TW442740B - Method for changing articulation speed - Google Patents
Method for changing articulation speed Download PDFInfo
- Publication number
- TW442740B TW442740B TW87121166A TW87121166A TW442740B TW 442740 B TW442740 B TW 442740B TW 87121166 A TW87121166 A TW 87121166A TW 87121166 A TW87121166 A TW 87121166A TW 442740 B TW442740 B TW 442740B
- Authority
- TW
- Taiwan
- Prior art keywords
- signal
- speed
- voice
- sound
- recording medium
- Prior art date
Links
Landscapes
- Signal Processing For Digital Recording And Reproducing (AREA)
Abstract
Description
442740 五、發明說明（1) 【發明的應用範圍】 本發明係有關一種改變發音速度的方法，應用於數位 化之語音資料的發音處理，用以在對數位化之語音資料進 行發音速度的改變後，不會使其發音之音調失真的方法。 【發明背景】 請參閱「第1圖」，無論是Microsoft開發的 ActiveMovie ’ MCI，還是其它公司開發的語音編緝軟件， 其在電腦中對語音的採集、存儲、播放的方式，是將各種 音源產生設備（如：麥克風、卡式錄音機等）1 〇，所產 生的語音信號，藉由一語音處理單元（如：語音卡）2 〇 對語音信號進行採樣，並透過邏輯處理單元3 〇轉換成相-對應之數位化的語音信號，請參閱「第2圖」，數位化的 語音k號4 0係由複數個音元信號段41、5 1、6 1所 組成’而且每個音元信號段4 1更包含有複數個信號採 點4 1 1，最後再將此數位化之語音信號4 〇存入一 媒體5 0的語音文件中；在播放語音時，只要將語杜 中的每個音兀信號段4 1内的信號採樣點4丄 出到語音處理單元3"，再由語音處理 信號採樣點4 i i放大輸出到聲音輸..出單元 = 聲音輸出單元6 0發出可聽到之聲音訊號^ 由 而其中與發音有密切關係的數據是信號採樣 1’信號採樣點4 1 1是按照預先設定 語音信號（係指由麥克風或卡式錄音機等設者= 行採樣，再將由這些信號接掸 Λ ， 叹價屋生者）進 琥铋樣點411所組成的音元信號442740 V. Description of the invention (1) [Scope of application of the invention] The present invention relates to a method for changing the pronunciation speed, which is applied to the pronunciation processing of digital voice data, and is used to change the pronunciation speed of digital voice data. Method that does not distort the pitch of its pronunciation. [Background of the Invention] Please refer to "Figure 1". Whether it is ActiveMovie 'MCI developed by Microsoft, or voice editing software developed by other companies, the way of collecting, storing, and playing voice in the computer is to convert various audio sources. A generating device (such as a microphone, a cassette recorder, etc.) 1 〇, the generated voice signal is sampled by a voice processing unit (such as a voice card) 2, and converted into a logical processing unit 3 〇 Phase-corresponding digitized voice signal, please refer to "Figure 2". The digitized voice k number 40 is composed of a plurality of vowel signal segments 41, 5 1, 6 1 'and each vowel signal Segment 4 1 further includes a plurality of signal acquisition points 4 1 1, and finally the digitized voice signal 4 0 is stored in a media 50 voice file. When playing a voice, as long as each The signal sampling point 4 in the sound signal segment 41 is output to the speech processing unit 3, and then the speech processing signal sampling point 4 ii is amplified and output to the sound output .. The output unit = the sound output unit 6 0 emits an audible sound Signal ^ Therefore, the data closely related to pronunciation is signal sampling 1 'signal sampling point 4 1 1 according to the preset voice signal (referred to by the microphone or cassette recorder, etc. = line sampling, and then these signals are connected to 掸 Λ , The sighing room) phonon signal composed of bismuth sample point 411
