525355 A7 B7 6337-1-0521 修.doc-3/30 經濟部智慧財產局員工消費合作社印製 五、發明說明（1 ) 本發明係有關一種類比對數位訊號之調變方法，尤指 一種以二次差分方法將類比訊號轉換成數位訊號者，該方 法係先對輸入之類比訊號取樣得數位訊號脈衝，將該數位 訊號脈衝與前次量化還原之數位訊號脈衝相減得一次差値 5 :再將該一次差値與還原器（Predictor)輸出之前次量化 還原差値相減，以取得二次差値；將該二次差値經量化器 (Quantizer)量化後得一壓縮之量化二次差値，以利傳 送或儲存；再利用還原器（Predictor)將前次之量化還原 差値與本次之量化二次差値相加得本次量化還原差値；最 10後再利用還原器（Predictor)將本次量化還原差値與前次 量化還原之數位訊號相加，即得本次量化還原之數位訊號 ，藉此方法調變所得之數位訊號其雜訊比及訊號失真最小 ，並具壓縮比最大而最簡省儲存空間等優點者。 按，習知用以將類比訊號轉換爲數位訊號之調變技術 15之一者，如1940年代所發展並應用於電話語音方面之一 次差分調變法（Delta Modulation，簡稱DM)，如第1 圖所不，該一次差分調變法係利用一加法器30、一量化 器40及一積分器50來達成將類比轉換成數位訊號之目的 。其係將類比輸入訊號Si與經由積分器50輸出之前次量 2〇化還原之數位訊號QSi-Ι於加法器30內相減得一差値Di ，該差値Di經量化器4〇後產生一量化差値QDi，作爲傳 送或儲存之訊號；該量化之差値QDi可回饋至積分器5〇 並還原成本次量化還原之數位訊號QSi，作爲下一次取樣 運算之參數値。易言之，該一次差分調變法（DM)係·一 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） (請先閱讀背面之注意事項再填寫本頁) -訂il 線麵 525355 A7 B7 6337-1-0521 修.doc-4/30 五、發明說明（2 ) 種將訊號差分量化的技術，即以訊號振幅之差値編碼替代 傳統直接以振幅編碼之方式，因此一次差分調變（DM) 又稱爲脈衝寬度調變（PWM)。當取樣頻率非常高（即 取樣週期很短）時，則取樣週期間之訊號變化就非常的小 5 ，其量化差値可壓縮至以極少位元數即能充份表現。一_ 以單一位元編碼之一次差分調變（1-bit Converters) ’ 如第2圖所示，以頻率1.5KHZ、最大振幅爲±1之正弦波 爲輸入訊號，其取樣訊號差値選擇0.125，經量化壓縮对 以1位元代表4量化位元之比，若以4個位元編碼，即对 10得24=16種量化値。則第2圖顯示以32倍取樣脈波訊號 C進行調變時，該積分器之輸出訊號D (即調變後之數位 訊號）很接近輸入之類比訊號A ;若以16倍取樣脈波_ 號C進行調變時（如第3圖所示），該積分器之輸出數位 訊號D將遠離輸入訊號A (代表訊號轉換之誤差較大）； 15若差値或取樣訊號比太低，則輸出訊號波形D將呈較大斜 率之傾斜狀且遠遠落後於輸入訊號A (如第4圖所示）’ 而產生訊號失真及訊號誤差。亦即利用該一次差分技術將 類比訊號轉換爲數位訊號，雖可以單一位元完成訊號之編 碼以節省硬體設備費用，但因其係採用取樣之振幅相減所 20得之一次差分値量化而成，其斜率固定（如第5A至5C 圖所示，其差値分別爲d=l，d=0及d=— 1 )故其量化# 差較大，由第2至4圖可知取樣時間越長（取樣頻率越低 )或取樣差値越小則產生之失真越嚴重，而使數位化後之 訊號與原類比訊號之誤差越大。因此，一次差分調變法（ -4- 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） (請先閲讀背面Μ涑意事項再碘寫本 iri,-------線 經濟部智慧財產局員工消費合作社印製 525355 A7 B7 6337-1-0521 修.doc-5/30 經濟部智慧財產局員工消費合作社印製 五、發明說明（3) DM)雖可利用單一位元完成訊號之編碼而將類比訊號轉 換成數位訊號，使其硬體設備成本較低，但訊號轉換過程 失値相當大，轉換效果並不理想，若要提高調變效果（降 低失真率），則需相對提高取樣頻率，故其整體設備成本 5仍過局。 另二種習知之類比對數位訊號之調變技術分別爲脈衝 編碼調變法（Pulse Code Modulation，簡稱PCM)及差 分脈衝編碼調變法（Differential Pulse Code Modulation，簡稱DPCM)，該脈衝編碼調變法（PCM 10 )係對類比訊號之取樣値量化編碼而成數位訊號；而差分 脈衝編碼調變法（DPCM)則爲前述一次差分DM技術與 脈衝編碼調變法（PCM)之綜合應用，即先對類比訊號取 樣，再求出實際取樣値與前次取樣値之差値，最後經量化 及編碼而成數位訊號。利用連續取樣訊號之取樣差値之集 15中性，可以較少之位元完成編碼。如第6A圖所示，爲以 脈衝編碼調變法（PCM)轉換類比聲音訊號爲數位訊號之 影像圖形，該圖顯示轉換所得之取樣値較集中分佈在20 至160之間，其訊號範圍包括160 - 20= 140，若以二進 位位元表示0至140値之範圍，則需至少8個位元（ 2〇 27 = 128 <140< 28=256 )，故其無法以較短之字元碼完 成訊號之編碼’即其壓縮比較小；反之，如第6B圖所示 ，以DPCM技術將類比訊號編碼轉換成數位訊號，所得 之取樣差値集中分佈在-20至20之間，其訊號範圍包括 20— ( —20) =40，因此以二進位位元表示〇至40範圍 (請先閱讀背面之注咅？事項再填寫本頁) 1---¾ Γ 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） 525355 A7 B7 6337-1_0521修.〇1〇(：-6/30 五、發明說明（4 ) 內之數位資料，只需6個位元（25 = 32<40<26 = 64 )， 故可以較短之字元碼完成訊號之編碼，其壓縮比較大。比 較第6A及6B圖得差分脈衝編碼調變法（DPCM)較脈 衝編碼調變法（PCM)更爲節省儲存設備之空間及降低硬 5體設備成本。 然而，該差分脈衝編碼調變法（DPCM)雖係改良脈 衝編碼調變法（PCM)而成，具有較節硬體設備省成本之 優點，惟其成本仍屬偏高因而無法廣泛應用於一般平價商 品（如較廉價之有聲玩具）。 10 本發明即爲克服習知類比對數位轉換技術之高成本及 局失真之缺點硏發而成者，其具有價格便宜及調變效果更 好之優點者。 易言之，本發明之主要目的在於提供一種類比訊號轉 換成數位訊號之新方法，其較換之誤差及訊號失真極小， I5可獲得良好之訊號轉換品質者。 本發明之次一目的在於提供一種類比訊號轉換成數位 訊號之新方法，因其量化後之訊號非常集中，故可以較少 之位元完成編碼，而能降低成本者。 經濟部智慧財產局員工消費合作社印製 -----------—— (請先閱讀背面之注意事項再填寫本頁) 本發明之上述及其他目的、功效及技術內容請參閱後 2〇附圖式及說明以獲致進一步之瞭解。 圖式說明如下： 第1圖係習知一次差分調變法（dm)之電路方塊圖 ， 第2圖係一次差分調變法以32倍取樣脈波訊號進行調 -6- 本紙張尺度賴巾® 0家標準（CNS)A4規格（210 X 297公f ) 525355 A7 B7 6337-1-0521 修.doc - 7/30 五、發明說明（5 ) 變時所得之類比與數位訊號關係圖； 第3圖係一次差分調變法以16倍取樣脈波訊號進行調 變時所得之類比與數位訊號關係圖； 第4圖係一次差分調變法以16倍取樣脈波訊號且其取 5 樣差値爲第2圖之一半進行調變時所得之類比 與數位訊號關係圖； 第5 A至5 C圖係一次差分調變法之三種固定斜率示 意圖； 第6 A圖係以脈衝編碼調變法（PCM)所得之數位訊 10 號分佈示影像圖； 第6 B圖係以差分脈衝編碼調變法（DPCM)所得之 數位訊號分佈影像圖； 第7圖係本發明之電路方塊圖； 第8圖係本發明二次差分調變法之訊號取樣示意圖； 15 第^係本發明二次差分調變法之三種變動斜率示意 圖； (請先閱讀背面之注意事項再填寫本頁) 裝---- 經濟部智慧財產局員工消費合作社印制衣 20 第1 0圖係本發明之調變程序方塊圖； 第11圖係本發明之調變方法流程圖； 第12A圖係以本發明之二次差分調變方法所得之數 位訊號分佈之影像圖； 第1 2 B圖係以本發明之二次差分調變方法所得之數 位訊號分佈示意圖； 第13圖係本發明應用於固定長度訊號碼與不定長度 訊號碼之硬體設備成本比較圖。 訂 j-------· 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） 525355 A7 B7 五、發明說明（6 ) 圖號說明如下： 加法器1 量化器2 積分器3 5 輸入訊號A 脈波訊號C 輸出訊號D 步驟10 步驟11 10 步驟I2 步驟13 步驟14 步驟15 步驟16 15 步驟17 步驟1 8 流程20 如第7圖所示 6337-1-0521 修.doc - 8/30 流程21 流程22 20 流程23 流程24 流程25 流程26 流程261 25 流程27 流程28 流程281 流程29 第一加法器30 30 第二加法器40 量化器50 還原器60 本發明所應用之電路包括一第一加法 器1、一第二加法器2、一量化器3及一還原器4 ;對類 35比訊號取樣得訊號Si，與自還原器4輸出之前次量化還原 數位訊號樣本QSi-Ι於第一加法器1相減得一差値Di， 將差値Di與還原器4輸出之前次量化還原之差値Q Di-1 於第二加法器2相減得一差値D Di，再經量化器3量化 得一量化値Q D Di，該量化値Q D Di經壓縮後用以儲存 -------\ ——丨·衣-------J訂 j—------ (請先閱讀背面之注意事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） 525355 A7 B7 6337-1-0521 修.doc - 9/30 $、發明說明（7) 或傳送；還原時，將該量化値Q D Di於還原器4與原前 次量化還原差値Q Di-i計算求得本次之量化還原差値Q Di，及以還原器4將本次量化還原差値Q Di與上次量化 還原之數位樣本Si-Ι計算求得本次量化還原之數位樣本 5 Q Si。 