499670 A7 B7__ 五、發明說明（/ ) 【發明之應用範疇】 i 本發明是關於一種聲音信號合成方法及裝置，特S提關於一 種以內插法運算來改善語音合成品質及降低雜訊之聲音信號合成499670 A7 B7__ V. Description of the invention (/) [Application scope of the invention] i The present invention relates to a method and device for synthesizing a sound signal, and particularly mentions a sound signal that uses interpolation to improve the quality of speech synthesis and reduce noise. synthesis
I 方法及裝置及依該方法而設計之聲音合成電路。 【發明之背景】 一般之聲音信號取樣的方法，是利用固定的頻率（取樣頻率 sampling mte )，對一聲音波形進行取樣，經_比數位轉獬|( A/D Converter)將取樣之類比波形直接轉換成數位編碼，此一編碼方 :式稱爲Pulse Code Modulation ( PCM )。而聲音信號合成的方法， 貝[J是依據該固定之頻率，透過數位類比較^器（D/A Converter) 將PCM的編碼_成類比信號。由於取樣時係依固定頻率，所以 取樣後之編碼資料會與實際之類比聲音波形，有某些程度上之失 寘現象。當取讎率愈低或是A/D Converter的解析度愈低都會使 得失真現象愈嚴重。 ， 第1圖顯示一個原始聲音信號波形與其利用PCM方式編碼 之取樣資料之關係圖。圖中，橫座標爲取樣時間，縱座標表其竣 形之振幢，曲線（1)表示該原始聲音信號波形，折線（2)袠笊 經以一定頻率（以XHz )，以PCM方式取樣後再合成之聲音波形。 3 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) 丨丨線丨-#- 經濟部智慧財產局員工消費合作社印製 Ί 經濟部智慧財產局員工消費合作社印製 499670 A7 B7 五、發明說明) 圖中顯示，類比聲音波开Μ 1)以XHz之取樣頻率（sampling rate ) 1 . ·. 取樣之後，以相同之PCM方式，在XHz頻率下進行語音合成後， 所得的波形與原聲音波形會有若干差異。該差異部份將造成輸出 聲音的失真。在一^的聲音合成裝置上，尤其是聲音合成1C之設 計上，爲節省記憶體，或爲延長儲存聲音之播放時間長度，可能 ------ 將該取樣頻率降低或是降低A/D轉換器之撕度。其結果麟成 合成聲音信號更進一步失真。 在習用技術中，爲改善輸出聲音的失真，可以提高取樣頻率 之方式來達成。第2圖即表示第1圖中之原始聲音信號（1)利用 4XHz之取樣頻率取樣後再合成所得之聲音信號波形。圖中（3 ) 代表經以4XHZ取樣後之聲音信號波形。如圖所示，經以4倍頻 一一-一 ·， 率取樣之後，確可使所得聲音波形趨近於原聲音波形，降低其失 真。但如此一來所需之聲音資料記憶空間卻是以XHz取樣頻率時 所需空間之4倍，導致設備成本的提高。 因此目前亟需有一種新穎的聲音信號合成方法及裝置，該方 法可以提供失真較少之聲音信號取樣，但又不增加占用之記憶體 ;【發明之目的】 4 本紙張尺度適用中國國家標準（CNS)A4規格（210 X 297公釐） -----------1» --------------^ I i^w— (請先閱讀背面之注意事項再填寫本頁) 499670 經濟部智慧財產局員工消費合作社印製 Α7 Β7 五、發明說明（彡） 本發明之目的乃在提供一種新穎的聲育信號合成方法及裝 , ; 置，該方法可以提供失真較小的聲音合成信號，並可節省儲存聲 音信號所佔用之記憶體空間。 【發明之簡述】 讎本發明之聲音信號合成方法及裝置，聲音信號係以較低 之取樣頻率取樣。但在再生時則利用內插法運算，算出兩取樣週 期間之聲音信號値，將該計算値塡入該兩取樣週期之間，加以再 ，生，即可獲得降低失真率之再生聲音。本發明提供一種大幅降低 失真且取樣頻率不需太高之聲音信號合成方法及裝置。 i：述及其他本發明之目的及優點，可由以下詳細說明並參照 下列圖式，而更形淸楚。 【圖式之說明】 第1圖表示一種聲音波形與以低取樣頻率取樣所得之合成聲 音Iff彡示意圖。 第2圖表示第1圖之聲音波形與以較第1圖之取樣頻率爲高 率取樣所得之合成聲音波形示意圖。 第3圖表示第1圖之聲音波形與以本發明之聲音信號合成方 • 裝置取樣後所得之合成聲音波形示意圖。； _ 5 本紙張尺度適用中國國家標準(CNS)f規格（210 X 297公爱） (請先閱讀背面之注意事項再填寫本頁) 一^J« H ϋ n I n ϋ n I n ϋ ϋ I I I I I I I I n ϋ n n n n I n I I n - 499670 經濟部智慧財產局員工消費合作社印製 A7 B7 五、發明說明（4) 第4圖表示本發明聲音信號合成方法及裝置之流程圖。 i ' 第5圖顯示一個以單位取樣頻率取樣所得之聲音信號餓， 與利用4倍單位播放頻率播放之聲音信號資料圖型示意圖。 表1顯示在兩種取樣/播放頻率下之取樣及播放聲音信號數 値表。 【發明之詳細說明】 以下依據圖式說明本發明之聲音信號合成方法及裝置。 第3圖顯示第1圖之聲音波形與利用本發明聲音信號合成方 法及裝置所取樣之合成聲音波形示意圖。如第3圖所示，第1圖 所示之聲音信號（1)利用XHz之頻率取樣之後，如在兩取樣週 期之間，***三筆聲音信號取樣資料計算値，再以4倍之再生頻 率（play back rate )播放（=4XHz)，貝(J所輸出之聲音波形將如第 3圖之（4 )所示。該波形與以4XHz取樣頻率所取樣之聲音波形 雖然有些差異，但已極爲_近，而可降低聲音取樣時所產生的失 真。而利用這種方式所產生的聲音信號波形，則僅需與以XHZ取 觀率所取得之聲音信號資料相同之記憶體空間，即足以儲存。 此外，利用本發明之方法所合成之聲音信號，可以降低PCM 1 聲音合成過程中，本身所產生之高，雜訊，並可有效減少因取樣 6 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i -線 i-----I---------I------- 499670 Α7 Β7 五、發明說明（夕） 頻率降低而產生之原音失真。 ^ 1 【實施例】 茲以實施例說明本發明之聲音信號合成方法及裝置之實施 例。第4圖表示本發明聲音信號合成方法及裝置之流程圖。第5 圖則表示一個以單位取樣頻率所取樣之聲音合成波形與以本發明 之方法在4倍單位播放頻率下合成聲音後所得之聲音信號波形示 意圖。以下依據圖式說明。 於（401 )，利用一單位取樣頻率（T)，每隔丄秒取得一聲音 τ 信號波形之PCM編碼資料，得到一聲音信號波形資料檔（D)， D=Di，D2, D3, ".Dn 〇 於（402 )，計算每組第以與第Di4筆資料之差値 Δ/V ΔΑ = A -1。於（403 )，將該除以4，得到I麵 4 △ Di4，Δϋί4= 士 ADi。於（404)在Di與Dw之間***三個聲音 4 信號D#士 ΔΑ，ΔΑ及DH+吞ΔΑ。如此獲得一個取樣 4 4 4 (播放）頻率爲叮丨之聲音信號檔（D，）。於（405 )，以4Τ之頻 率，依據所得之聲音信號檔D’之內容，播放該聲音。此時，其每 一單位播放時間爲^。 4Τ 【發明之效果】 7 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) ·· , - * ϋ i_i n -ϋ n aal· ϋ a n emm9 n 經濟部智慧財產局員工消費合作社印製 線丨·-----：-------- 經濟部智慧財產局員工消費合作社印製 499670 A7 B7 五、發明說明（g) 第5圖顯示一個以單位取樣頻率取樣所得之聲音信號資料’ i ，； - - | 與利用4倍單位播放頻率播放之聲音信號資料圖型示意圖。表1 則顯示在兩種取樣/播放頻率下之取樣及播放聲音信號數値表。 顯示利用提高之播放頻率及使用內插法運算獲得所需之聲音信 號，可以獲得更爲接近原始聲音之波形。 在本發明中，因爲每一筆PCM編碼資料間之差異大爲減小’ 因此可以有效減少聲音合成時所產生之背景高頻噪音。同時因爲 所得之波形比較接近原始聲音波形，因此本發明之方法確可改善 聲音信號播放之品質。 