201009613 九、發明說明： 【發明所屬之技術領域】 -種語音翻譯系統及其方法，特別是指―種透 系統及其方法 言翻譯的普通話與多種方言之間的語音二 【先前技術】201009613 IX. Description of the invention: [Technical field to which the invention pertains] - A speech translation system and method thereof, in particular, a "transparent system" and a method thereof. Speech between a translated Mandarin and a plurality of dialects [Prior Art]
近年來，隨著漢語學習的蓬勃發展，除了通用的普通 話之外，各種地區性方言逐漸受到重視，方言亦成為^解 中國各地風俗、經濟及文化所不可或缺的語言。 然而，不同方言之間雖然有其關聯性，但是其差異性 亦不可小覷，以普通話和粤語為例’若僅懂得普通話而沒 有學過粤語的人，在初:墟齡騎n㈣對方所^ 表達的意思，所以難免會因語言不通而產生不必要的誤 會。因此，便有廠商提出方言翻譯軟體，藉由提供使用者 輸入普通話的文字進行方言翻譯，用以輔助使用者了解相 應的方言涵義及方言語音。 不過’礙於人類語言的複雜性，文字的涵義及語音會 隨著文章的内容而有所變化。因此，透過市面上的方言翻 譯軟體翻譯所得到的方言語音之正確性仍然不盡理想，而 且方言語音的聲調更是單調呆板。故如何解決翻譯後的方 言語音正確性不佳且單調呆板的問題，便成為各家廠商急 欲解決的課題之一。 綜上所述’可知先前技術中長期以來一直存在翻譯後 的方言語音正確性不佳且單調呆板的問題，因此實有必要 201009613 提出改進的技術手段，來解決此一間題。 【發明内容】 有鑒於先前技術存在翻譯後的方言語音正確性不佳 且單調呆板的問題，本發明遂揭露一種普通話與多種方言 ' 之間的語音翻譯系統及其方法。 . /本發明所揭露之普通話與多種方言之間的語音翻譯 系統，包含：語音資料庫、接收模組、詞彙比對模組、詞 参彙篩選模組、語音载入模組及聲調調節模組。其中，語音 貝料庫用以儲存普通話語音及方言語音，·接收模組用以接 收待翻文字及_選擇條件；詞彙比對模組根據翻譯選擇 條件載入相應的詞彙對照表，並將詞彙對照表與待翻文字 進行比對後’生成普通糊彙及減的方言财；詞囊筛 選模組用關斷各普通話触或各方調彙之雛，並透 過翻譯選擇條件選擇相應_選規·，根據騎規則篩 選方言詞彙或普通話詞彙；語音載入模組用以根據筛選規 • :筛選的方言詞彙或普通話詞彙，依序载入相應的方言 ' 語音鱗職語音；聲綱賴_據鱗條件選擇 ‘她_節參數，収調節方言語音或普通話語音的聲調 後輸出。 至於本發明所揭露之普通話與多種方言之間的語音 翻譯方法，其包含下列步驟：預先儲存普通話語音及方言 語音；接收待翻文字及翻譯選擇條件；根據翻譯選擇條件 載入相應的詞囊對照表，並將詞囊對照表與待翻文字進行 比對後，生成普通話詞彙及相應的方言詞彙·’判斷各普通 201009613 活1或各方言詞彙之詞性，並透過翻譯選擇條件選擇相 應的篩選規則後，根據篩選規則篩選方言詞彙或普通話詞 彙，根據轉規晰騎的方言詞彙或普通闕彙，依序 載入相應的方5语音或普通話語音；根據翻譯選擇條件選 擇相應的轉參數’用以·方言語音或普通話語音的聲 調後輸出。 本發明所揭露之系統與方法如上，與先前技術之間的 差異在於本發财過㈣酬選取正柄方言語音並 行聲調調節。 透過上述的技術手段，本伽可崎献高方言語音 的準確性及擬真程度的技術功效。 【實施方式】 以下將配合圖式及實施例來詳細說明本發明之實施 方式’藉此對本發明如何應用技術手段來解決技術問題並 達成技術功⑽實騎雜充練麵_實施。 立在說明本發明所揭露的普通話與多種方言之間的語 :翻澤系統及其方法之前，先對本發明中所自行定義的名 2如下之解釋。本發明中所提到的普通話語音及方言語 曰疋預先錄製好的普通話及方言之語音檔案，所述方言即 為地方性的語言’如：閩南語及上海話。舉例來說，普通 ^吾音為麵話的語音職，而方言語韻鱗語的語音 3案:閩南語的語音或上海話的語音鄕...等。所述 j话語音及方言語音均預先設置於語音資料庫中，方言 《更包含第-方言語音(例如：細話)及第二方言語音 201009613 2如：客家語），以可透過來新增第三語言語音(例 二四川話)於語音資料庫，並以此類推新增各種不同的 方言語音。 接著’請參閱「第！圖」，「第】圖」為本發明普通話 ’、夕種方言之間的語音翻譯系統之方塊圖，包含··語音資 . 料庫1〇卜接收模、组1〇2、詞彙比對模組103、詞彙篩賴 組⑼、語音載入模組105、聲調調節模組1〇6及更新模 e ^ 1〇7。其中’語音資料庫101用以儲存普通話語音及方 言語音，所述的普通話語音及方言語音已於前面自行定義 的名詞中進行說明，故在此不再作贅述。 接收模組102用以接收待翻文字及翻譯選擇條件，所 述的待翻文字可為普通話的文字(例如：「他在學校」）或方 p的文子(例如·閩南語文字為「依佇學校」），而翻譯選 擇條件則包含普通話翻譯成方言及方言翻譯成普通話之 條件設定，在實際實施上，可將普通話翻譯為閩南語以數 • 值1代表、普通話翻譯為粤語以數值2代表及普通話翻譯 為上海話以數值3代表，以及將閩南話翻譯為普通話以數 值4代表、粵語翻譯為普通話以數值5代表及上海話翻譯 為普通話以數值6代表，並以此類推，這樣設置方便通過 鍵盤及滑鼠等多種方式進行輸入；另外，當要將普通話翻 譯成其它方言時，接收模組102亦可藉由語音轉文字引擎 (Speech To Text，STT)直接透過收音裝置接收使用者的語 音’並且轉換為普通話的文字作為待翻文字，由於上述語 音轉文字引擎為習知技術，故在此不多作贅述。 201009613 詞彙比對模組103根據 載入相應的_懷表，州物：數值^ 贿6；]彙對照表與待翻文字進行 比對後，生成普通話詞彙及相 照表為記錄普通話詞彙及方；彙。所述詞彙對 ” 讀的同義對照列表，也就 疋說，备方W彙為閩南語時，其相應In recent years, with the vigorous development of Chinese language learning, in addition to the common Mandarin, various regional dialects have gradually received attention, and dialects have become an indispensable language for the customs, economy and culture of China. However, although there are relatedities between different dialects, the difference can not be underestimated. Take Mandarin and Cantonese as examples. If you only know Mandarin and have not learned Cantonese, in the beginning: the age of the ride n (four) each other ^ The meaning of expression, so it is inevitable that there will be unnecessary misunderstandings due to language barriers. Therefore, some manufacturers have