其計算式爲：525355 A7 B7 6337-1-0521 Rev. doc-3 / 30 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the invention (1) The invention relates to a method for modulating analog digital signals, especially a modulation method Those who convert the analog signal into a digital signal by a quadratic difference method. This method first samples the input analog signal to obtain a digital signal pulse, and subtracts the digital signal pulse from the digital signal pulse of the previous quantization reduction to obtain a difference of 5 : Subtract the primary difference from the previous quantized reduction difference output of the Predictor to obtain the secondary difference; quantize the secondary difference to obtain a compressed quantization 2 The second rate is used for transmission or storage; the reducer (Predictor) is used to add the previous quantized reduction rate to the current quantized second rate to obtain the current quantized reduction rate. (Predictor) adds the current quantized reduction rate to the digital signal of the previous quantized reduction to obtain the digital signal of this quantized reduction. By this method, the digital signal obtained by modulation has the smallest noise ratio and signal distortion. And with a compression ratio of the largest province and the most simple storage space, etc. are. Press, one of the modulation techniques 15 used to convert analog signals into digital signals, such as the Delta Modulation (DM) developed in the 1940s and applied to telephone speech, as shown in Figure 1 No, the primary differential modulation method uses an adder 30, a quantizer 40, and an integrator 50 to achieve the purpose of converting the analog to a digital signal. It is the subtraction of the analog input signal Si and the digital signal QSi-1, which was reduced to 20 times before the output through the integrator 50, in the adder 30 to subtract a difference 値 Di, which is generated by the quantizer 40. A quantized difference QDi is used as a signal for transmission or storage; the quantized difference QDi can be fed back to the integrator 50 and restored to the digital signal QSi restored by the next quantization as the parameter of the next sampling operation. In other words, the one-time differential modulation method (DM) is a paper size applicable to the Chinese National Standard (CNS) A4 (210 X 297 mm) (please read the precautions on the back before filling this page)-Order il Line surface 525355 A7 B7 6337-1-0521 Rev.doc-4 / 30 V. Description of the invention (2) A technique for differentially quantizing a signal, that is, to replace the traditional direct amplitude coding method with the differential coding of the signal amplitude, so One-time differential modulation (DM) is also called pulse width modulation (PWM). When the sampling frequency is very high (that is, the sampling period is very short), the signal change during the sampling period is very small 5 and its quantization difference can be compressed to a minimum number of bits to fully perform. _ One-bit converters with single-bit encoding (1-bit Converters) 'As shown in Figure 2, a sine wave with a frequency of 1.5KHZ and a maximum amplitude of ± 1 is used as the input signal, and the sampling signal difference is selected to be 0.125 The ratio of quantized compression to 1 bit represents 4 quantized bits. If it is encoded with 4 bits, 24 = 16 types of quantization are obtained for 10. Then Figure 2 shows that when the 32-times sampled pulse wave signal C is modulated, the output signal D of the integrator (that is, the digital signal after modulation) is very close to the analog signal A of the input; if the 16-times sampled pulse wave_ When the signal C is modulated (as shown in Figure 3), the digital signal D output by the integrator will be far away from the input signal A (representing a larger error in signal conversion); 15 If the rate or sampling signal ratio is too low, then The output signal waveform D will be inclined with a large slope and far behind the input signal A (as shown in Figure 4) ', resulting in signal distortion and signal error. That is, the one-time difference technology is used to convert the analog signal into a digital signal. Although the signal can be encoded in a single bit to save hardware equipment costs, it is a one-time difference