442740 五、發明說明（2) '"""" 段4 1經過處理後存入記錄媒體5 〇内的語音文件中。然 後再，與採樣頻率相同的頻率通過語音處理單元3 〇將這 些信號採樣點還原播放之。在目前的語音信號的格式中 22kHz、8bit的格式為單聲道收音機音質，44kHz、丨 ^式為立體聲CD音質；其中2驗（44kHz)就是指採樣 頻率，8blt (16bit)就是指存放一個信號採樣點4工丄 的位兀Ϊ/、而語音處理單元3 0就是以-既定的播 ' 根據月'j述的語音格式來播放聲音，且立體聲CD音 為一…聲道收音機音質的二 极ί ί ί變語音發音的方法，是以每個信號採樣點4 1 it:，$行信號採樣點411複製或删減以實 加快或減慢。因此如果要將原語音的播放 採二二都；個中的每個信號 :ί在號段41的波形週期就被拉長-倍， 來的也立m f #果保持採樣頻率不變，則播放出 :十同時聲音就會變低、變粗。請 如£1所_#為1古為原始之音几信號段4 1 1的波形圖， 如圖所不係為含有一幅度為156的採樣作 時 =!?信號段41，今若要以慢」倍：速度播放音元 據前述之傳统變迷處理方式，就須i 曰兀“唬1中的每個信號採樣點 將複製後的信號採樣點41“***音心…中並442740 V. Description of the invention (2) '" " " " Paragraph 4 1 After processing, it is stored in the voice file in the recording medium 50. Then, these signal sampling points are restored and played back by the speech processing unit 30 at the same frequency as the sampling frequency. In the format of the current voice signal, the format of 22kHz and 8bit is the sound quality of the mono radio, and the format of 44kHz and ^^ is the sound quality of the stereo CD; where the 2 test (44kHz) refers to the sampling frequency and 8blt (16bit) refers to the storage of a signal The sampling point is 4 bits, and the voice processing unit 30 is to play the sound according to the speech format described in "Preset Broadcasting", and the stereo CD sound is a two-channel radio quality. ί ί The method of changing the pronunciation of a voice is to copy or delete the signal sampling point 411 for each signal sampling point 411 to speed up or slow down. Therefore, if you want to play the original voice, use two or two; each signal: ί The waveform period at number 41 is stretched-times, and the next one is mf # If the sampling frequency is maintained, the playback Out: At the same time, the sound will become lower and thicker. Please refer to the waveform diagram of the signal section 4 1 1 for the original sound, such as £ 1. As shown in the figure, it is not a sample with an amplitude of 156. = !? Signal section 41. "Slow" times: according to the traditional obfuscation processing method described above, it is necessary to say that "every signal sampling point in Bluff 1 will insert the copied signal sampling point 41" into the heart ...
442740 五、發明說明（3) 置於原6號採樣％ 4 1 1 ^ 信號段4 1 a將^「^1的後面，那麼經過處理後的音元 鄰且採樣頻率相同的所㊉’包含有多組兩兩個相 如果按照預定的It:;採樣點411、4113 ’所以 Φ ® ^ ^ ^ ^诛樣頻率進行聲音的還原和播放，則原來 十^ 一個振動週期的音元信號段4 1，就變成 — /1疋成一個振動週期的音元信號段41a ; 廷樣一來，語音的播妓— + Λ 们播放速度固然減慢了’但由於改變了原 t聲：！振動週期和頻率，所以語音就產生了變調的現 象。这疋因為假如原本是以221^2採樣頻率錄製的音元信 號段4 1，經過上述的處理後就轉換成了 一個以44kHz採 樣頻率iflL的音元信號段41 a，但是由於仍是按原來的 22kb/s速度播放，所以還原後的聲音的頻率比錄製時慢 了一借，再加上發聲的聲調與聲波旳振動頻率有直接的關 係’所以就會出現變調的現象。 【發明欲解決之問題】 目前的語音變速的技術，在對原語音文件的採樣信號 進行處理的過程中，改變了還原後語音聲波的振動頻^广 所以會出現變調的現象；因此目前的語音變速技術不論是 頻率變低或變高，均會在變速後使聲音變得模糊不清w $ 成使用者在聽覺上的不悦。特別是在進行語言教學過程 中，學習者一般都對口語和聽力學習感覺困難。其中—部 份原因是對方說話的語速過快，初學者來不及反應。如^ 能夠將聲音的速度減慢將可以大大提高訓練的效果。 【發明之概述】442740 V. Description of the invention (3) Placed on the original sample No. 6 4 1 1 ^ The signal segment 4 1 a will be after ^ "^ 1, then the processed phonons are adjacent to each other and have the same sampling frequency. If two groups of two phases are used to restore and play the sound according to the predetermined It :; sampling points 411, 4113 'so Φ ® ^ ^ ^ ^ 诛 sample frequency, the original ten ^ one vibration period of the vowel signal segment 4 1 , It becomes — / 1, a vowel signal segment 41a that oscillates into a vibration cycle; in the same way, the voice of the prostitute — + Λ is slowed down, but because the original sound is changed: the vibration cycle and Frequency, so the voice has a tone change phenomenon. This is because if the original phonon signal segment 41 was recorded at the 221 ^ 2 sampling frequency, after the above processing, it will be converted into a phonon with a sampling frequency of 44kHz iflL. The signal segment 41 a, but because it is still playing at the original 22kb / s speed, the frequency of the restored sound is a bit slower than when recording, plus the tone of the sound has a direct relationship with