544654 ί ^ ! C7 D7 五、創作説明（/) · 發明背景 1. 發明類別 本發明與語音合成的方法，如脈沖調制碼（P^seCode Modulation，簡稱 PCM )，調適性脈沖調制碼（Adaptive Differential Pulse Code modulation 簡稱 ADPCM )，線性預估碼（Linear Predition Code 簡稱 LPC )等，時域（time domain )或頻域（frequency domain) 組碼，及雜訊消除方法有所關連· 2. 先前的技藝 一些與聲音有關並附有喇叭的電子裝置，常常會、在開 機時產生，’嘟”，的一聲雜訊，如何消除這個雜訊一庫是技 術人員的目標. 曰本昭和58年（ 1983 )4月26日公開之昭58-70297號”聲音 輸出方式”專利公開案中，說其專利的特徵在於，聲音合成 積体電路輸出聲音訊號時·於該積体電路所輸出之聲音訊 號輸出波形前後，加入一自邏輯準位至中心準位之具有特 定斜率之波形，及加入一自該中心準位至該邏輯準位之具 有特定斜率之波形·但昭58-70297號專利並未告知其斜率的 範圍，及此特定斜率的波形，自邏輯位準至中心位準需要 多久時間（時間才是真正的關鍵），才能有消除”卡答”聲的 效果. 同樣中華民國專利第78101301號π語音合成系統消除瞬 間雜音之編碼方法”，其專利範圍也是在語音訊號開始時加 入一段平緩碼，從0開始至靜音位階·並在語音訊號結束 再加入一段從靜音位階至〇之平緩碼，以有效消除電流瞬 _ / __ 本紙伕尺度適用中國國家梂準（CNS ) Α4規格（210 X 297公釐） (請先閲讀背面之注意事項再填寫本頁)544654 ί ^! C7 D7 V. Creation instructions (/) · Background of the invention 1. Category of the invention The method of the present invention and speech synthesis, such as pulse modulation code (P ^ seCode Modulation (PCM), adaptive pulse modulation code (Adaptive Differential Pulse Code modulation (ADPCM for short), Linear Predition Code (LPC for short), etc., the time domain (frequency domain) or frequency domain (frequency domain) group codes, and noise cancellation methods are related. 2. Previous techniques Some electronic devices related to sound and with a horn are often produced when the machine is turned on, a beep, and how to eliminate this noise and a bank is the goal of the technical staff. This year, Showa 58 (1983 ) In the “Sound Output Method” patent publication No. 58-70297 published on April 26, the patent is characterized in that when a sound synthesis integrated circuit outputs a sound signal, the sound signal output by the integrated circuit is output. Before and after the waveform, a waveform with a specific slope from the logical level to the center level is added, and a waveform with a specific slope from the center level to the logic level is added. Waveform · But Zhao 58-70297 patent does not tell the range of its slope and the waveform of this specific slope. How long does it take from the logic level to the center level (time is the real key) to eliminate it? "The effect of sound. Similarly, the Republic of China Patent No. 78101301 π speech synthesis system coding method to eliminate transient noise", the scope of its patent is to add a smooth code at the beginning of the speech signal, starting from 0 to the mute level and ending the speech signal Then add a gentle code from mute level to 0 to effectively eliminate the current transient. _ / __ The paper size is applicable to China National Standard (CNS) Α4 size (210 X 297 mm) (Please read the precautions on the back before filling (This page)