quantization obtained by subtracting the amplitude of the sample from 20 It has a fixed slope (as shown in Figures 5A to 5C, and the rates are d = 1, d = 0, and d = — 1). Therefore, its quantization # difference is large. From Figures 2 to 4, we can know the sampling time. The longer (the lower the sampling frequency) or the smaller the sampling difference is, the more serious the distortion is, and the larger the error between the digitized signal and the original analog signal. Therefore, the one-time differential modulation method (-4- This paper size is applicable to the Chinese National Standard (CNS) A4 specification (210 X 297 mm)) (Please read the M-notes on the back before writing the iri, ------- Printed by the Consumers 'Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 525355 A7 B7 6337-1-0521 Rev. doc-5 / 30 Printed by the Consumers' Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the invention (3) DM) Yuan completes the signal encoding and converts the analog signal into a digital signal, which makes the hardware equipment cost lower, but the signal conversion process is greatly lost, and the conversion effect is not ideal. To improve the modulation effect (reduce the distortion rate), It is necessary to relatively increase the sampling frequency, so its overall equipment cost is still 5. The other two conventional analog digital signal modulation techniques are Pulse Code Modulation (PCM) and differential pulse code modulation. (Differential Pulse Code Modulation, DPCM for short), the pulse code modulation method (PCM 10) is a digital signal that is quantized and encoded by sampling analog signals; and the differential pulse code modulation method (DPCM) It is a comprehensive application of the previous differential DM technology and pulse code modulation (PCM). That is, the analog signal is sampled first, and then the difference between the actual sampling value and the previous sampling value is obtained. Finally, the digital signal is quantized and encoded. Utilizing the neutrality of the sampling rate of the continuously sampled signal is 15 neutral, and encoding can be completed in fewer bits. As shown in Figure 6A, it is an image pattern of a digital signal converted by the pulse code modulation (PCM) analog sound signal, The figure shows that the sampled 値 resulting from the conversion is more concentratedly distributed between 20 and 160, and its signal range includes 160-20 = 140. If the range of 0 to 140 値 is represented by binary bits, at least 8 bits are required ( 2〇27 = 128 < 140 < 28 = 256), so it ca n’t complete the signal encoding with shorter character codes, that is, its compression is smaller; otherwise, as shown in Figure 6B, the analog signal is DPCM technology The code is converted into a digital signal, and the sampling rate obtained is concentrated between -20 and 20, and its signal range includes 20— (-20) = 40, so it is expressed in binary bits from 0 to 40. (Please read the back first Note? Please fill in this matter ) 1 --- ¾ Γ This paper size applies to Chinese National Standard (CNS) A4 specification (210 X 297 mm) 525355 A7 B7 6337-1_0521 repair. 〇1〇 (: -6/30 V. Description of the invention (4) The digital data only needs 6 bits (25 = 32 < 40 < 26 = 64), so the signal can be encoded with a shorter character code, and its compression is relatively large. Compared with Figures 6A and 6B, the differential pulse code modulation (DPCM) method saves space in storage equipment and reduces the cost of hardware equipment compared to pulse code modulation (PCM). However, although the differential pulse code modulation method (DPCM) is an improved pulse code modulation method (PCM), it has the advantage of saving costs compared to hardware equipment, but its cost is still high and it cannot be widely used in general cheap goods ( Such as cheaper sound toys). 10 The present invention was developed to overcome the shortcomings of the high cost and local distortion of the conventional analog-to-digital conversion technology, and it has the advantages of cheaper price and better modulation effect. In other words, the main purpose of the present invention is to provide a new method for converting analog signals into digital signals. Compared with digital signals, the error and signal distortion are extremely small, and I5 can obtain good signal conversion quality. A second object of the present invention is to provide a new method for converting analog signals into digital signals. Because the quantized signals are very concentrated, the coding can be completed with fewer bits, and the cost can be reduced. Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs --------------- (Please read the notes on the back before filling this page) For the above and other purposes, effects and technical contents of the present invention, please refer to The following 20 drawings and descriptions for further understanding. The diagram is explained as follows: Figure 1 is a block diagram of the conventional primary differential modulation method (dm), and Figure 2 is a primary differential modulation method with 32 times the sampling pulse signal. Standard (CNS) A4 (210 X 297 male f) 525355 A7 B7 6337-1-0521 Rev. doc-7/30 V. Description of the invention (5) Relationship between analog and digital signal obtained by changing time; Figure 3 Figure 1 shows the relationship between the analog and digital signals obtained when the 16-times sampling pulse wave signal is modulated by the one-time differential modulation method. Figure 4 shows that the 16-times pulse wave signal is sampled by the one-time differential modulation method and the difference is 5. Figure 1 shows the relationship between the analog and digital signals obtained during modulation. Figures 5 A to 5 C are schematic diagrams of three fixed slopes of the one-time differential modulation method. Figure 6 A is the digits obtained by pulse code modulation (PCM). Figure 10 shows the image distribution; Figure 6B is a digital signal distribution image obtained by the differential pulse code modulation (DPCM) method; Figure 7 is a block diagram of the circuit of the present invention; Figure 8 is the second difference of the present invention Signal sampling diagram of modulation method; 15th is the second invention Schematic diagram of three types of variation slopes of the differential modulation method; (Please read the precautions on the back before filling in this page.) Equipment ---- Printed by the Consumers ’Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs. Block diagram; Figure 11 is a flowchart of the modulation method of the present invention; Figure 12A is an image diagram of the digital signal distribution obtained by the second differential modulation method of the present invention; Figure 1 2B is the second of the present invention Schematic diagram of digital signal distribution obtained by the sub-differential modulation method; FIG. 13 is a comparison diagram of hardware equipment cost of the present invention applied to fixed-length signal numbers and variable-length signal numbers. Order j ------- · This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) 525355 A7 B7 V. Description of the invention (6) The drawing numbers are explained as follows: Adder 1 Quantizer 2 Integrator 3 5 Input signal A Pulse signal C Output signal D Step 10 Step 11 10 Step I2 Step 13 Step 14 Step 15 Step 16 15 Step 17 Step 1 8 Process 20 Repair as shown in Figure 6337-1-0521. doc-8/30 Flow 21 Flow 22 20 Flow 23 Flow 24 Flow 25 Flow 26 Flow 261 25 Flow 27 Flow 28 Flow 281 Flow 29 First Adder 30 30 30 Second Adder 40 Quantizer 50 Reducer 60 Application of the invention The circuit includes a first adder 1, a second adder 2, a quantizer 3, and a reducer 4; the signal Si is sampled from a 35-level analog signal, and the previous quantized and restored digital signal sample is output from the reducer 4. QSi-1 subtracts a difference 値 Di from the first adder 1 and subtracts the difference 値 Di from the previous quantization reduction 输出 Q Di-1 from the reducer 4 to subtract a difference 値 D from the second adder 2. Di, and then quantized by quantizer 3 to obtain a quantized 値 QD Di, the quantized 値 QD Di is compressed for storage-- ----- \ —— 丨 · Clothing ------- J Order j ------- (Please read the notes on the back before filling this page) The paper size of the paper is applicable to the Chinese National Standard (CNS) A4 (210 X 297 mm) 525355 A7 B7 6337-1-0521 Rev. doc-9/30 $, invention description (7) or transmission; when it is restored, the The quantization 値 QD Di in the reducer 4 and the previous quantization reduction difference 値 Q Di-i are calculated to obtain the current quantization reduction difference 値 Q Di, and the reduction unit 4 compares the current quantization reduction difference DiQ Di with the previous The quantized reduced digital sample Si-1 is calculated to obtain the digital sample 5 Q Si of the quantized reduction. Its calculation formula is:
Di = Si-QSi-l.......( a ) 取樣時須 有Q Di-Ι及Q Di-Ι之回饋， DDi = Di — QDi-1......( b ) Si 爲輸 10入訊號之第i個取樣，Di爲 Q D Di = Q Table (D Di) ···(〇) 輸入訊 號與QSi-1之差，D Di爲Di與 Q Di = Q Di-1 +Q D Di · · · · ( d ) QDi-1 之差，用D DI來做查表得Q D Di 15 Q Si = QSi-l+Q Di.....( e ) 以Q D DI壓縮後儲存或傳送 * 初始値 QS-1 =Q D-l = 0 解壓縮時以Q D Di作爲還原器4之輸入訊號’利用 (d )、（ e )兩式即可求得Q Di及Q Si値。 20 請參閱第8圖所示，對一弦波訊號取樣，本發明二次 差分調變方法之理原爲：Di = Si-QSi-l ....... (a) Q Di-I and Q Di-I must be provided when sampling. DDi = Di — QDi-1 ...... (b) Si To input the i-th sample of the input signal, Di is QD Di = Q Table (D Di) ··· (〇) The difference between the input signal and QSi-1, D Di is Di and Q Di = Q Di-1 + QD Di · · · · (d) The difference between QDi-1. Use D DI to look up the table. QD Di 15 Q Si = QSi-l + Q Di ..... (e) Compressed with QD DI or stored or Transmission * Initial 値 QS-1 = Q Dl = 0 When decompressing, QD Di is used as the input signal of the restorer 4 'Q Di and Q Si 値 can be obtained by using (d) and (e). 20 Please refer to FIG. 8 for sampling a sine wave signal. The principle of the secondary differential modulation method of the present invention is as follows:
取樣初始値S-l = d-l = d d_l=〇 經計算得一次差分値d0 = s0 - s-l = s0 dl=sl— sO 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公t ) (請先閱讀背面之注音？事項再填寫本頁) •丨-丨丨?訂—.-----丨 · 經濟部智慧財產局員工消費合作社印製 525355 A7 B7 6337-1-0521 修.doc - 10/30 經濟部智慧財產局員工消費合作社印製 五、發明說明（8 ) d 2 = s 2 — s 1 d3=S3-s2 di = si — si-l = si 經計算得二次差分値：Sampling initial 値 Sl = dl = d d_l = 〇 once calculated the difference 値 d0 = s0-sl = s0 dl = sl — sO This paper size applies to China National Standard (CNS) A4 (210 X 297 mm t) (please Read the phonetic on the back? Matters and then fill out this page) • 丨-丨 丨? Order —.----- 丨 · Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 525355 A7 B7 6337-1-0521 Repair.doc- 10/30 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the invention (8) d 2 = s 2 — s 1 d3 = S3-s2 di = si — si-l = si The calculated quadratic difference 値:
5 d dO = dO — d-l = dO d d 1 = d 1 - d 0 dd2=d2—dl d d3=d3—d2 即 d di = di — di-1 10 = ( si-si-1) - (si-1-si-2) 一 si — 2si-l~(~ si-2 利用本發明之二次差分調變方法可求得取樣訊號間之 斜率變化，包括有値線（d d = 〇，如第9B圖所示）、上 拋物曲線（d d =1，如第9A圖所示）及下拋物曲線（ 15 d d = 一1 ’如第9C圖所示）三種，因而其二次差値幾乎 呈連續性’差値較小，所調變之數位訊號分佈如第12A 圖及第12B圖所示，大致集中在一 1〇至1〇之間，只需五 個位兀即可完成編碼，且其分布甚爲集中（即壓縮比較大 )，若忽略兩端極微小之訊號，則只需4個位元（甚至更 20少')即可滿足’因而較上述習知—次差分調變法及差分脈 衝碼調變法均較精簡位元，且取樣頻率亦不必特別高，故 其成本較低，且無失真之虞。 此間應予以g兌明者，乃在於：圖9A到9(：所欲表達的 疋，雖然只利用最小的+1、-1、0，卻可以：（1 )、產 -10 - 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公璧) ------ ----------I 衣---------^il------線 (請先閱讀背面之注音？事項再填寫本頁) 525355 A7 B7 6337-1-0521 修.doc - 11/30 經濟部智慧財產局員工消費合作社印製 五、發明說明（9) 生拋物線，迅速調整斜律；（2)、維持在不同之斜律， 如圖9B; (3)、在不同之斜律下隨時以拋物線更改斜律 ，如圖9A，由負變正，9C，由正變負；所以在高壓縮比 外，又能維持失真很小，尤其是對低頻大訊號如定音鼓加 5上高頻小訊號如小提琴之狀況，比傳統的DPCM、 DM(Delta Modulation)等方式好出很多很多，此類狀況 俗稱Intermodulation Distortion，爲評估壓縮方發的一 項重要指標；比傳統之ADPCM更有彈性、性能更佳、卻 更簡單便宜許多；實爲本發明之一大特色。 10 如第1〇圖所示，本發明之步驟包括： 步驟1 :對類比輸入訊號取樣Si ; 步驟2 :取輸出之前次量化還原之數位樣本QSi-1 ; 步驟3 :計算取樣訊號Si與前次量化還原之數位樣 本QSM之差値Di ; 15 步驟4 :取輸出之前次量化還原差値Q Di-1 ; 步驟5 :計算差値Di與前次量化還原差値Q Di-1之 差値得二次差値D Di ; 步驟6 ;將二次差値D Di値量化得本次量化差値Q D Di ； 20 步驟7 ;計算前次量化還原差値Q Di-1與本次量化 差値Q D Di，得本次量化還原差値Q Di ; 步驟8 :計算前次量化還原之樣本QSi-1與本次量化 還原差値Di，得本次量化還原之數位樣本QSi。 步驟9 :儲存或傳送量化差値Q D Di。 -11 - 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） (請先閱讀背面之注意事項再填寫本頁) --------Γ 訂--------· 525355 A7 B7 6337-1-0521 修.doc - 12/30 五、發明說明（10) (請先閱讀背面之注意事項再填寫本頁) 本發明之方法之流程如第11圖所示，其流程與上述 步驟相當，包括： stepl :開始； step2 :設定初始値 i =0，QS-1=0，Q D-l^O，並 5設定參數N値； step3 :取輸入訊號樣本Si ; steP4 :計算輸入訊號樣本與前次量化還原之數位訊 號樣本之差値（Di=Si —QSi-Ι); step5 ··計算該差値與二次差値D Di = Di —Q Di-1 10前次量化還原差値，得二次差値（D Di = Di — Q Di-1 )5 d dO = dO — dl = dO dd 1 = d 1-d 0 dd2 = d2—dl d d3 = d3—d2 ie d di = di — di-1 10 = (si-si-1)-(si- 1-si-2) si — 2si-l ~ (~ si-2 The slope difference between the sampling signals can be obtained by using the second differential modulation method of the present invention, including the chirp line (dd = 〇, as in Section 9B Figure), the upper parabolic curve (dd = 1, as shown in Figure 9A) and the lower parabolic curve (15 dd = 1 1 'as shown in Figure 