the frequency of the sound wave 旳 vibration ' There will be a phenomenon of tone change. [Invent The problem to be solved] The current technology of voice speed change, in the process of processing the sampled signal of the original voice file, changes the vibration frequency of the restored sound wave so that the phenomenon of tone change will occur; therefore, the current technology of voice speed change Regardless of whether the frequency becomes lower or higher, the sound will become blurred after changing the speed w $ becomes a user's hearing dissatisfaction. Especially in the process of language teaching, learners are generally speaking and listening Learning is difficult. Part of the reason is that the other party speaks too fast, and beginners have no time to respond. For example, ^ can slow down the speed of the voice will greatly improve the training effect. [Overview of the invention]
C:\Program Files\Patent\P-229TW.ptd 第 6 頁 442 7 4 Ο 五、發明說明（4) 本發明的主要目的在於提出 速度的快速播放或者慢速播放時 法’使得在調整語音的播放速度 不變、聲音不失真。 本發明處理語音變速的原理 的每個信號採樣點4 1 χ作為複 疋以原#音信號4 〇中的音元作 遇期）41作為一個基 參閱「第3圖」和「第5圖」， 4 1具有較標準之播放速度慢一 圖」中曰元k號段4 1進行複製 段4 1 a置於原音元信號段4 1 信號段42 (如「第5圖」所示 0以原來取樣頻率的播放速度進 放，這樣一來就不會改變每個音 原有頻率’而且還可在改變語音 語音的語調（頻率）。 有關本發明之詳細内容及技 下： 【圖式簡單說明】 第1圖，為語音變速播放處理裝 第2圖’為語音信號的波形圖。 第3圖、為以原始之音元信號段 第4圖’為以傳統方法經慢速播 一種對語音 ’不會出現 後，語音清 信號進 變調現 淅、語 ’並不是以 製或刪減的 「第1 基本單 號.段（即一個完整 製或刪 的音元 ，係對 後的音 ’來進行複 若要使輸出 倍的效果時 ，並把複製 的後面，構 )，再由語 行音元信號 元信號段4 播放速度後 成一新 音處理 段4 21、4 ，仍維 行任意 象的方 調保持 圖」中 元，而 的振動 減。請 信號段 「第3 元信號 的音元 單元3 的播 1 a的 持原来 術，茲就配合圖式說明如 置的方塊圖 的波形圖° 放處理後的 波形圖C: \ Program Files \ Patent \ P-229TW.ptd Page 6 442 7 4 〇 V. Description of the invention (4) The main purpose of the present invention is to propose a method of fast playback or slow playback of the speed, so as to adjust the voice The playback speed does not change and the sound is not distorted. Each signal sampling point 4 1 χ of the present invention that handles the principle of speech shifting is used as the complex period, and the period of the original # tone signal 4 〇 is used as a reference period. 41 is used as a basis. See "Figure 3" and "Figure 5" , 4 1 has a picture that is slower than the standard playback speed. ”In the Chinese paragraph k section 4 1 copy section 4 1 a placed in the original sound signal section 4 1 signal section 42 (as shown in the" Figure 5 "0 to the original The playback speed of the sampling frequency is advanced, so that the original frequency of each tone will not be changed, and the tone (frequency) of the speech can also be changed. Details and techniques related to the present invention: [Schematic illustration ] Figure 1, for voice variable-speed playback processing, Figure 2 is the waveform diagram of the voice signal. Figure 3, the original vowel signal segment When it does not appear, the voiceless signal is changed into tone, and the language 'is not based on the "basic number 1. paragraph (ie, a complete or deleted syllable, which is the complete syllable). When you want to make the output double the effect, and put the copy on the back, construct) Then the speech line signal element signal segment 4 plays a new tone processing segment 4 21, 4 after playing speed, and still maintains the square tone retention map of any image ", while the vibration is reduced. Please signal segment" element 3 The original technique of broadcasting 1a of the phonon unit 3 of the signal will be explained with reference to the waveform diagram of the block diagram as shown in the figure.
