550476 A7 2 經濟部智慧財產局員工消費合作社印制衣 五、發明說明( 本發明之領域及背景; 本發明是利用於多媒體產生動畫之領域,特別是驅動具有 唇形之圖形。 習知在多媒體製作中,如電腦動畫或如利用電腦製作之卡 通謝,由於虛擬之人物甚至是動物有相當多需要説話 <場合,而如何以逼眞之方式控制嘴巴之動作及嘴巴附近 之肌肉變化是一項非常耗時且耗力之工作。 目前可行之方式係眞將複數之感應偵測器貼在—眞膏人的 嘴巴附近之肌肉,利用感應偵測器去感應嘴巴肌肉之變 彳匕,而再將此數據驅動具有唇形之圖形。 當然這是一種相當好之方法,但是有如下之缺點:. 1 ·此設備非常昂貴,且其製作流程複雜,需有專業之人士 協助,一般僅適用於大型之製片公司,或是遊戲軟體製 造大公司。 不適用於電腦動畫中,特別是遊戲軟體,因為所有之且 有唇形之圖形經製作後,其動作(包括嘴巴之肌肉變/、 化)即已固定。然而如遊戲軟體,圖形之動作會根據使 用者輸入之不同而有所改變,因此在遊戲軟體中,若採 用習知方法,至多僅能以預先儲存之資料庫作為模擬, 但若對於無法預期之文字發聲時,則無法模擬;譬如使 用者輸入又字資料或是使用者輸入一段文字經由發聲引 擎產生文字資料,則習知方法根本無法應用。 本紙張尺度適用中國國家標準(CNS)A4規格⑵〇x 297公爱- 雇- ^--------^--------- (請先閱讀背面之注意事項再填寫本頁) 550476 A7 号务明說明(二) 3 ·不適用於轉換不同夕丄 、 ,譬如一電腦動畫原本是以中 文製作’當要將其雷 /、%恥動畫内部之文字發聲轉變為英文 發聲,貝1J依習知之大、土 , 乃决,有關嘴巴附近肌肉之控制資料 (請先閱讀背面之注意事項再填寫本頁) 必須重新製作,膏ρχ /、1令上吓不可行。 去邊j之簡果: ·- 本發明之主要目的係te也 ^ ^ 供一間易之方式利用文字資料直接 驅動具有唇形之圖形。 本發明之再一目的係可划田 士、 一 j竹」利用現有邵分之技術,而能在不同 ¥吾言之地區仍能輕易利用本發明之方法。 本Μ月I更目的係適用於各種電腦動畫,包括電腦動畫 電影,卡通影片,遊戲軟體之製作,甚至如互動式之遊戲 體等。 經濟部智慧財產局員工消費合作社印製 本發明之又-目的係適用於需要轉換不同語言之場合,由 於本發明利用文字資料直接驅動具有唇形之圖形,因此僅 需要改變文字資料以及内部之配套軟體程式即可。. 為完成本發明之主要目的,本發明之方法 料^中本發明在較佳之應用中是针對文字資枓之内容為 言敘述之文字資料;接著再將文字資料轉成複數之 曰私付號組,其中各晋標符號組相對代表一個字,而各立 麵組包括單一或複數之音標符號;在得到音標符^ 本紙張尺度適用中國國家標準(CNS)A4規格(210 X 297公爱) 經濟部智慧財產局員工消費合作社印製 550476 A7 B7 五、号务明說明(3 ) 後1,依據骨標-唇形轉換表將各晋標符號組轉成唇形指令, 最後依據唇形指令控制具有唇形圖形之動作。 其中在較佳之實施利中文字分析引擎負責將文字資料轉成 音標符號及合成語音,則該文字分析引擎主要包括產生音 才票符號之語意語境分析引擎,以及產生合成語音之語音合 成引擎◦文字分析引擎另一個可選擇之功能是分配各音標 符號組所相對使用之時間並記錄下來,或者亦可包括設定 各音標符號組之發聲強度資料。 其中文字分析引擎不限定用於文字分析引擎,由於文字分 才斤引擎已為各國目前已普遍發展之技術,因此本方法很容 易用於其他不同語言之國家。 另外若文字分析引擎所得到之音標符號組資料不足以精確 i也代表唇形之變化時,可選擇性地再執行轉換成聲韻音標 符號之步驟,以更精確地代表唇形之變化。 由於本發明之方法確有增進相當之功效,故依法申請發明 專利◦ 圖式簡單説明: 第1圖係本發明方法之流程圖。 第2圖係本發明有關音標符號轉成聲韻音標符號之對照表參 考實施例〇 本紙張瓦度適用中國國家標準(CNS)A4規格(210 x 297公釐) -----------裝 -------訂.-------- (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 550476 A7 B7 五、發明說明(彳) 第3圖係本發明有關聲韻-唇形轉換表之參考實施例。 第4圖係本發明有關處理發聲時間及發聲強度之示意圖。 較佳具體實施例之詳細説明: 請參見第1圖有關本發明方法之流程圖◦本發明最佳之實 彡包環境是在電腦或具有如電腦架構之其他裝置中處理,因 4匕本發明方法之流程圖以電腦軟體流程之架構敘述。 步驟S1 :先由使用者輸入文字資料,該文字資料可為已編 寫好之文字檔,或者由使用者當場輸入之文字資料;甚至 在如中文輸入中,以如中文注音符號輸入之文字資料◦本 發明所適用之文字資料可為一個或複數個字,譬如文字資 料為『梁』一個字,或是在絕大部分之狀況下譬如『您好 η鬲』之具有語言敘述之文字資料。 步驟S2,S3-1,S3-2 :經接收文字資料後,藉由文字分 析引擎轉成音標符號(S3-1 )及合成語音(S3-2 ) ◦文 字分析引擎主要包括產生音標符號之語意語境分析引擎, 以及產生合成語音之語音合成引擎◦語意語境分析引擎將 文字資料轉成複數之音標符號組,其中各音標符號組相對 4戈表-個字。譬如輸入之文字資料為中文『您好嗎』,則 利用語意語境分析引擎辨識後得到『^一―/』,『厂幺 v』及『_Π 丫·』三個音標符號組(中文注音符號),而 各音標符號組包括單一或複數之音標符號,如『好』相對 之音標符號組為『厂幺ν』,而該音標符號組之音標符號 ____6_ 本紙張&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) 經濟部智慧財產局員工消費合作社印製 550476 A7 B7 五、發明說明(3 ) 由『厂』,『幺』及『v』所組成。當然在輸入文字資料 曰寺,本來就是以晋標符號輸入時(譬如以中文注晋符 號),則若輸入文字工具能保留原音標符號,則文字分析 引擎不需要語意語境分析引擎◦語音合成引擎可由文字檔 中之文字碼或者音標符號產生合成語音◦另外需注意的是,文 字分析引擎當然亦可為針對英文、日文等其他外籍語言所設計的, 由於語意語境分析引擎,以及語音合成引擎已為目前已發展之技 射ί,因此本説明書不再贅述該等引擎之技術原理及步驟。 文字分析引擎另一個可選擇之功能是分配各音標符號組所 相I對使用之時間並記錄下來,或者亦可包括設定各音標符 號組之發聲強度資料,有關此部分請一併參考步驟S8 ◦ 另外為了更逼眞模擬產生合成語音之效果,文字分析引擎 更可包括斷詞引擎,斷詞引擎可針對一句話分析決定該句 話中的每一個字之停頓時間;譬如文字資料為『智慧財產 局於民國八十八年成立』,經斷詞後變為『智慧財產局-於 —民國八十八年-成JL』,其中『-』代表停頓時間較長Q由 於斷詞引擎之本身技術並非本發明之探討重點,因此本説明 書不再贅述此技術。 步驟S4 :在步驟S2、S3-1得到各音標符號組之後,以目 前台灣所採用之中文注音符號而言,可再進一步將其轉成 聲韻音標符號組,主要原因是中文注音符號無法精細地代 表唇形之變化,故經過轉換成聲韻音標符號,更能精確地 代表唇形之變化。同樣的每一聲韻音標符號組包括單一或 ____7_ 本紙張尺度適用中國國家標準(CNS)A4規格(210x 297公釐) --------訂·-------- (請先閱讀背面之注意事項再填寫本頁) 立 發明說明(6 ) =之聲韻音標符號。請配合參見第 :表,譬如『梁』之相對之音標符號組為『=—唇形轉 _之影響),音標符號組為『力 以對〜唇〜明 標符號則為『UaN』。’’ /』而對照 < 聲韻音 參見第3圖,依據音標,形轉換表將夂立 二形指令之實: 『為讓輸入之文字資料與具有唇形之圖形能以 冋步頭不屋生影音動畫』,需經過—『同步機制、素 成’此為本發明後續之應用,由於 』= 整合『聲音』及『動畫』之技術,且由;重已= =過又字貧枓直接驅動具有唇形之圖形,故不在此贊 步驟S8 :在步驟S2之過程中,文孛八姑 、、 τ 又子刀析引擎將文字資料朝 成衩數 < 首標符號組及合成言丑音的 义上 风口曰的冋時,為配合步騾S6、 S 7惑衫貫動畫』同步機制,可在牛 i Γ在步驟S2對各音標符號每 彳旨疋將要發出語首炙時間以;3你、A后…〜 及作為唇形指令執行之時間表 數。如果文字分析引擎更包括斷叫 " 手又匕栝引擎時,對於模擬時間 更可達到逼眞之效果。另外亦可# . , J』包括紀錄各晉標符號組之 發聲強度資料,請配合參見第4 R 辟』『 兄罘4圖,譬如『梁』對照之聲旬 55〇476 五 B7 号务明說明(τ ) ==號=『LiANj總共佔则.5秒,各聲韻音標符號可 數^^式依據『母音及子音及前後音等特性予以分配秒 ° 1』、『A』及『N』分別佔用0 . 