、1T *曹 經濟部中夬榡準局員工消費合作杜印製 544654 經濟部中夬樣準局員工消費合作社印製 C7 D7 五、創作説明（2) · 間變化所可能導致之雜音. 專利第78101301號，除了說這組碼可以消除雜音外，並 未說明其可使用的範圍，如需要輸入多久時間，才有消去 雜訊的功效，及該碼的特徵.因爲該碼決不可能適用於所 有頻率範圍，如果用在取樣及輸出頻率6khz時，在8khz時 就不一定能用.同樣該碼只是一 6位元碼如適用於6位元 的訊號·在8位元時就不能用.同時該碼只能用在脈碼調 制法中. 本發明與上面所說的兩種方法不一樣，乃是以生理學 ，語音學，數學，爲基礎發明出一個解決因電壓突然ΐ文變 ，所引起雜訊的方法. > 表說明 表1.自0位準至l/2Vd，6khz取樣，1/4週期餘弦波，十進位 ，二進位數値表. 表2·自0位準至1/2 Vd，8 khz取樣，1/4週期餘弦波，十進位 ，二進位數値表. 表3.自0位準至l/2Vd，6khz取樣，1/4週期拋物線波，十進 位，二進位數値表. 圖式說明 圖1.自〇位準至l/2Vd，6khz取樣，1/4週期餘弦波，波形圖· 圖2.自0位準至l/2Vd，8khz取樣，1/4週期餘弦波，波形圖· 圖3.自0位準至1/2 Vd，6khz取樣，1/4週期拋物線波，波形圖· 圖4.自l/2Vd至0位準，6khz取樣，1/4週期餘弦波，波形圖. 圖5. 6khz取樣，聲音尾音，波形圖. I------'—·裝-----—1T----- (請先閔讀背面之注意事項再填寫本頁) 本紙張尺度適用中國國家梂準（CNS) A4規格（210X29*7公釐） 5446541T * Printed by the Consumers 'Cooperative Bureau of the Ministry of Economic Affairs of the Ministry of Economic Affairs of the People's Republic of China Printed 544654 Printed by the Consumers' Cooperative of the State Council of the Ministry of Economy of the People's Republic of China Printed by C7 D7 V. Creation Instructions (2) Noise that may be caused by changes in time. No. 78101301, in addition to saying that this group of codes can eliminate noise, it does not explain its range of use, such as how long it takes to enter the effect of eliminating noise, and the characteristics of the code. Because this code can never be applied to All frequency ranges, if used at the sampling and output frequency of 6khz, may not be available at 8khz. The same code is only a 6-bit code, which is suitable for 6-bit signals. It cannot be used at 8-bit. At the same time, the code can only be used in the pulse code modulation method. The present invention is different from the two methods mentioned above. It is based on physiology, phonetics, and mathematics. Methods of Noise Caused. ≫ Table Description Table 1. Table from 0 level to l / 2Vd, 6khz sampling, 1/4 period cosine wave, decimal, binary number. Table 2. From 0 level to 1/2 Vd, 8 khz sampling, 1/4 period cosine wave, decimal, two Number of digits table. Table 3. From 0 level to l / 2Vd, 6khz sampling, 1/4 period parabolic wave, decimal, binary digits table. Schematic description Figure 1. From 0 level to l / 2Vd , 6khz sampling, 1/4 period cosine wave, waveform diagram. Figure 2. From 0 level to l / 2Vd, 8khz sampling, 1/4 period cosine wave, waveform diagram. Figure 3. From 0 level to 1/2. Vd, 6khz sampling, 1/4 period parabolic wave, waveform diagram. Figure 4. From l / 2Vd to 0 level, 6khz sampling, 1/4 period cosine wave, waveform diagram. Figure 5. 6khz sampling, voice tail, waveform Fig. I ------'— · Installation -----— 1T ----- (Please read the notes on the back before filling out this page) This paper size is applicable to China National Standards (CNS) A4 size (210X29 * 7mm) 544654
C7 D7 五、創作説明（3) ’ 圖6. 6 khz取樣，自1/2 Vd至0位準，1/4週期三角波+尾音， 波形圖. 發明槪要 ' 一個附有揚聲器的電子裝置中，在電源開啓時，常常 會因電位的突變，而發出"嘟”的一聲雜聲.同樣在關機或 訊號電位突然改變時，也會產生〃嘟”的一聲雜聲. 本發明乃是以生理學，語音學，數學，爲基礎.發明 出一個解決瞬間雜音的方法.本發明所提出的解決方法是 在電源開啓時，聲音訊號輸入揚聲器前，立輸入一自〇電 位到二分之一工作電位的低頻訊號.同樣聲音結朿後也輸 入一從二分之一工作電位到0電位的低頻訊號.這類低頻 訊號，按本發明所提的法則，很容昜求得.本發明更進一 步的，將原始訊號和低頻訊號波合而爲一，如此不但可消 除雜音，也不會因儲存低頻訊號而消耗部分記憶体. 詳細說明 一般而言，聲音的大小，與聲音振幅有關.而音調的 高低則與頻率有關.振幅越大聲音越大，頻率越大，音調 越高.人類能聽見的聲音頻率自20到20000週/秒（hz). 