C:\ProgramFiles\Patent\P-229TW.ptd 第 7 頁 442740 五、發明說明（5) 第5圖’為第3圖之音元信號段經慢一倍之速度播放處理 後的波形圖。 第6圖’為第2圖之語音信號經慢一倍之速度播放處理後 的波形圖。 第7圖，為第2圖之語音信號經慢二分之一倍之速度播放 處理後的波形圖 第8圖，為第2圖之語音信號經快一倍之速度播放處理後 的波形圖。 第9圖，為結構鏈表的示意圖。 第1 0 — 1圖，為本發明處理語音變速播放之方法的部份 流程圖。 第1 0 — 2圖，為本發明處理語音變速播放之方法的 流程圖。 ， 。刀 第1 0 — 3圊，為本發明處理語音變速播放之方法的 流程圖。 β切 【發明之詳細說明】 請參閱「第2圖」，本發明所採用的方法是在進> $ 音信號4 0的變速播放時，不是複製或刪除語音作號=， 中的每一個信號採樣點4 i X ，而是根據要將語^ := 0以加快或變慢方式播放的要求，對其内的音元信^二 (聲波的一個完整振動週期）4 1作複製或删除&動= 所以在對語音信號4 0作變速播放的處理之前， ， 找出語音信號4 0中的每個音元信號段4 1 ，以下g 、 決定語音信號内之音元信號段的條件： 匈C: \ ProgramFiles \ Patent \ P-229TW.ptd Page 7 442740 V. Description of the invention (5) Figure 5 'is a waveform diagram of the vowel signal segment of Figure 3 after being processed twice as slowly. Fig. 6 'is a waveform diagram of the voice signal of Fig. 2 after being processed twice as slowly. Fig. 7 is a waveform diagram of the voice signal of Fig. 2 after being played at half the slower speed. Fig. 8 is a waveform diagram of the voice signal of Fig. 2 after being played at a speed twice as fast. Figure 9 is a schematic diagram of a structured linked list. Figures 10-1 are partial flowcharts of the method for processing variable-speed speech playback of the present invention. Figs. 10 to 2 are flowcharts of a method for processing variable-speed speech playback according to the present invention. ,. Knife Nos. 10 to 3 are flowcharts of the method for processing variable-speed speech playback of the present invention. β cut [Detailed description of the invention] Please refer to "Fig. 2". The method adopted in the present invention is not to copy or delete each of the voice marks during the variable-speed playback of the> $ tone signal 40. The signal sampling point 4 i X, but according to the requirement to play the language ^: = 0 in a faster or slower way, copy or delete the phonetic letter ^ 2 (a complete vibration cycle of the sound wave) 4 1 & Dynamic = So before processing the variable speed playback of the speech signal 40, find each phonetic signal segment 4 1 in the speech signal 40, and the following g determines the condition of the speech signal segment in the speech signal : Hungary
;44274ο 五、發明說明（6) i .疼個音元信號段的起始點4 4 須疋中心點或者它和它的一 止點4 5的必 中心線4 6彳目a 、，，下一個信號採樣，點組成的連線與 r。琢4D相父，並且起拎 號與它們下一個_號採M ^ "點4 5的採樣信 或同為下降趨勢樣點組成的變化趨勢同為上升趨勢 該以/“Hz起為始丄和終止點4 5之間在時間上的間良應 終止點間的時間相隔為2-3毫秒。 P起-始點與 右一1的二元信號段和鄰近的下一個音元信號段，應 ΐ = : Ρ兩個音元信號段的中心線4 6以上的 =-大：作：線4 6以下的最小值之間的差距小於中心線 到取大變化範圍的十分之—。 4不滿足以上條件的不能作為一個音元信號段，而 且對於不滿足條件的數據在語音變速處理時保持不變，既 不複製也不刪減。 请參閱「第1〇 一丄圖」至「第1〇 — 3圖」，為本 發明語音變速播放處理的流程圖，其變速的處理步騍依序 為： 步轉A 1 ..於數位化之語音信號中，以比較每兩個信 號採樣點4 1 1的方式進行掃描，並將所有中心綠4 6上 的信號採樣點4 1 1 ，或與其後之信號採樣點的連線與中 心線4 6相交的採樣點’以及所有拐點（即指波峰、波谷 的轉折點）的信息記錄到一個結構鏈表4 7内，其中每個 鍵表4 7 1的結構如表一所示；.44274ο V. Description of the invention (6) i. The starting point 4 4 of the vowel signal segment must be the center point of it or its one-stop point 4 5 and the necessary center line 4 6 彳 目 a ,,,, A signal sample, a line of points and r. 4D phase fathers, and the 拎 number and their next _ number to take M ^ " point 4 5 sampling letter or the same as the downward trend, the change trend of the sample point composition is the same as the upward trend should start with / "Hz" The time between the time between the termination point 45 and the termination point 5 should be 2-3 milliseconds. The P-start point is the binary signal segment from the right to the first 1 and the next next sound signal segment, Should ΐ =: Ρ above the center line of the two tone signal segments 4 6 =-large: operation: the difference between the minimum value below the line 4 6 is less than the center line to take a tenth of a large change range.