proposed dialect translation software to provide dialect translation by providing the user to input the words of Putonghua to help the user understand the corresponding dialect meaning and dialect voice. However, due to the complexity of human language, the meaning and voice of the text will change with the content of the article. Therefore, the correctness of the dialect voice obtained through the translation of the dialect translation software on the market is still not satisfactory, and the tone of the dialect voice is monotonous. Therefore, how to solve the problem of poor speech correctness and monotonousness in the translated language has become one of the topics that various manufacturers are eager to solve. In summary, it can be seen that in the prior art, there has been a problem that the translated dialect has poor speech correctness and monotony for a long time, so it is necessary to propose an improved technical means to solve this problem. SUMMARY OF THE INVENTION In view of the prior art, there is a problem that the translated dialect voice is not correct and monotonous, and the present invention discloses a speech translation system between a Mandarin and a plurality of dialects and a method thereof. The speech translation system between the Mandarin and the plurality of dialects disclosed in the present invention comprises: a voice database, a receiving module, a vocabulary comparison module, a word sputum screening module, a voice loading module, and a tone adjustment module. group. The voice shell library is used for storing the Mandarin voice and the dialect voice, and the receiving module is configured to receive the text to be flipped and the _ selection condition; the lexical comparison module loads the corresponding vocabulary comparison table according to the translation selection condition, and the vocabulary After comparing the comparison table with the text to be flipped, 'generate the common paste and reduce the dialect; the capsule filter module shuts down the common words or the parties, and selects the corresponding conditions through the translation selection criteria. Regulations, screening dialect vocabulary or Mandarin vocabulary according to the riding rules; the voice loading module is used to load the corresponding dialects according to the screening rules: the dialect vocabulary or the Mandarin vocabulary. Lai _ According to the scale condition, select the 'her _ section parameter, and output the tone of the dialect or Mandarin voice. As for the speech translation method between the Mandarin and the plurality of dialects disclosed in the present invention, the method includes the following steps: pre-storing the Mandarin speech and the dialect speech; receiving the to-be-turned text and the translation selection condition; and loading the corresponding word capsule according to the translation selection condition. Table, and compare the word capsule comparison table with the text to be flipped, generate the Mandarin vocabulary and the corresponding dialect vocabulary. 