1秒、 、·” G.2秒及G.3秒,而發聲強度資料則以曲線表示 ^ ’其中時間資料比較重要,以便在步驟S5中每-唇形指 =括時間 < 參數,使得在後續之步驟37中更精確控制且 =形圖形之動作。同樣的若有發聲強度資料,則在步驟 5中母-唇形指令包括發聲強度之參數, 則嘴形越張開,譬如若將發聲強度分為十等級,配合= 九種基本唇形,則可衍伸成九十種唇形。 訂 需〉王意的是,上述僅為實施例,而非限制於實施例。譬如 在步驟^中/ 文字分析引擎所得到之音標符號組即 經 濟 部 智 慧 財 產 局 員 工 消 費 合 作 社 印 製 :使用而直接跳到步驟S5,譬如語意語境分析引擎辨識後 得到『羅馬拼音』則不需要再有步騾S4之動作,或如英文 語意語境分析引擎辨識完所得到之英文音標亦不需要再有 步驟S4,亦即步驟S4之意義在於當步驟S2所得到之資料 不足以精確地代表唇形之變化時,可選擇性地再執行步驟 S 4。又譬如步驟s 8不一定要在步騾s 2之過程中產生,而 只要在步驟S 6足前或同時產生即可,此不脱離本發明基本 架構者,皆應為本專利所主張之權利範圍,而應以專利申 請範圍為準。550476 A7 2 Printed clothing by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 5. Description of the invention (Field and background of the invention; The invention is used in the field of multimedia to produce animation, especially to drive graphics with a lip shape. Known in multimedia In production, such as computer animation or cartoons made with computers, because virtual characters and even animals have a lot to speak & occasions, how to control the movement of the mouth and the muscle changes near the mouth is an Very time-consuming and labor-intensive work. At present, the feasible method is to stick a plurality of inductive detectors to oint the muscles near the mouth of the person, use the inductive detectors to sense the changes in the mouth muscles, and then Drive this data with lip-shaped graphics. Of course, this is a fairly good method, but it has the following disadvantages: 1 · This equipment is very expensive, and its production process is complex, requiring professional assistance, generally only applicable to Large production company or game software manufacturing company. Not suitable for computer animation, especially game software, because After the graphics with lip shape have been made, their movements (including muscle changes in the mouth) can be fixed. However, in game software, the movements of the graphics will be changed according to user input. In the game software, if a conventional method is used, at most, only a pre-stored database can be used as a simulation, but if an unexpected text is spoken, the simulation cannot be performed; for example, the user inputs a word data or the user enters a text The text data is generated by the sound engine, and the conventional method cannot be applied at all. This paper size applies the Chinese National Standard (CNS) A4 specification4〇x 297 公 爱---^ -------- ^ ---- ----- (Please read the precautions on the back before filling this page) 550476 A7 Policy Note (2) 3 · Not applicable to the conversion of different nights, such as a computer animation originally produced in Chinese Change the sound of the text inside the thunder / shame animation into English sound. Bei 1J is familiar with the size, soil, and resolution of the muscles near the mouth (please read the precautions on the back before filling this page). The new production, cream ρχ /, 1 makes it impossible to frighten. The simple result of removing edge j: ·-The main purpose of the present invention is also te ^ ^ Provides an easy way to directly use text data to drive graphics with lips. Another purpose of the present invention is to make use of existing Shaofen technology, and can still easily use the method of the present invention in different regions. This month, the purpose is more applicable to Production of various computer animations, including computer animation movies, cartoon films, game software, and even interactive game bodies, etc. Printed by the Consumer Cooperative of the Intellectual Property Bureau of the Ministry of Economic Affairs-The purpose of this invention is to apply to those who need to convert different languages In this case, since the present