-4 not Those that meet the above conditions cannot be used as a vowel signal segment, and the data that does not meet the conditions remains unchanged during speech shift processing, and is neither copied nor deleted. Please refer to the "No. 10 figure" to "No. 1" 〇-3 "is a flowchart of the voice variable speed playback processing of the present invention. The sequence of the variable speed processing steps is: Step A 1 .. in the digital voice signal to compare every two signal sampling points 4 1 1 to scan, and all the signal sampling points 4 1 1 on the center green 4 6 or The information about the connection point of the signal sampling point and the center line 4 6 'and the information of all inflection points (that is, the turning points of the crests and troughs) are recorded in a structural linked list 4 7 of which each key table 4 7 1 The structure is shown in Table 1.
442 74 Ο442 74 Ο
表一、鏈表的結構 號採樣點與中心線之間的差值 if號採樣否為中心點 耜始點的 ^£~^ψ 趕.'表結構的指$：~ 步驟A 2 . 在兩個相鄰的中 遠的拐點； 於結構鏈表4 6中濾除多餘的拐點記錄， 心點間最多只保留一個距離中心線4 5最 一個上升趨 步驟A 3 .從結構鏈表4 6的頭向後尋找 勢或下降趨勢的中心點； 步驟A4.判斷是否存在一個上升趨勢或下降趨勢的 '^點若為是跳至步A6，若為否執行下一步驟； 步驟A5·尋找下一個為上升趨勢或下降趨勢的中心 點，並跳至步驟A 4 ; 步騾A 6 .判斷是有中心點的記錄，若為是執行下一 步驟；若為否執行步驟A 8 ; 步驟A 7 .記錄中心點的記錄，並跳至步驟a 9 ; 步驟A 8 .記錄中心點的位置於記錄媒體中.； 步驟A 9 判斷記錄媒體中是否有兩個具有相同特徵 的中心點，若為是跳至夕驟A i丄，若為否，執行下一 驟；Table 1. The difference between the sample number of the structure number of the linked list and the center line. If the sample of if number is ^ £ ~ ^ ψ of the starting point of the center point, hurry. 'The structure of the table means $: ~ Step A 2. Two adjacent COSCO inflection points; filter out the excess inflection point records in the structure link list 46, at most, only one distance from the center line 45 to the heart point is the most ascending step A 3. From the head of the structure link list 4 6 Step backward to find the center point of the trend or down trend; Step A4. Determine if there is a '^ point of an up trend or down trend. If yes, go to step A6, if no, go to the next step; Step A5. Find the next up The center point of the trend or downtrend, and skip to step A 4; Step 骡 A 6. Judge that there is a record of the center point, if yes, go to the next step; if no, go to step A 8; Step A 7. Record Center Step A 9; Step A 8. Record the position of the center point in the recording medium; Step A 9 Determine whether there are two center points with the same characteristics in the recording medium, if it is skipped to the evening Step A i 丄, if not, go to the next step;