'Determine the general word of the 201009613 live 1 or the vocabulary of the parties, and select the corresponding screening rules through the translation selection criteria. After that, according to the screening rules, the dialect vocabulary or the Mandarin vocabulary is filtered, and according to the dialect vocabulary or the ordinary vocabulary of the circumstantial ride, the corresponding party 5 voice or Mandarin voice is sequentially loaded; and the corresponding turn parameter is selected according to the translation selection condition. • The tone of the dialect or Mandarin voice is output. The system and method disclosed by the present invention are as above, and the difference between the prior art and the prior art is that the present invention has been used to adjust the tone of the positive dialect speech. Through the above-mentioned technical means, Benjakozaki contributes to the technical accuracy of the accuracy and immersiveness of the dialect speech. [Embodiment] Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings and the embodiments, and thus, the present invention is applied to the technical means to solve the technical problems and to achieve the technical work (10). Before explaining the language between the Mandarin and the various dialects disclosed in the present invention: before the system and the method thereof, the following explanation of the name 2 which is self-defined in the present invention is made. The Putonghua speech and dialect mentioned in the present invention are pre-recorded voice files of Mandarin and dialect, which are local languages such as: Min Nan and Shanghai dialect. For example, the ordinary ^wuyin is the voice of the face-to-face, and the voice of the dialect of the dialect is 3: the voice of the Minnan dialect or the voice of the Shanghai dialect...etc. The j voice and the dialect voice are pre-set in the voice database, and the dialect "includes the first dialect voice (for example: whisper) and the second dialect voice 201009613 2 such as: Hakka language), which can be added by The third language speech (example 2 Sichuan dialect) is in the voice database, and various different dialect voices are added in this way. Then, please refer to the "No.!", "The first picture" is a block diagram of the speech translation system between the Mandarin and the Chinese dialects of the present invention, including the voice resource. The library 1 receiving mode, group 1 2. The vocabulary comparison module 103, the vocabulary screening group (9), the voice loading module 105, the tone adjustment module 1〇6, and the update mode e^1〇7. The voice database 101 is used for storing the Mandarin voice and the dialect voice. The Mandarin voice and the dialect voice have been described in the previously defined nouns, and therefore will not be described herein. The receiving module 102 is configured to receive the text to be flipped and the selection condition of the translation, and the text to be flipped can be a character of a Mandarin (for example, "he is at school") or a text of a square p (for example, the text of the Minnan language is "伫The school's translation criteria include the translation of Mandarin into dialects and the translation of dialects into Mandarin. In practice, Mandarin can be translated into Minnan with a value of 1 and a Mandarin for Cantonese with a value of 2. The translation of Mandarin into Shanghai dialect is represented by a value of 3, and the translation of Minnan dialect into Mandarin is represented by a value of 4, the translation