invention directly uses the text data to drive the graphic with a lip shape, it is only necessary to change the text data and the internal supporting software program. In order to accomplish the main purpose of the present invention, in the method material of the present invention, in a better application, the content of the text resource is narrative text data; then the text data is converted into a plurality of private payments. No. group, in which each group of Jinbiao symbol represents a single word, and each façade group includes a single or plural phonetic symbol; after obtaining the phonetic symbol ^ This paper size applies the Chinese National Standard (CNS) A4 specification (210 X 297) ) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 550476 A7 B7 V. Numbering instructions (3) After 1, convert each group of award symbols into lip instructions according to the bone label-lip shape conversion table, and finally according to the lip shape Commands control actions with lip shapes. Among them, in a better implementation, the Chinese and Chinese character analysis engine is responsible for converting text data into phonetic symbols and synthesized speech. The text analysis engine mainly includes a semantic context analysis engine that generates phonetic symbols, and a speech synthesis engine that generates synthesized speech. Another optional function of the text analysis engine is to allocate and record the relative use time of each phonetic symbol group, or it can also include setting the sound intensity data of each phonetic symbol group. The text analysis engine is not limited to the text analysis engine. Since the text analysis engine is a technology that has been generally developed in various countries, this method is easily applicable to other countries with different languages. In addition, if the phonetic symbol set data obtained by the text analysis engine is not accurate enough i also represents the change of the lip shape, the step of converting into the phonetic phonetic symbol can be optionally performed to represent the change of the lip shape more accurately. Since the method of the present invention does have an equivalent effect, the invention patent is applied in accordance with the law. Schematic description: Figure 1 is a flowchart of the method of the present invention. Figure 2 is a reference example of the comparison table of phonetic symbols into phonological phonetic symbols according to the present invention. The paper wattage is in accordance with Chinese National Standard (CNS) A4 (210 x 297 mm) --------- --Equipment ------- Order .-------- (Please read the notes on the back before filling this page) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 550476 A7 B7 V. Description of the invention (Ii) FIG. 3 is a reference embodiment of the rhyme-lip shape conversion table of the present invention. FIG. 4 is a schematic diagram of processing the utterance time and utterance intensity according to the present invention. Detailed description of the preferred embodiment: Please refer to FIG. 1 for a flowchart of the method of the present invention. The best practical package environment of the present invention is to be processed in a computer or other device with a computer architecture. The flow chart of the method is described by the framework of computer software flow. Step S1: The user first enters text data, and the text data can be a written text file or text data input by the user on the spot; even in the Chinese input, for example, the text data input in Chinese phonetic symbols. The text data applicable to the present invention may be one or a plurality of words, for example, the