經濟部中夬標隼局員工消費合作社印策 (請先閲讀背面之注意事項再填寫本頁) 聲音自聲學訊號經微音器，轉換成電子訊號時，如果 工作電位是Vd，則聲音以Vd/2爲基準，在0與Vd之間振 動（聲音開始及結朿都在Vd/2或接近Vd/2處）.這個訊號稱 爲類比訊號，一個類比訊號可以在短時間內取樣，每一個 取樣訊號的振幅可以量化成有限的位階.將類比訊號轉換 成數位訊號的技術，己經是一種熟知的技巧.轉換時，每 本紙伕尺度適用中國國家梂準（CNS ) A4说格（210 X 297公釐） 544654C7 D7 V. Creation instructions (3) 'Figure 6. 6 khz sampling, from 1/2 Vd to 0 level, 1/4 period triangle wave + tail, waveform diagram. Invention summary' In an electronic device with speakers "When the power is turned on, a" beep "noise is often issued due to a sudden change in potential. Similarly, a" beep "sound is also generated when the power is turned off or the signal potential changes suddenly. The present invention is Based on physiology, phonetics, and mathematics. Invented a method to solve transient noise. The solution proposed by the present invention is to input a signal from 0 potential to two points immediately before the sound signal is input to the speaker when the power is turned on. One of the low-frequency signals of the working potential. After the same sound, a low-frequency signal from one-half of the working potential to 0 potential is also input. Such low-frequency signals are very easy to obtain according to the rules of the present invention. The invention further took the original signal and the low-frequency signal into one, which not only eliminates noise, but also does not consume part of the memory due to the storage of low-frequency signals. Detailed description Generally speaking, the size of the sound is related to the amplitude of the sound . The pitch is related to frequency. The greater the amplitude, the louder the frequency, the greater the frequency, the higher the pitch. The frequency of human audible sounds is from 20 to 20000 cycles / second (hz). Yin Ce (please read the notes on the back before filling this page). When the acoustic signal is converted into an electronic signal by a microphone, if the working potential is Vd, the sound is based on Vd / 2, between 0 and Vd. Inter-vibration (the beginning and end of the sound are at or near Vd / 2). This signal is called an analog signal. An analog signal can be sampled in a short period of time. The amplitude of each sampled signal can be quantified to a limited level. . The technology of converting analog signals into digital signals has been a well-known technique. At the time of conversion, each paper size applies the China National Standard (CNS) A4 format (210 X 297 mm) 544654
C7 D7 經濟部中央標準局員工消費合作社印袈 五、創作説明) 一數位訊號對應於一聲音樣本的振幅·這種將類比轉換成 數ίϋ的方法有很多種.如脈沖調制碼（p^lse Code Modulation， 簡稱 PCM)，調適性脈沖調制碼（Adaptive Differential Pulse Code Mo dulation簡稱ADPCM)，三角調制法（Delta Modulation)等，這些調 制法也爲人所熟悉· 欲將儲存在電子裝置內的聲音訊號重現時，當電源開 啓後，訊號突然自0跳至Vd/2，產生一隨機瞬間電流，輸 入揚聲器中，產生〃嘟” 一聲雜訊·如何消除此聲雜訊，從 生理學言，有二種方法，一種是電源開啓時，立刻輸入一 •人耳所聽不到的高頻，即高於20k(lk== 1000)hz的頻率··另一 個方法，則是立即輸入一人耳所聽不到的低頻訊號，即小 於20hz的訊號，一般揚聲器，在低頻的頻率反應很差，30 hz幾乎聽不見（助聽器的頻率在400-4khz)，這也晏本發明重點 的所在. 傅氏級數（Fourier series )的理論，如杲一個三角級數對所 有的X値收歛，它的和f(x)，是一個以2兀爲週期的函數· f(x)= aO/2 + ( alcosx + blsinx) + ( a2cos2x + b2sin2x) +............ (1) 爲了降低其他諧波所產生的雜音，本發明先輸入一個振 幅自〇到Vd/2的餘弦波，如果聲音訊號是6khz取樣，6位元表 示自0到Vd的電位差，代入公式(2)及(3)可得到表1，及圖1的 輸入波.（1/4是輸入1/4週期） 1 fs诹樣頻率） N阳樣數)------------------ (2) 4 fe认耳聽不見的頻率） - " 备 - · Λ I _— ____ 本紙杀尺度適用中國國家梂準（CNS ) Α4規格（210 X 297公釐） ---------φ袭-----—1Τ----- (請先閲讀背面之注意事項再填寫本頁) 544654 C7 D7 五、創作説明（5)— 設經揚聲器後，人耳聽不見的頻率爲25hz，代入(2)式. fs=6⑻0 fe=25 得 N=60 y=A-Acos( 1.57 x t/T) T^s x N 7 t=fs x n y=A-Acos( 1.57 xn/N) 0< n < 60 (3) (3)式中A=kcxVd/2，kc是比率常數，T是輸入低頻訊號所 需時間，在6位元中設A=32，取y的整數値並將y轉換 成二進位，則得到一餘弦的低頻訊號.見表1.圖1.從(2) 知如輸入頻率fe太低，則N就會很大，這樣會浪費記憶体 ，因此輸入頻率不必過低. ， > 聲音訊號如杲是8khz取樣，輸入低頻25hz訊號波經 (2)，（3)式的運算· N=80，並可得到表2,圖2.的結果.表1 . ，表2.數據雖不一樣_，但是是同樣的一個訊號波（頻率是25 hz的餘弦波）.如果自0到Vd的電位差以8位元表示，同樣 可得到數據不同，可是訊號卻是一樣的餘弦波. 經濟部中央標準局員工消費合作社印製 (請先閲讀背面之注意事項再填寫本頁) 現將表1.的二進位資料或其變体，存入聲音儲存及再 生裝置中，電源開啓後先輸入此低頻率訊號，再輸入聲音 訊號，到揚聲器，則聽不到任何因電源開啓所引起的雜訊 .同理以6khz取樣，週期是25hz軌跡是拋弧線的訊號. y=A(t / T) =A(n/N)2 0 < η < 60 (4) 把(3)中所用的資料代入(4)，得到表3.及圖3.同樣將表3的 二進位資料或其變体，存入聲音儲存及再生裝置中，電源 開啓後先輸人此低頻率訊號，再輸入聲音訊號，到揚聲器 厂 本紙張尺度適用中國國家梂準（CNS ) Α4規格（210 X 2W公釐） 5446C7 D7 Printed by the Consumer Cooperatives of the Central Standards Bureau, Ministry of Economic Affairs 5. Creation instructions) A digital signal corresponds to the amplitude of a sound sample. There are many ways to convert an analog to a number. For example, pulse modulation code (p ^ lse Code Modulation (abbreviated as PCM), Adaptive Differential Pulse Code Modulation (ADPCM), Delta Modulation, etc. These modulation methods are also well-known. I want to store sound signals in electronic devices. During reproduction, when the power is turned on, the signal suddenly jumps from 0 to Vd / 2, generating a random instantaneous current, which is input into the speaker, producing a beeping sound. "How to eliminate this noise. From physiological doctrine, There are two methods. One is to input a high frequency that the human ear cannot hear immediately when the power is turned on, that is, a frequency higher than 20k (lk == 1000) hz. The other method is to input one human ear immediately. Inaudible low-frequency signals, that is, signals less than 20hz, general speakers respond poorly at low frequencies, and are almost inaudible at 30hz (hearing aid frequency is 400-4khz). The point is. The theory of Fourier series, such as 杲 a triangle series converges on all X 値, and its sum f (x) is a function with a period of 2 兀 f (x) = aO / 2 + (alcosx + blsinx) + (a2cos2x + b2sin2x) + ............ (1) In order to reduce the noise generated by other harmonics, the present invention first inputs an amplitude from 〇 The cosine wave to Vd / 2, if the sound signal is 6khz samples, the 6 bits represent the potential difference from 0 to Vd. Substituting into equations (2) and (3) can get Table 1 and the input wave of Figure 1. (1 / 4 is the input 1/4 period) 1 fs sample frequency) N number of positive samples) ------------------ (2) 4 fe inaudible frequency)- " Prepare-· Λ I _— ____ The paper size is applicable to China National