of Cantonese into Mandarin is represented by a value of 5, and the translation of a Shanghai dialect into a Mandarin is represented by a value of 6, and so on. The input is performed by a keyboard, a mouse, or the like. In addition, when the Mandarin is to be translated into other dialects, the receiving module 102 can also receive the user directly through the radio device through a Speech To Text (STT). Voice 'and convert to Mandarin text as text to be flipped, since the above-mentioned voice-to-text engine is a conventional technique, This much dwell on them. 201009613 The vocabulary comparison module 103 loads the corresponding _ pocket watch, the state object: the value ^ bribe 6;] the comparison table and the text to be flipped, the Mandarin vocabulary and the comparison table are generated for recording the Mandarin vocabulary and the party; exchange. The vocabulary pair reads the synonymous list of readings, that is to say, when the preparation party W is in Minnan, the corresponding
錄同義的普通話及間南語之詞囊，如：「讀書」及:讀I 以此例而言，當待翻文字為「他在讀書」、而翻譯選擇條Record the synonymous Mandarin and the slang of the Sino-Southern language, such as: "Reading" and: Reading I. In this case, when the text to be translated is "He is reading", the translation selection
件為數值K即普通話翻譯為閩南語)時，載入相應於此翻 譯選擇條件_彙對職(即普通話與_語之詞彙的同 義對照列表），並將詞彙對照表(例如··表中記錄「他」對 應「伊」；「在」對應「符」及「n列」；「讀書」對應「讀冊」） 與待翻文字進行比對後’生成普通話詞彙(例如：「他」、 「在」及「讀書」）及方言詞彙(例如：「伊」、「作」、「〇列」 及「讀冊」）。 詞彙篩選模組104用以判斷各普通話詞彙或各方言詞 彙之詞性，並透過翻譯選擇條件選擇相應的篩選規則後， 根據篩選規則篩選方言詞彙或普通話詞彙。所述判斷詞性 的技術為翻譯軟體之習知技術，故在此不再作贅述，而篩 選規則為根據普通話詞彙或方言詞彙之構句詞性組合(例 如·正確的句子可由主詞、副詞及動詞組合…），選擇適 當方言詞彙或普通話詞彙之規則。以上述待翻文字為「他 在讀書」的普通話翻譯成方言為例：由於每一普通話詞彙 (例如：「他」）會預設對應一個適當的方言詞彙(例如： 「伊」）’甚至一個普通話詞彙(例如：「在」）可預設對應 201009613 多個適當的方言詞彙(例如：「仵」及 情況下根據構句詞性組合來選縣便在不冋 ;組1〇3所產生的普通話詞彙(例如:「在」) 中，、有對應多個方言詞彙(例如：「件」及「例」）時 了提升所選擇的方言詞囊 、 ⑽所刪㈣她1議模組 磡 如]用啸據筛選規則 彙^正確雛組合之_，選㈣當的方言詞 上例而言，普通賴彙如：「他」、及 性組合分別為主詞、副詞及動詞，因此 選擇普通話詞棄為「在」且詞性為副詞所對應的 一二如·「口列」’並且自方言詞彙中刪除另一相應的方 吕岡菜如：「件」0 語音載人顯1G5取根據_選規騎_的方言詞 1立-!v伊」、「％」及「讀冊」）或普通話詞彙，依序自 料庫！01中載入相應的方言語音或普通話語音；聲 I郎^組1G6根據翻譯選擇條件選擇相應的調節參數， ^周即方言語音或#通話語音的聲調後輸出。所述調節 =為分職據不同方言(例如：閩南語、粤語及上海話） I調所預設的參數值，其調節參數亦可用來作為轉換男 聲及女聲够驗_彳如：料般語音進行男聲及女 互相轉換）’而方言語音之輸出方式可透過揚聲裝置 '令八)及寫入語音檔案(例如：壓縮過#邮語音谢 …式進行。除此之外，該系統更可包含更新模組用 、、’J<由網路新增第二语言語音(例如：四川話的語音)及相 201009613 應的詞彙對照表，以及更新翻譯選擇條件。 =第2圖」所示意’「第2圖」為本發明普通話與 二語音翻料法之流糊，包含下列步驟·· ❹ =子曰通話語音及方言語音(步驟2〇1);接收待翻文 ^翻=選擇條件(步驟撕）；根據翻譯選擇條件載入相 應的列彙對照表，並將詞彙對照表與待翻文字進行比對 後，生成普通話詞彙及相應的方言詞彙(步驟2〇3);判斷 各曰通么§]彙或各方言詞彙之詞性，並_翻譯選擇條件 選擇相應_選規職，根據篩選規蘭選方言詞彙或普 通話詞彙(步驟2〇4);根_選規朗_選的方言詞棄或 普通話詞彙’依序載入相應的方言語音或普通話語音(步 驟2〇5) ’根據_選擇條件選擇相應_節參數，用以調 節方σ »。曰或普通話語音的聲調後輸出(步驟2%)。其中， »亥方法更包3透過更新模組，用以經由網路新增第三 語言語音及械的詞彙對照表，以及更新翻譯選擇條件的 步驟。透過上述步卿卩可根據普通話詞彙或方言詞彙的詞 性’對相應的方言詞彙或普通話詞彙進行_選並調節聲 凋，解決以往翻譯後的方言語音正確性不佳且單調呆板的 問題。 以下將配合「第3圖」至「第7圖」以實施例方式進 行如下說明，首先，「第3圖」及「第4圖」為應用本發 明進行普通話轉閩南語之示意圖，請先參閱「第3圖」， 其包含··翻譯視窗300、文字輸入框3〇1、語音播放元件 302、語言選擇元件一 311及語言選擇元件二312。