C:\Program Files\Patent\P-229TW. ptd 第 10 頁 442 74 Ο 五、發明說明（8) 步驟A 1 0 一步驟，若為否 步驟A 1 1 步驟A 1 2 時間上的間隔； 步驟A 1 3 行下一步騍，若 步驟A 1 4 音元信號段，並 步驟A 1 5 號段； 步驟A 1 6 大點的偏移值， 線與最大值的 出狀態，在變 至步驟1 9， 步驟A 1 大點的偏移值 大值的偏移量 •判斷是否全部 跳至步騍A 5 ; .计算兩個中心 •再根據採樣頻 •判斷間隔是否 為否則跳至步驟 .將兩個中心點 記錄到一個臨時 .重覆步驟8〜 搜尋完畢，若 為 是執行 下 :比較次一個音 是否遠遠小於前 移量，若為是， 處理時將不對此 為否’執行下一 比較次一個音 ，是否與前一個音 近似，若為是，跳 偏 速 若 7 行下一步驟 步驟A 基準，跳至 步驟A 了比較辨認 點之間的偏移量. 率計算出兩個中心點之間 h於2至3亳秒 A 5 ； 右馬是執 間的信號採樣點作為— 的記錄媒體中；為個 14尋找出次—個音元信 兀信號段中’中心線與最 —個音元信號段中，中心 則可以認定此為語音的淡 段聲音做特殊處理，並跳 步驟； 元信號段中，中心線與最 元信號段中，中心線與最 至步驟19，若為否，執C: \ Program Files \ Patent \ P-229TW. Ptd Page 10 442 74 Ο V. Description of the invention (8) Step A 1 0 One step, if not Step A 1 1 Step A 1 2 Time interval; Step Step A 1 to the next line. If step A 1 4 is the vowel signal section and step A 1 5; step A 1 6 is the offset value of the large point, and the state of the line and the maximum value is changed to step 1. 9. Step A 1 Large point offset value Large value offset • Determine whether all jump to step 骒 A 5;. Calculate two centers • Then according to the sampling frequency • Determine if the interval is otherwise skip to step. A center point is recorded to a temporary. Repeat step 8 ~ After the search is completed, if it is executed, compare whether the next note is far less than the amount of forward movement. If so, do not perform this comparison when processing. The next tone, is it similar to the previous tone, if yes, skip speed if 7 lines Next step step A benchmark, skip to step A to compare the offset between the identified points. The rate calculates the two center points H between 2 and 3 leap seconds A 5; the right horse is the record of the signal sampling point as — In the media; find out the center line and the most phonetic signal segment of the sub-single signal segment for each of the 14; the center can identify this as the light voice of the voice to do special processing and skip the steps; In the meta signal segment, the center line and the most meta signal segment, the center line and the most up to step 19, if not, execute
1 8 ·以第一個音元信號段的第二個中心點 步驟A 5 ; 1 9 ·判斷結構表中的所有記錄點是否都經過 ，若為是執行下步驟，若為否，則跳至步驟A1 8 · Step A 5 with the second center point of the first vowel signal segment; 1 9 · Determine whether all recorded points in the structure table have passed. If yes, go to the next step. If no, skip to Step A
442 7 4 0 五、發明說明（9)5 ； 步驟A 2 0 -確定語音信號中音元信號段； 步驟A 2 1 ·根據發音速度的設定將所有的音元信號 段於一記錄媒體中進行複製； 步驟A2 2 ·以語音處理單元2 0，將複製於記錄媒 體中的音元信號段轉換成可聽之聲音訊號； 步驟A 2 3，判斷是否已處理完所有複製後的音元信 號段，若為是，執行步驟A2 5，若為否，執行下一步 驟； 步驟A 2 4 ·取出下一筆複製後的音元信號段，並跳 至步驟A 2 2 ;以及 步驟A 2 5 ·將語音處理單元2 ◦置於等待狀態。 在上述步驟A2 1中，若所設定發音速度較標準的發 音速度的慢一倍，則「第2圖」中的語音信號4 0經處理 後將如「第6圖」所示，將每個音元信號段4 1 、5 1 、 6 1在記錄媒體中做兩次的複製，於是在原來的每個音元 信號段4 1、5 1、6 1之後將分別產生音元信號段4 1 a、5 1 a、6 1 a ;但是若所設定發音速度較標準的發 音速度慢二分之一倍，則會如「第7圖j所示，將語音信 號中奇數的音元信號段4 1、.6 1 ，在記錄媒體中做兩次 的複製，產生音元信號段41a、61a ，偶數的音元信 號段5 1 ，則在記錄媒體中只做一次的複製；再者，請參 閱「第8圖」，若是設定發音速度較標準的發音速度快一 倍，則是每隔一個音元信號段，在記錄媒體中複製一個音442 7 4 0 V. Description of the invention (9) 5; Step A 2 0-Determine the vowel signal segments in the speech signal; Step A 2 1 · Perform all vowel signal segments in a recording medium according to the setting of the pronunciation speed Copy; Step A2 2 · Use the voice processing unit 20 to convert the vowel signal segments copied in the recording medium into audible sound signals; Step A 2 3 to determine whether all the copied vowel signal segments have been processed If yes, go to step A2 5; if no, go to next step; Step A 2 4 · Take out the next copied vowel signal segment and skip to step A 2 2; and step A 2 5 · will Voice processing unit 2 ◦ Put on standby. In the above step A21, if the set pronunciation speed is twice as slow as the standard pronunciation speed, the voice signal 40 in the "picture 2" will be processed