上述說明雖以4等分方式在兩筆聲音信號中***三筆聲音信 號資料，但以較多或較少等分方式，***聲音信號資料，也屬可 行。此外，不以等分方式，而以不等分方式***聲音信號資料， 在某些實施例I中，可以獲得類似或更優異之效果。 以上是對本發明聲音信號合成方法及裝置實施例之說明。習 於斯藝之人士，應可由以上說明得知本發明之精神，並據以作出 各種變化與衍生。唯只要不脫離本發明之精神，均應包含於其申 請發明專利範圍之內。I method and device and sound synthesis circuit designed according to the method. [Background of the invention] The general method for sampling sound signals is to use a fixed frequency (sampling frequency sampling) to sample a sound waveform, and then use the A / D Converter to convert the sampled analog waveform. It is directly converted into digital code. This coding method is called Pulse Code Modulation (PCM). In the method of synthesizing the sound signal, [J is based on the fixed frequency and uses a digital class comparator (D / A Converter) to encode the PCM into an analog signal. Because the sampling is based on a fixed frequency, the encoded data after sampling will be compared with the actual analog sound waveform, and there will be a certain degree of loss. The lower the sampling rate or the lower the resolution of the A / D Converter, the more severe the distortion. Figure 1 shows the relationship between the original sound signal waveform and the sampled data encoded by the PCM method. In the figure, the horizontal coordinate is the sampling time, and the vertical coordinate is its completed vibration. The curve (1) indicates the original sound signal waveform. Resynthesized sound waveform. 3 This paper size applies to China National Standard (CNS) A4 (210 X 297 mm) (Please read the precautions on the back before filling out this page) 丨 丨 丨-#-Printed by the Consumers ’Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 499 Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs 499670 A7 B7 V. Description of the invention) The figure shows that the analog sound wave is opened 1) At the sampling rate of XHz 1. .. After sampling, use the same PCM Method, after speech synthesis is performed at XHz frequency, the obtained waveform will have some differences from the original sound waveform. This difference will cause distortion of the output sound. On a sound synthesis device, especially the design of sound synthesis 1C, in order to save memory or to extend the playback time of stored sound, it is possible to reduce the sampling frequency or reduce A / D converter tear. As a result, Lincheng's synthesized sound signal is further distorted. In conventional technology, in order to improve the distortion of the output sound, it can be achieved by increasing the sampling frequency. Figure 2 shows the waveform of the original sound signal (1) in Figure 1 after sampling with a sampling frequency of 4XHz. (3) in the figure represents the sound signal waveform after sampling at 4XHZ. As shown in the figure, after sampling at 4 times the frequency one-to-one, the obtained sound waveform can be brought closer to the original sound waveform and its distortion can be reduced. However, the required memory space for sound data is four times the space required at the XHz sampling frequency, which leads to an increase in equipment costs. Therefore, there is an urgent need for a novel method and device for synthesizing sound signals, which can provide sound signal sampling with less distortion, but without increasing the occupied memory; [Objective of the Invention] 4 This paper size applies to Chinese national standards ( CNS) A4 specification (210 X 297 mm) ----------- 1 »-------------- ^ I i ^ w— (Please read the Note: Please fill in this page again) 499670 Printed by the Consumers' Cooperative of Intellectual Property Bureau of the Ministry of Economic Affairs Α7 Β7 V. Description of the