text data is a word "liang", or in most cases, such as "hello hello" text data with language description. Steps S2, S3-1, S3-2: After receiving the text data, it is converted into phonetic symbols (S3-1) and synthesized speech (S3-2) by the text analysis engine. The text analysis engine mainly includes the semantic meaning of the phonetic symbols. Context analysis engine, and speech synthesis engine that generates synthetic speech. Contextual context analysis engine converts text data into plural phonetic symbol groups, where each phonetic symbol group is 4 characters per word. For example, if the input text data is Chinese "How are you?", You can use the semantic context analysis engine to identify three groups of phonetic symbols: "^ 一 ― /", "Factory 幺 v" and "_Π 丫 ·" (Chinese phonetic symbols) ), And each phonetic symbol group includes a single or plural phonetic symbol, such as "Good", the relative phonetic symbol group is "factory 幺 ν", and the phonetic symbol of the phonetic symbol group ____6_ This paper & degree applies Chinese national standards (CNS) A4 size (210 x 297 mm) ----------- ^ -------- ^ --------- (Please read the precautions on the back first Refill this page) Printed by the Consumer Cooperatives of the Intellectual Property Bureau of the Ministry of Economic Affairs 550476 A7 B7 V. Invention Description (3) It consists of "factory", "幺" and "v". Of course, when inputting text data, the temple was originally entered with Jin symbols (such as the Jin symbol in Chinese). If the text input tool can retain the original phonetic symbols, the text analysis engine does not need a semantic context analysis engine. ◦ Speech synthesis The engine can generate synthesized speech from the text code or phonetic symbols in the text file. Note also that the text analysis engine can also be designed for English, Japanese and other foreign languages. The semantic context analysis engine and speech synthesis The engines have been developed so far, so this specification will not repeat the technical principles and steps of these engines. Another optional function of the text analysis engine is to allocate and record the relative use time of each phonetic symbol group, or it can also include setting the sound intensity data of each phonetic symbol group. For this part, please refer to step S8 together. In addition, in order to more effectively simulate the effect of synthesizing speech, the text analysis engine can further include a word segmentation engine. The word segmentation engine can determine the pause time of each word in the sentence for a sentence analysis; for example, the text data is "intelligent property "The Bureau was established in the eighty-eight years of the Republic of China", after the word segmentation, it became "Intellectual Property Bureau-Yu-eighty-eight years of the Republic of China-JL", where "-" stands for a longer pause Q due to the technology of the word segmentation engine itself It is not the focus of the present invention, so this description will not repeat this technology. Step S4: After obtaining each phonetic symbol group in steps S2 and S3-1, the Chinese phonetic symbols currently used in Taiwan can be further converted into a phonetic phonetic symbol group, mainly because the Chinese phonetic symbols cannot be refined. Represents the change of the lip shape, so it can more accurately represent the change of the lip shape after being converted into a phonetic symbol. The same set of phonetic symbols for each rhyme includes a single or ____7_ This paper size applies the Chinese National Standard (CNS) A4 specification (210x 297 mm) -------- Order · -------- ( (Please read the notes on