as shown in "picture 6", and each The phonetic signal segments 4 1, 5 1, and 6 1 are duplicated twice in the recording medium, so that each of the original phoneme signal segments 4 1, 5 1, and 6 1 will be separately generated after each of the phoneme signal segments 4 1 a, 5 1 a, 6 1 a; However, if the set pronunciation speed is one-half times slower than the standard pronunciation speed, as shown in "Figure 7 j, the odd number of vowel signal segments in the speech signal will be 4 1, .6 1, making two copies in the recording medium to generate the vowel signal segments 41a, 61a, and the even number of vowel signal segments 5 1, then making the copy only once in the recording medium; moreover, see "Figure 8", if the pronunciation speed is set to be twice as fast as the standard pronunciation speed, then every other phoneme signal segment is copied in the recording medium
C:\ProgramFiles\Patent\P-229TW.ptd 第 12 頁 442 74 Ο 五、發明說明αο) 元信號段’也就是只對在該語音信號中為奇數壙位的音元 信號段4 1、6 1進行複製，便可以實現語音的快速播 放。 以上所述僅為本發明之較佳實施例，並不限於以上述 之硬體的裝置實施’舉凡任何熟習此項技藝者在本發明之 領域内所做的任何修飾’具有同等之功效者，均 下列之申請專利範圍内。 ‘ 【發明之效果】 本發明之方法對各種格式的5吾音文件都可以做變速處 理’以使得在調整語音的播放速度後，所產生的語音清 淅、語調保持不變、聲音不失真。 【圖示符號說明】 1 0音源產生設備 20語音處理單元 3 0邏輯處理單元 4 0語音信號 4 1音元信號段 4 1 a音元信號段 4 1 1信號採樣點 4 1 1 a信號採樣點 4 2音元信號段 4 4起始點 4 5終止點 4 6中心線C: \ ProgramFiles \ Patent \ P-229TW.ptd Page 12 442 74 Ο V. Description of the invention αο) Meta signal segment 'that is, only for the vowel signal segment which is an odd number of bits in the voice signal 4 1, 6 1 Make a copy to achieve fast voice playback. The above is only a preferred embodiment of the present invention, and is not limited to the implementation of 'any modification made by any person skilled in the art in the field of the present invention' with the hardware device described above, which has equivalent efficacy, Within the scope of the following patent applications. [Effects of the invention] The method of the present invention can perform variable-speed processing on various vocal files in various formats' so that after adjusting the playback speed of the voice, the resulting voice is clear, the tone remains unchanged, and the sound is not distorted. [Illustration of symbols] 1 0 sound source generating device 20 voice processing unit 3 0 logic processing unit 4 0 voice signal 4 1 phoneme signal segment 4 1 a phoneme signal segment 4 1 1 signal sampling point 4 1 1 a signal sampling point 4 2 phonetic signal segment 4 4 start point 4 5 end point 4 6 center line
C:\Program Files\Patent\P-229TW. ptd 第 13 頁 4 42 74 Ο 五、發明說明αu 4 7結構鏈表 4 7 1鏈表 5 0記錄媒體 5 1音元信號段 5 1 a音元信號段 6 0聲音輸出單元 6 1音元信號段 6 1 a音元信號段 C:\Program Files\Patent\P-229TW. ptd 第 14 頁C: \ Program Files \ Patent \ P-229TW. Ptd Page 13 4 42 74 〇 V. Description of the invention αu 4 7 Structure linked list 4 7 1 linked list 5 0 recording medium 5 1 phoneme signal segment 5 1 a phoneme Signal section 6 0 sound output unit 6 1 phoneme signal section 6 1 a phoneme signal section C: \ Program Files \ Patent \ P-229TW. Ptd page 14
Claims (1)
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW87121166A TW442740B (en) | 1998-12-18 | 1998-12-18 | Method for changing articulation speed |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| TW87121166A TW442740B (en) | 1998-12-18 | 1998-12-18 | Method for changing articulation speed |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| TW442740B true TW442740B (en) | 2001-06-23 |