Invention (本) The purpose of the present invention is to provide a novel method and device for synthesizing acoustic signals. It can provide a low-distortion sound synthesis signal, and save the memory space occupied by the stored sound signal. [Brief description of the invention] 雠 The sound signal synthesis method and device of the present invention, the sound signal is sampled at a lower sampling frequency. However, during reproduction, an interpolation method is used to calculate the sound signal 两 during the two sampling cycles, and the calculation is inserted between the two sampling cycles, and then regenerated to obtain a reproduced sound with a reduced distortion rate. The invention provides a sound signal synthesizing method and device that greatly reduce distortion and does not need to be too high in sampling frequency. i: Refers to other objects and advantages of the present invention, which can be made more detailed by the following detailed description and reference to the following drawings. [Explanation of the drawing] Fig. 1 shows a schematic diagram of a sound waveform and a synthesized sound Iff 彡 sampled at a low sampling frequency. Fig. 2 is a diagram showing the sound waveform of Fig. 1 and the synthesized sound waveform obtained by sampling at a higher rate than the sampling frequency of Fig. 1. Figure 3 shows the sound waveform of Figure 1 and the sound signal synthesis method of the present invention. _ 5 This paper size is in accordance with Chinese National Standard (CNS) f specifications (210 X 297 public love) (Please read the precautions on the back before filling out this page) ^ J «H ϋ n I n ϋ n I n ϋ ϋ IIIIIIII n ϋ nnnn I n II n-499670 Printed by A7 B7, Consumer Cooperative of Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the invention (4) Figure 4 shows a flowchart of the sound signal synthesis method and device of the present invention. i 'Figure 5 shows a schematic diagram of the sound signal data sampled at the unit sampling frequency and the sound signal data played at 4 times the unit playback frequency. Table 1 shows the number of samples and playback sound signals at two sampling / playback frequencies. [Detailed description of the invention] The method and device for synthesizing a sound signal according to the present invention are described below with reference to the drawings. Fig. 3 is a schematic diagram showing the sound waveform of Fig. 1 and the synthesized sound waveform sampled by the sound signal synthesizing method and device of the present invention. As shown in Figure 3, after the sound signal shown in Figure 1 (1) is sampled at a frequency of XHz, if between two sampling periods, three pieces of sound signal sampling data are inserted to calculate 値, and then 4 times the reproduction frequency (Play back rate) Play (= 4XHz), the sound waveform output by J (J) will be as shown in (4) of Figure 3. Although this waveform is slightly different from the sound waveform sampled at the 4XHz sampling frequency, it is extremely different. _ Near, and can reduce the distortion generated when the sound is sampled. The sound signal waveform generated by this method requires only the same memory space as the sound signal data obtained at the viewing rate of XHZ, which is enough to store In addition, the sound signal synthesized by the method of the present invention can reduce the high and noisy noise generated by the PCM 1 sound synthesis process, and can effectively reduce the sampling due to the 6 national paper standards (CNS) A4. Specifications (210 X 297 mm) (Please read the precautions on the back before filling this