9C), so the quadratic difference is almost continuous 'The difference is small. The modulated digital signal distribution is shown in Figure 12A and 12B, which is roughly concentrated between 10 and 10, only five bits are required to complete the coding, and its distribution It is very concentrated (that is, the compression is relatively large). If you ignore the extremely small signals at both ends, only 4 bits (or even less than 20) can be satisfied. Therefore, it is better than the conventional method of sub-differential modulation and differential pulse. The code modulation method is more streamlined, and the sampling frequency does not have to be particularly high, so its cost is low, and there is no risk of distortion. Those who should be given g here are: Figures 9A to 9 (: What you want to express Alas, although only profit The smallest +1, -1, 0, but can: (1), production -10-This paper size applies to China National Standard (CNS) A4 specifications (210 X 297 cm) ------ ---- ------ I clothing --------- ^ il ------ line (Please read the phonetic on the back? Matters before filling out this page) 525355 A7 B7 6337-1-0521 Repair. doc-11/30 Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the invention (9) Parabola is generated, and the slope is adjusted quickly; (2) Maintained at a different slope, as shown in Figure 9B; (3), in Change the slope with a parabola at any time under different slopes, as shown in Figure 9A, from negative to positive, 9C, from positive to negative; so outside of high compression ratio, the distortion can be kept small, especially for large low-frequency signals such as timpani. Add 5 high frequency small signals such as the condition of the violin, which is much better than the traditional DPCM, DM (Delta Modulation) and other methods. Such conditions are commonly known as Intermodulation Distortion, which is an important indicator for evaluating the compression side. ADPCM is more flexible, has better performance, but is much simpler and cheaper; it is a major feature of the present invention. 10 As shown in Figure 10, the steps of the present invention include: Step 1: Sample Si for the analog input signal; Step 2: Take the digital sample QSi-1 that was previously quantized and restored; Step 3: Calculate the difference 値 Di between the sampled signal Si and the digital sample QSM that was previously quantized and restored; 15 Step 4 : Take the output of the previous quantization reduction difference DiQ Di-1; Step 5: Calculate the difference of the difference 値 Di and the previous quantization reduction difference 値 Q Di-1 to obtain the secondary difference DiD Di; Step 6;値 D Di 値 is quantized to obtain the current quantization difference 値 QD Di; 20 Step 7; Calculate the previous quantized difference 値 Q Di-1 and the current quantized difference 値 QD Di to obtain the current quantized difference 値 Q Di; Step 8 : Calculate the sample QSi-1 of the previous quantization reduction and the current differential 値 Di to obtain the digital sample QSi of this quantization reduction. Step 9: Store or transmit the quantized difference Q D Di. -11-This paper size is in accordance with Chinese National Standard (CNS) A4 (210 X 297 mm) (Please read the precautions on the back before filling this page) -------- Γ Order ----- --- · 525355 A7 B7 6337-1-0521 Rev. doc-12/30 V. Description of the invention (10) (Please read the notes on the back before filling this page) The process of the method of the present invention is as shown in Figure 11 As shown, the flow is equivalent to the above steps, including: stepl: start; step2: set initial 値 i = 0, QS-1 = 0, Q Dl ^ O, and 5 set parameters N 値; step3: take the input signal sample Si; steP4: Calculate the difference between the input signal sample and the digital signal sample from the previous quantization reduction (Di = Si — QSi-I); step5 ·· Calculate the difference and secondary difference D Di = Di — Q Di-1 10 The previous quantization reduction of the rate difference yields the second rate difference (D Di = Di — Q Di-1)
steP6 :利用預設表對照求（Table)得量化値Q DsteP6: Use the preset table to find the quantization 値 Q D