Standard (CNS) A4 specification (210 X 297 mm) --------- φ 袭 -----— 1Τ-- --- (Please read the precautions on the back before filling this page) 544654 C7 D7 V. Creative Instructions (5) — After setting up the speaker, the frequency that the human ear can't hear is 25hz, which is substituted into (2). Fs = 6⑻0 fe = 25 gives N = 60 y = A-Acos (1.57 xt / T) T ^ sx N 7 t = fs xny = A-Acos (1.57 xn / N) 0 < n < 60 (3) (3) Where A = kcxVd / 2, kc is the ratio constant, and T is the time required to input the low-frequency signal. Set A = 32 in 6 bits, take the integer y of y and convert y to binary, and get a cosine low-frequency The signal. See Table 1. Figure 1. From (2), if the input frequency fe is too low, N will be very large, which will waste memory, so the input frequency does not have to be too low. ≫ The sound signal such as 杲 is 8khz Sampling, input the low-frequency 25hz signal wave through the operation of formula (2), (3) · N = 80, and can get the results of Table 2, Figure 2. Table 1.., Table 2. Although the data is not the same _, but The same signal wave (frequency is a cosine wave of 25 hz). If the potential difference from 0 to Vd is expressed in 8 bits, the same data can be obtained, but the signal is the same cosine wave. Employees of the Central Standards Bureau of the Ministry of Economic Affairs Printed by the cooperative (please read the notes on the back before filling this page) The binary data of Table 1. or its variants are stored in the sound storage and reproduction device. After the power is turned on, input this low-frequency signal first, and then Input the sound signal and go to the speaker, you will not hear any noise caused by the power on. Similarly, take 6khz Similarly, the period is a signal at 25hz and the trajectory is a parabola. Y = A (t / T) = A (n / N) 2 0 < η < 60 (4) Substitute the data used in (3) into (4) , Get Table 3. and Figure 3. Also store the binary data of Table 3 or its variant into the sound storage and reproduction device. After the power is turned on, input this low-frequency signal first, then input the sound signal to the speaker factory This paper size applies to China National Standard (CNS) Α4 size (210 X 2W mm) 5446
I ^"1 27 月 C7 D7 經濟部中夬標準局員工消費合作社印製 五、創作説明（6) ，一樣聽不到任何因電源開啓所引起的雜音· 同理在聲音儲存及再生糸統中，在聲音訊號開始輸出 前，以低於30hz的其他訊號，如三角波，圓弧形波（方形 波除外）輸入.等都可消去”嘟"聲雜音“ 當聲音結束時，聲音訊號多停留在1/2糾處·如果關 閉電源，或聲音訊號自1/2 Vd處，突掉至0位準時，聲音儲’ 存及再生裝置，一樣會產生”嘟”的一聲雜音，其解決方 法和上面所述一樣，只是改爲自1/2 Vd處，向0電位輸入一 低頻率訊號.見圖4 . 聲音儲存及再生糸統中，如何以有限的記億体，儲存 最多的資料，一直是技術人員所追求的目標·爲了消除電： 源開啓，或關閉時所生的雜音，預先儲存一段低頻訊號在 記憶体中、多少會佔用一些記憶体，影響資料的儲存.本ί 發明則將低頻訊號和原始訊號合而爲一，如此不但可消除 雜音，還可保有原來儲存資料的空間·其運算公式如(5):: se(x) = s⑻-l/2Vd + sl(x) (5) se(X):最終訊號 s⑻：原始訊號 sl(x):低頻訊號 圖5·是某聲音號訊號的結尾，圖6.則是圖5·的訊號按公式 (5)和一低頻（25hz)自1/2 Vd至0電位的三角波所合組的訊號 •該方法在聲音開始及結尾爲輕音（unvoice)時特別有效· 綜合以上所述，本發明可簡化成一數理法則，其規則 如下： 本紙張尺度逋用中國國家揉準（CNS ) A4規格（210 X 297公釐） (請先閲讀背面之注意事項再填寫本頁) 訂 •會 544654 月 經濟部中央揉隼局員工消費合作社印製 C1 D7 五、創作説明（7 ) ‘ 1. 決定聲音儲存及再生糸統中，聲音的取樣頻率fs · 2. 