the back before filling out this page) Li Invention Note (6) = phonological phonetic symbol. Please refer to the following table for cooperation. For example, the relative phonetic symbol set of "Beam" is "= —the influence of lip-shaped turn _", and the phonetic symbol set is "Force to Lip ~ Marked symbol is" UaN ". ”/” And the contrast < vowel sound see Figure 3, according to the phonetic symbol, the shape conversion table will stand up to the shape of the two-shape instruction: "In order to make the entered text data and lip-shaped graphics "Animation of live video and audio" needs to go through-"Synchronization mechanism, Su Cheng" This is the subsequent application of the present invention, because "= integrates the technology of" sound "and" animation ", and the reason is; heavy has = = over and the word poor directly The driver has a lip-shaped graphic, so it is not recommended here. Step S8: In the process of step S2, the text 孛 姑, τ, and 子 刀 analysis engine will direct the text data into a number < header symbol group and synthesize ugliness The meaning of the sound is the time of the mouth, in order to cooperate with the synchronization mechanism of steps S6 and S7, you can use the initial time for each of the phonetic symbols at step S2 in Niu Γ; 3 you, after A ... ~ and the number of schedules to execute as a lip instruction. If the text analysis engine even includes a broken call " hand and dagger engine, the simulation time can be even better. In addition, you can also record the sound intensity data of each symbol group in the “#., J”, please refer to the 4th figure of the “Picture of R R” “Brother 4”, for example, “Beam” contrast voice Xuan 55〇476 Wu B7 Ming Ming Explanation (τ) == number = "LiANj accounts for a total of 5 seconds. Each phonological phonetic symbol can be counted ^^ The formula assigns seconds according to the characteristics of vowels, consonants and antonyms ° 1", "A" and "N" It takes 0.1 seconds,…, G.2 seconds and G.3 seconds, respectively, and the vocal intensity data is represented by a curve ^ 'wherein time data is more important, so that in step S5 each-lip finger = bracket time < Parameters, making it more precise to control the action of the shape figure in the subsequent step 37. Similarly, if there is sound intensity data, in step 5, the mother-lip command includes the sound intensity parameter, the mouth shape will open more For example, if the vocal intensity is divided into ten grades, and cooperation = nine basic lip shapes, it can be extended to ninety lip shapes. Ordering> The meaning of the king is that the above is only an embodiment, not limited to the embodiment . For example, in step ^ / the phonetic symbol set obtained by the text analysis engine is the wisdom of the Ministry of Economy Printed by the Production Cooperative Consumer Cooperative: Use it and skip directly to step S5. For example, if the "Roman Pinyin" is obtained after the recognition of the semantic context analysis engine, you do not need to take the step of S4, or if the English context analysis engine recognizes The English phonetic symbol obtained does not need to have step S4, that is, the meaning of step S4 is that when the data obtained in step S2 is not sufficient to accurately represent the change of the lip shape, step S4 can be selectively performed again. Step s 8 does not have to be generated in the process of step s 2, but only needs to be generated before or at the same time as step S 6. Those who do not depart from the basic structure of the present invention should all be within the scope of the rights claimed by this patent. Instead, the scope of the patent application shall prevail.