Di ； steP7 :計算本次量化差値與前次量化還原差値，得 15 —本次量化還原差値（QDi = QDi-l+QDDi); steP8 :計算本次量化還原差値與前次量化還原之數 位訊號樣本，得本次量化還原之數位訊號樣本（QSi = QSi-1 +Q Di)； 經濟部智慧財產局員工消費合作社印製 steP9 :儲存或傳送量化差値Q D Di ; 20 step 1 0 ··設 I 二 i + 1 ; stepll :比較I是否小於N (I<N);若I<N，貝fj 執行step3重複上述步驟（step3至stepll)，否則執行 stepl2 ； stepl2 :結束。 -12- 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） 525355 A7 B7 6337-1-0521 修.doc - 13/30 經濟部智慧財產局員工消費合作社印製 五、發明說明（11) 舉例說明本發明之訊號調變方法，若以3個位兀記錄 數位訊號時可得8種値： 100 101 110 111 000 001 010 011 一 4 一 3 一 2 一 1 0 1 2 3Di; steP7: Calculate the current quantization difference and the previous quantization reduction difference, and get 15—the current quantization reduction difference (QDi = QDi-l + QDDi); steP8: calculate the current quantization reduction difference and the previous quantization reduction. Reduced digital signal samples to get the quantized reduced digital signal samples (QSi = QSi-1 + Q Di); SteP9 printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs: Store or transmit the quantized differential QD Di; 20 step 1 0 ·· Let I 2 i + 1; stepll: compare whether I is less than N (I <N); if I < N, execute step3 and repeat the above steps (step3 to stepll), otherwise execute stepl2; stepl2: end. -12- This paper size is in accordance with Chinese National Standard (CNS) A4 (210 X 297 mm) 525355 A7 B7 6337-1-0521 Rev. doc-13/30 Printed by the Consumer Cooperative of Intellectual Property Bureau of the Ministry of Economic Affairs Explanation (11) Explains the signal modulation method of the present invention as an example. If digital signals are recorded in 3 bits, 8 types can be obtained: 100 101 110 111 000 001 010 011-4-3-2-1 0 1 2 3
若出現之値>3，則以3表示；若出現之値< 一4，貝[J 以-4表示；如此雖有極微小之訊號損失，惟如第12A、 5 12B圖所示，該兩端邊緣之訊號出現率微乎其微，故可忽 略不計，而不影響訊號數位化之真實度。 如下表所示，利用一個bite記錄二次差分規則： ①數位訊號（symbol )連續4次1 ( 1111 ) 則二次差分之位階（step)加1 ; 10 ②數位訊號（symbol )連續4次0 ( 0000 ) 則二次差分之位階（step)減1 ; ③數位訊號（symbol)由1變0時，則二次差分之位 階減1 ; 數位訊號（symbol)由0變1時，則二次差分之位 15階加1。 其判斷方法爲： 當第i次取樣之差値大於第i-Ι次取樣差値，即Dig Di-1時，則取數位訊號（symbol ) = 1 ; 當第i次取樣之差値小< 於第i-1次取樣差値，即Di 20 < Di-1時，則取數位訊號（symbol ) = 0 ; 其數位訊號（symbol)與二次差分値（Delta of Delta)之對應關係如下表： -13- 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公爱） (請先閱讀背面之注意事項再填寫本頁) 訂： 525355 A7 B7 6337-1-0521 修.doc - 14/30 五、發明說明（12) symbol DeTtT~ of Delta 13 12 13 12 11 10 由上表可得，以本發明之二次差分調變方法將類比訊 5虎轉換成數位訊號時’可適度地略過部分差値（如在連續 6個1時其第4、5、6個1處依上述二次差分規則①將差 値D之位階加1，而略過5、7、8、10、11、12;同理， 5於連續5個0時其第4及第5個〇處將差値D減1，而略 過9、7、6，即得上表），而不會提高訊號數位化過程之 失真或誤差。 本發明之二次差分調變方法所用之數位訊號包括固定 長度及不定長度兩種，其中固定長度者依照上表之略過方 10式，可以5個位元（25=32種訊號）來表示習知技術需8 個位元之訊號長度：0、1、2、3、4、5、7、10、14、 19、25、32、40、49、59、70 及—1、—2、—3、—-4 、一 5 一 7、一 10、一 14、一 19、一 25、一 32、一 40、 一 49、一 59、一 70，即本發明能以較少位元數完成訊號 15數位化之編碼，而節省記憶體容量。 不定長度之數位訊號（symbol)則因二次差分量化差 値之分布如第12A及12B圖所示，越靠近0點訊號出現 率越高，其壓縮比越大，以較多位元數表示之（例如5個 位元）’越遠離0點訊號出現率越低，其壓縮比越小，以 2〇較少之位元數表示之（例如3個位元），故量化差値 QDDi越靠近〇點其使用之位元（bit)數越少，越簡省記 -14- 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公爱） I I------- (請先閱讀背面之注意事項再填寫本頁) ~tr· · 經濟部智慧財產局員工消費合作社印製 525355 A7 B7 6337-1-0521 修.doc - 15/30 五、發明說明（13) 憶體。 比較固定長度及不定長度之數位訊號表示方式之成本 如第13圖所示，圖中虛線代表固定長度之數位訊號所需 之硬體設備成本，實線代表不定長度之數位訊號所需之硬 5體設備成本，經實驗所得第13圖中當訊號長度（例如錄 音長度）超過3分鐘以上時，不定長度之數位訊號表不方 式其調變後之訊號失較少，壓縮比較高，較簡省記憶體， 但其基本硬體設備較貴，因此，使用者可依據個人對品質 之要求標準而選用不定長度之數位訊號表示方式或固定長 10度之數位訊號表示方式。 本發明可應用於有聲書本或發聲玩具等方面，以提供 使用者多元化之學習工具或趣味化遊戲者。 綜上所述，本發明之訊號調變方法其技術內容較習知 訊號調變方法具革命性之創舉，且具最佳之調變品質及最 I5高之壓縮比，所得之數位訊號失真最小，且最節省記憶體 空間，而充份符合發明專利之申請要件者。 ------1---ίφΜ--------------^#1 - , (請先閱讀背面之注音？事項再填寫本頁) 經濟部智慧財產局員工消費合作社印製 本紙張尺度適用中國國家標準（CNS)A4規格（21〇 χ 297公餐）If the occurrence of 値 > 3 is represented by 3; if the occurrence of 値 < -1, 4 [J is represented by -4; so although there is minimal signal loss, as shown in Figures 12A and 5 12B, The occurrence rate of the signals at the edges of the two ends is negligible, so it can be ignored without affecting the authenticity of the signal digitization. As shown in the following table, a bite is used to record the rule of quadratic difference: ① The digital signal (symbol) is continuously 4 times 1 (1111), then the step of the quadratic difference (step) plus 1; 10 ② the digital signal (symbol) is 4 times consecutive 0 (0000) The step of the quadratic difference is reduced by 1; ③ When the digital signal (symbol) changes from 1 to 0, the level of the quadratic difference decreases by 1; When the digital signal (symbol) changes from 0 to 1, the quadratic 15th order plus one. The judgment method is: when the difference between the i-th sampling is greater than the difference between the i-l and the first sampling, that is, Dig Di-1, then the digital signal (symbol) = 1; when