從揚聲器的頻率反應中決定人耳所聽不見的低頻fe . 3. 決定自0電位至1/2工作電位，或自1/2工作電位至0電位 的低頻訊號，及波形. 4. 從步驟1.及步驟2.依公式(2)算出低頻訊號所需的時間T. (時間T = N X fs N是取樣點） 5. 將n=l，2,3..............N値代入低頻訊號，求出相對應的振幅 ，或軌跡.並將這些値轉換成二進位，存入聲音儲存及再 生糸統中. 6. 在聲音儲存及再生糸統中.聲音訊號輸出前或結束後， 輸出這些二進位數値，經數位對類比轉換器（Digitaltoana logy Coavter簡稱DAC)，轉換成低頻訊號輸出.如此消除掉 -訊號突然自低電位跳至高電位，或由高電位突降至低電 位所產生的雜音· 本發明所敘述的這種在訊號儲存及再生糸統中·消去雜 音的法則，不僅適用於聲音處理方法中的時域(timedomain)， 也適用於頻域（frequency domain). 本發明所敘述的這種在訊號儲存及再生糸統中.消去 雜音的法則，不僅適用於DAC中.也適用於脈衝寬度調變 法（Pulse Width Modulation 簡稱 PWM)中. 本發明所敘述的這種在聲音儲存及再生糸統中·消去 雜音的法則，不僅適用於預先將低頻訊號存在記憶体中· 也可應用在用微電腦，隨時計算低頻訊號軌跡的即時糸統 (realtime system)中· f ------1T------ (請先閲讀背面之注意事項再填寫本頁) 本纸伕尺度適用中國國家梂準（CNS ) Α4規格（210Χ297公釐） 544854 π 终习 ;…：丨 C7 r...... D7 -五、創作説明（g ) 本發明所敘述的這種在聲音儲存及再生糸統中.消去 雜音的法則，不僅適用於聲音儲存及再生糸統中，也可應 用在聲音傳輸糸統中.如電傳會議. 語音合成積体電路，就是一種聲音儲存及再生系統，因 此這種預存一人耳聽不見的低頻訊號，以消除雜訊的方法 可應用在語音合成積體電路中. (請先Η讀背面之注意事項再填寫本頁} 經濟部中夬標準局員工消費合作社印製 P0 本紙法尺度適用中國國家梂準（CNS ) Α4洗格（210Χ 297公釐）I ^ " 1 Printed on July 27 C7 D7 Printed by the Consumer Cooperatives of the China Standards Bureau of the Ministry of Economic Affairs 5. Creative Instructions (6), I still ca n’t hear any noise caused by power on. Similarly, in the sound storage and reproduction system In the beginning, before the sound signal starts to output, input with other signals lower than 30hz, such as triangle wave, arc wave (except for square wave). You can eliminate the "beep" sound noise. When the sound is over, the sound signal is more Stay at 1/2 correction. If the power is turned off, or the sound signal suddenly drops to 0 level from 1/2 Vd, the sound storage and playback device will also generate a “beep” noise, which will solve the problem. The method is the same as the above, except that a low-frequency signal is input to the 0 potential from 1/2 Vd. See Figure 4. In the sound storage and reproduction system, how to store the most data with limited memory , Has always been the goal pursued by technical staff. In order to eliminate electricity: noise generated when the source is turned on or off, a low-frequency signal is stored in memory in advance, which will take up some memory and affect the storage of data. This invention Low frequency The signal and the original signal are merged into one, so not only can eliminate noise, but also keep the original data storage space. Its calculation formula is (5) :: se (x) = s⑻-l / 2Vd + sl (x) (5 ) se (X): final signal s⑻: original signal sl (x): low-frequency signal Figure 5. · is the end of a sound signal, Figure 6. is the signal of Figure 5 · according to formula (5) and a low-frequency (25hz ) Signals combined by triangle waves from 1/2 Vd to 0 potential. This method is particularly effective when the beginning and end of the sound are unvoice. Based on the above, the present invention can be simplified into a mathematical rule, the rules are as follows : This paper size is in accordance with China National Standard (CNS) A4 (210 X 297 mm) (Please read the notes on the back before filling out this page). System C1 D7 V. Creative Instructions (7) '1. Determine the sampling frequency fs of the sound in the sound storage and reproduction system. 