Family
ID=21632364
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| TW87121166A TW442740B (en) | 1998-12-18 | 1998-12-18 | Method for changing articulation speed |
Country Status (1)
| Country | Link |
|---|---|
| TW (1) | TW442740B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103258552A (en) * | 2012-02-20 | 2013-08-21 | 扬智科技股份有限公司 | Method for adjusting play speed |
| CN110798327A (en) * | 2019-09-04 | 2020-02-14 | 腾讯科技（深圳）有限公司 | Message processing method, device and storage medium |
| CN114363713A (en) * | 2022-01-12 | 2022-04-15 | 维沃移动通信有限公司 | Sound adjusting method and device |
-
1998
- 1998-12-18 TW TW87121166A patent/TW442740B/en not_active IP Right Cessation
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103258552A (en) * | 2012-02-20 | 2013-08-21 | 扬智科技股份有限公司 | Method for adjusting play speed |
| CN103258552B (en) * | 2012-02-20 | 2015-12-16 | 扬智科技股份有限公司 | The method of adjustment broadcasting speed |
| CN110798327A (en) * | 2019-09-04 | 2020-02-14 | 腾讯科技（深圳）有限公司 | Message processing method, device and storage medium |
| CN114363713A (en) * | 2022-01-12 | 2022-04-15 | 维沃移动通信有限公司 | Sound adjusting method and device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9601029B2 (en) | Method of presenting a piece of music to a user of an electronic device | |
| Zattra | The Assembling of" Stria" by John Chowning: A Philological Investigation | |
| Crockett | High quality multi-channel time-scaling and pitch-shifting using auditory scene analysis | |
| TW442740B (en) | Method for changing articulation speed | |
| Kane | Relays: Audiotape, material affordances, and cultural practice | |
| US20040182228A1 (en) | Method for teaching individual parts in a musical ensemble | |
| JP4994890B2 (en) | A karaoke device that allows you to strictly compare your recorded singing voice with a model song | |
| Komara | Édouard-Léon Scott de Martinville, Inventor of Sound Recording: A Bicentennial Tribute | |
| JP3809537B2 (en) | Language learning system | |
| Jones | Rock formation: popular music and the technology of sound recording | |
| Haley | The Complete Josef Lhevinne | |
| JP2000099308A (en) | Electronic book player | |
| TW200521898A (en) | Formation method of interactive learning software/firmware review program | |
| Haley | Black Swans | |
| Bultmann | New Music String Quartet: The Complete Columbia Album Collection | |
| Feaster | Phonography and the recording in popular music | |
| Colby | Sound Recordings in the Music Library: With Special Reference to Record Archives | |
| Lucia et al. | Jürgen Bräuninger Remembered | |
| Galo | Stokowski: The Acoustic Recordings–Volumes 1–4 | |
| Wahl | Recording the Classical Guitar | |
| Miller | The influence of recording technology on music performance and production | |
| Haley | Sound Recording Reviews in Brief | |
| JPS6032678Y2 (en) | sheet music sheet | |
| Lewis | Brahms: Recaptured by Pupils and Colleagues | |
| Dankner | Edison, Musicians, and the Phonograph: A Century in Retrospect |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| GD4A | Issue of patent certificate for granted invention patent | ||
| MM4A | Annulment or lapse of patent due to non-payment of fees |