the difference between the i-th sampling is small < At the i-1th sampling difference, that is, when Di 20 < Di-1, the digital signal (symbol) = 0; the corresponding relationship between the digital signal (symbol) and the second difference (Delta of Delta) The following table: -13- This paper size is applicable to China National Standard (CNS) A4 specification (210 X 297 public love) (Please read the precautions on the back before filling this page) Order: 525355 A7 B7 6337-1-0521 Repair. doc-14/30 V. Description of the invention (12) symbol DeTtT ~ of Delta 13 12 13 12 11 10 It can be obtained from the above table, when the analog signal 5 tiger is converted into a digital signal by the second differential modulation method of the present invention ' Partial rates can be skipped moderately (for example, at the 4th, 5th, and 6th positions of 6 consecutive 1s, according to the above-mentioned quadratic difference rule ①, the rank of the rate D is increased by 1, and skipped 5, 7, 8 , 10, 11, 12; Similarly, 5 will reduce the rate D by 1 at the 4th and 5th 0s for 5 consecutive 0s, and skip 9, 7, 6 to get the above table), instead of Will increase the signal digitization process Distortion or error. The digital signals used in the secondary differential modulation method of the present invention include two types of fixed length and indefinite length. Among them, the fixed length can be expressed in 5 bits (25 = 32 types of signals) according to the skip formula of the above table. Known technology requires 8-bit signal length: 0, 1, 2, 3, 4, 5, 7, 10, 14, 19, 25, 32, 40, 49, 59, 70, and -1, -2, —3, —-4, one 5 one 7, one 10, one 14, one 19, one 25, one 32, one 40, one 49, one 59, one 70, that is, the present invention can be completed with fewer bits Signals are digitally encoded to save memory capacity. Digital signals of indefinite length (symbol) are distributed as shown in Figures 12A and 12B due to the secondary differential quantization difference. The closer the signal is to 0, the higher the signal occurrence rate is, the larger the compression ratio is, and it is expressed in more bits. (For example, 5 bits), the farther away from 0 points, the lower the signal occurrence rate, the smaller the compression ratio, which is represented by 20 fewer bits (for example, 3 bits), so the more the quantization difference QDDi The closer it is to 0, the fewer the number of bits it uses, the simpler it is to save. -14- This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297 public love). I I ------- (Please read the precautions on the back before filling this page) ~ tr · · Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 525355 A7 B7 6337-1-0521 Rev. doc-15/30 V. Description of Invention (13) Recall body. The cost of comparing fixed-length and variable-length digital signal representations is shown in Figure 13. The dashed line in the figure represents the hardware equipment cost required for a fixed-length digital signal, and the solid line represents the hardware required for a variable-length digital signal. The cost of the external equipment is shown in the experiment. When the signal length (such as the recording length) exceeds 3 minutes, the digital signal of indeterminate length will show less signal loss after modulation. Memory, but its basic hardware equipment is more expensive. Therefore, users can choose a digital signal representation of variable length or a digital signal representation of a fixed length of 10 degrees according to personal quality standards. The invention can be applied to audio books or audio toys to provide users with a variety of learning tools or fun games. In summary, the technical content of the signal modulation method of the present invention is revolutionary than the conventional signal modulation method, and has the best modulation quality and the highest compression ratio of I5. The resulting digital signal distortion is the smallest. , And save the most memory space, and fully meet the requirements of the invention patent application. ------ 1 --- ίφΜ -------------- ^ # 1-, (Please read the note on the back? Matters before filling out this page) Staff of the Intellectual Property Bureau of the Ministry of Economic Affairs The paper size printed by the consumer cooperative is applicable to China National Standard (CNS) A4 (21〇χ 297 meals)