2. Determine the low-frequency fe inaudible to the human ear from the frequency response of the speaker. 0 potential to 1/2 working potential, or low frequency signal from 1/2 working potential to 0 potential, and waveform 4. Calculate the time T required for the low-frequency signal from step 1. and step 2. according to formula (2) (time T = NX fs N is the sampling point) 5. Set n = l, 2, 3 .... ......... N 値 Substitute low frequency signals, find the corresponding amplitude, or trajectory. Convert these 値 into binary, and store them in the sound storage and reproduction system. 6. In the sound storage In the reproduction system, these binary digits are output before or after the audio signal is output, and then converted to a low-frequency signal output by a digital analog converter (Digitaltoanalogy Coavter for short). In this way, the signal suddenly disappears from a low potential. Noise generated by jumping to a high potential or suddenly falling from a high potential to a low potential · The principle of eliminating noise in the signal storage and reproduction system described in the present invention is not only applicable to the time domain in sound processing methods ( time domain), also applicable to the frequency domain (frequency domain). This invention described in the signal storage and reproduction system. The rule of noise reduction is not only applicable to DAC. It is also applicable to Pulse Width Modulation (Pulse Width Modulation) Modulation (PWM for short). This kind of sound described in the present invention The principle of saving and reproducing noise in the system is not only suitable for storing low-frequency signals in memory in advance, it can also be applied to a real-time system using a microcomputer to calculate the low-frequency signal track at any time. F --- --- 1T ------ (Please read the notes on the back before filling this page) The paper size is applicable to China National Standard (CNS) A4 specification (210 × 297 mm) 544854 π C7 r ...... D7-V. Creation instructions (g) The sound storage and reproduction system described in the present invention. The rule of eliminating noise is not only applicable to the sound storage and reproduction system, but also Can be used in voice transmission systems. For example, teleconference. The speech synthesis integrated circuit is a sound storage and reproduction system. Therefore, this method stores a low-frequency signal that is inaudible to one's ears. The method of eliminating noise can be applied to speech. Synthetic integrated circuit. (Please read the precautions on the back before filling out this page} Printed by the Consumers' Cooperative of the China Standards Bureau of the Ministry of Economic Affairs P0 This paper method is applicable to China National Standards (CNS) %)