page) Order i -line i ----- I --------- I ------- 499670 Α7 Β7 V. Description of the invention (Even) Original sound distortion caused by lower frequency ^ 1 [Embodiment] The following describes the embodiment of the sound signal synthesis method and device of the present invention with reference to the embodiment. Fig. 4 shows a flowchart of the sound signal synthesis method and device of the present invention. Fig. 5 shows a unit sampling frequency. The sampled sound synthesis waveform and the sound signal waveform obtained after the sound is synthesized at 4 times the unit playback frequency by the method of the present invention. The following description is based on the diagram. At (401), a unit sampling frequency (T) is used. Acquire the PCM coded data of a sound τ signal waveform every leap second to obtain a sound signal waveform data file (D), D = Di, D2, D3, " .Dn 〇 in (402), calculate the first and second of each group The difference of Di4 data 値 Δ / V ΔΑ = A -1. At (403), divide this by 4 to get I plane 4 △ Di4, Δϋί4 = taxi ADi. At (404) insert three between Di and Dw Each sound 4 signal D # is ΔΔ, ΔΑ and DH + swallows ΔΑ. In this way, a sound signal file (D,) with a sampling frequency of 4 4 4 (play) is obtained. At (405), the frequency is 4T, according to the obtained The content of the sound signal file D 'plays the sound. At this time The playback time of each unit is ^. 4T [Effect of the invention] 7 This paper size is applicable to China National Standard (CNS) A4 (210 X 297 mm) (Please read the precautions on the back before filling this page) ·· ,-* ϋ i_i n -ϋ n aal · ϋ an emm9 n Printed line of employee consumer cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 丨 · ---------------- Printed 499670 A7 B7 V. Description of the invention (g) Figure 5 shows a sound signal data 'i' sampled at a unit sampling frequency;--| and a schematic diagram of sound signal data played at 4 times the unit playback frequency. Table 1 shows the number of samples and playback sound signals under two sampling / playback frequencies. The display uses the increased playback frequency and uses interpolation to obtain the required sound signal, which can obtain a waveform closer to the original sound. In the present invention, because the difference between each piece of PCM coded data is greatly reduced ', the background high-frequency noise generated during sound synthesis can be effectively reduced. At the same time, because the obtained waveform is relatively close to the original sound waveform, the method of the present invention can indeed improve the quality of sound signal playback. Although the above description inserts three pieces of sound signal data into two sound signals in a four-division manner, it is also possible to insert sound signal data in more or less equal divisions. In addition, instead of inserting the sound signal data in an unequal manner, in some embodiments I, a similar or superior effect can be obtained. The foregoing is an explanation of the embodiments of the method and device for synthesizing a sound signal according to the present invention. Those skilled in Siyi should know the spirit of the present invention from the above description, and make various changes and derivatives based on it. As long as it does not depart from the spirit of the present invention, it should be included in the scope of its application for a patent for invention.
I 8 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)I 8 This paper size applies to Chinese National Standard (CNS) A4 (210 X 297 mm) (Please read the precautions on the back before filling this page)