology Chuan Kang implemented, please, apply for the scope of patents such as ten years with patents, and Mai filed a patent application according to law

200412570 Schematic illustration of the case The following figure and detailed description can be used to gain a deeper understanding: Figure 1: System block diagram of the preferred embodiment of the method of changing the tone of the sound in the case. Figure 2: When the sound data Frame (third): Corresponding to the weighting of the frame data in the second image, the figure numbers of the main components are as follows:

1 1: Analog / digital converter (ADC) 1 2: Tone change algorithm 13: Digital / analog converter (DAC) SR: Sampling frequency N: Length of time frame X [〇], ..., X [N-1 ]: Data in the time frame p 1: First index p2: Second index δ: Offsets w [pl], w [p2]: Weighting

Page 10

Claims (1)

200412570 6. Scope of patent application 1. A method for changing the tone of a sound, including the following steps: providing a time frame; sequentially storing a digital data into the time frame with a sampling frequency; and synthesizing a synthesis with the sampling frequency The value is converted into an analog signal output, where the composite value is obtained from a first indicator and a second indicator from the time frame, and is obtained after a weighting operation, and the first indicator and the first indicator The two indicators change in response to an offset. 2. The method for changing the tone of a sound as described in item 1 of the scope of the patent application, wherein the time frame is a sound data time frame (f r a m e). 3. The method for changing the tone of a sound as described in item 1 of the scope of the patent application, wherein the sampling frequency is an analog to digital sampling frequency. 4. The method for changing the tone of a sound as described in item 1 of the scope of the patent application, wherein the sampling frequency is a digital frequency converted to an analog sampling frequency. 5. The method of changing the tone of a sound as described in item 1 of the scope of the patent application, wherein the digital data is sequentially and repeatedly stored in the time frame. 6. The method for changing the tone of a sound as described in item 1 of the scope of the patent application, wherein the height of the tone of the sound can be adjusted through the change of the offset 5, wherein 5 > 0 means that the tone of the output sound is increased, 5 < 0 Indicates that the tone is decreasing, and 5 = 0 does not change the tone. 7. The method for changing the tone of a sound as described in item 6 of the scope of the patent application, wherein the first index pl = pl + 5 is set every time data is taken from the time frame in response to the change of the offset 5. Let the second index p2 = pl + (N / 2), where N is the length of the time frame. 8. The method for changing the tone of a sound as described in item 7 of the scope of patent application, which 200412570 6. In the scope of patent application, in response to the first index p 1 > (N-1), let p 1 = p BU N, where N is the length of the time frame. 9. The method for changing the tone of a sound as described in item 7 of the scope of the patent application, where in response to the first index p 1 < 0, let p 1 = p 1 + N, where N is the length of the time frame. 10. The method for changing the tone of a sound as described in item 7 of the scope of the patent application, wherein in response to the second index p2 > (N-1), let p2 = p2-N, where N is the length of the time frame. 11 1. The method for changing the tone of a sound as described in item 7 of the scope of the patent application, wherein in response to the second index p 2 < 0, let p 2 = p 2 + N, where N is the length of the time frame. 1 2. The method for changing the tone of a sound as described in item 1 of the scope of the patent application, wherein the weight calculation is a Triangular Window conversion. Page 12