其中， 201009613 翻譯視窗300是提供使用者進行翻譯的查詢介面·文字輸 入框3〇1提供使用者鍵入待翻文字(例如：「他在學校」广 :音播放元件302提供使用者於鍵入待翻文字且進行翻譯 後’用以播放相應於所鍵入的待翻文字之方言語音或普通 2音；語言選擇蔣—311是提供使用者選擇所鍵入的 待翻文字·節物··普触或方言），秘言選擇元件 - 312則是提供個者選擇欲崎摘語言麵，例如：When the value is the value K, that is, the translation of Mandarin into Taiwanese, the corresponding selection condition is _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ Record "he" corresponds to "i"; "in" corresponds to "character" and "n column"; "reading" corresponds to "reading book") After comparing with the text to be flipped, 'generate a vocabulary word (for example: "he", "Yes" and "Reading" and dialect vocabulary (eg "Yi", "Work", "Queue" and "Reading"). The vocabulary screening module 104 is configured to determine the part of speech of each Mandarin vocabulary or the words of the parties, and select a corresponding screening rule through the translation selection condition, and then filter the dialect vocabulary or the Mandarin vocabulary according to the screening rule. The technique for determining part of speech is a well-known technique for translating software, and therefore will not be described again here, and the screening rule is a composition of words according to the vocabulary of a Mandarin or a dialect (for example, the correct sentence can be composed of a subject, an adverb and a verb combination. ...), the rules for choosing the appropriate dialect vocabulary or Mandarin vocabulary. For example, the translation of the Mandarin to be translated into a dialect is as follows: since every Mandarin word (for example, "he") is preset to correspond to an appropriate dialect vocabulary (for example: "I")" even one Putonghua vocabulary (for example, "at") can be preset to correspond to 201009613 multiple appropriate dialect vocabulary words (for example: "仵" and in the case of selecting a county based on a combination of sentence formations; the Mandarin produced by group 1〇3 In vocabulary (for example, "in"), when there are multiple dialect vocabulary words (for example, "pieces" and "examples"), the selected dialect vocabulary is promoted, (10) deleted (4) her 1 module is as follows] Use the whistle-blowing data to filter the rules of the correct combination of _, select (4) as the dialect word, for example, the ordinary reliance such as: "he", and the sexual combination are the main words, adverbs and verbs, so choose the common words Abandoned as "in" and the part of speech is an adverb corresponding to the "two" "mouth" and delete another corresponding party from the dialect vocabulary such as: "piece" 0 voice manned display 1G5 according to _ selection The dialect word of the ride _ 1 --v", "%" and "read ") Or Putonghua vocabulary, sequentially from the material library! 01 is loaded with the corresponding dialect speech or Mandarin speech; the sound I Lang ^ group 1G6 selects the corresponding adjustment parameters according to the translation selection conditions, ^ week is the dialect voice or the voice of the # call voice is output. The adjustment = is the default parameter value of the different dialects (for example: Min Nan, Cantonese and Shanghai dialect), and the adjustment parameters can also be used as a conversion of male and female voices. The male voice and the female convert each other) 'The dialect voice output mode can be transmitted through the speaker device 'Ed 8' and written to the voice file (for example: compressed #邮音谢...), in addition, the system can be Including the update module, 'J< new second-language speech (such as: Sichuan dialect) and the 201009613 vocabulary comparison table, and update translation selection conditions. = Figure 2" "Fig. 2" is a flow of the Mandarin and the two-voice reversal method of the present invention, including the following steps: · ❹ = sub-call voice and dialect speech (step 2〇1); receiving to be transcribed ^ flip = selection condition ( Step tearing; loading the corresponding column comparison table according to the translation selection condition, and comparing the vocabulary comparison table with the text to be flipped, generating a Mandarin vocabulary and a corresponding dialect vocabulary (step 2〇3); § §] the words of the vocabulary And _ translation selection conditions select the corresponding _ election rules, according to the screening rules to select dialect vocabulary or Mandarin vocabulary (step 2 〇 4); root _ selection lang _ selected dialect words or Mandarin vocabulary 'sequentially loaded corresponding Dialect speech or Mandarin speech (step 2〇5) 'Select the corresponding _ section parameter according to _ selection condition to adjust the square σ ». or the tone of the Mandarin voice output (step 2%). Among them, » Hai method is more The package 3 is used to update the vocabulary comparison table of the third language voice and equipment via the network, and to update the translation selection conditions. Through the above steps, the words can be based on the vocabulary of the Mandarin vocabulary or the dialect vocabulary. The dialect vocabulary or the Mandarin vocabulary is used to select and adjust the sound, to solve the problem of poorly correct and monotonous voices in the translated dialects. The following will be carried out in the following way with "3rd" to "7th" As explained below, firstly, "3rd picture" and "4th picture" are schematic diagrams for applying the present invention to transfer Mandarin to Minnan. Please refer to "3rd picture" first, which includes · translation window 300, text input a frame 〇1, a voice playback component 302, a language selection component 311, and a language selection component 312. The 201009613 translation window 300 is a query interface for providing translations by the user. The text input box 3〇1 provides the user to type Text (for example: "He is at school" wide: The audio playback component 302 provides the user with the dialect speech or the ordinary 2 tone corresponding to the typed text to be flipped after typing the to-be-turned text and translating; - 311 is to provide the user to select the type of text to be flipped, the festival, the general touch, or the dialect. The secret selection component - 312 is to provide the individual to choose the language, for example:
^過滑鼠點選語言選擇元件二312來選擇將待翻文字翻譯 為閩南語或客家話。 當使用者透過鍵盤於文字輸入框3〇1中鍵入待翻 (例如：「他在學校）後，可如「第 」乃交J如第4圖」所不意，透過滑 乳氣點選語言選擇元件—311，収選擇待翻文字的類 型(例如：普通話），以及點選語言選擇元件二312來選擇 欲翻譯成絲·r_w如·· _語），雜此辦(普通話 翻譯成閩南語)記錄於翻譯選擇條件(例如··以數值i代表 普通話翻譯成閩南語）。 此時，接收模組102接收待翻文字(例如：「他在學校」） 及翻譯選擇條件⑽如：數值1)後，透過詞彙比對模組10」3 載入相應於翻譯選擇條件的詞彙對照表(即預設的普通話 與閩南語之峨龍表），並將此詞料 記錄「「他」對應「伊」;「在」對應「件」及 對，「學校」〉與待翻文字進行比對後，生成普通話詞彙心 ,·「他」、「在」及「學校」)及相應的方言詞彙(例如：「伊」、 「件」、「σ列」及「學校」）。 12 201009613 「然後，詞彙篩選模組104判斷各普通話詞彙(例如： 九他」、在」及「學校」）之詞性(例如·主詞、動詞及名 屑）’並透過翻譯選擇條件選擇相應的筛選規則(例如：普 閩南語之間的構句詞性組合之規則液，根據篩選 一的記錄如:「主詞與名詞之間需為動詞」來薛選方 ^彙，以此例而言，將騎村詞彙中代表 「咧」。 ❹ ❹ 接著，語音載入模組105根據筛選規則所篩選後方令 如：「伊」、「作」及「學校」），依序自語音_ 墟_載入相應的方言語音後’透過聲調調節模組106根 擇條件選擇相應的調節參數(例如：間南語的聲 =形數據），用以調節(例如：進行聲波的處理)方古达立 的聲調後輸出’其輸出方式可直接透補 甚至兩者皆可同時進行。除此之外，= (例1^3更賴組107用以經_路新增第三語言語音 丨、’四川話的語音)於語音資料庫101中，且下 的照表及更新翻譯選擇條件，以提供擴充其它方言 圖ft閱「第5圖」及「第6圖」，「第5圖」及「第6 二發明進行間南語轉普通話之示意圖。其詳細 此心 料通話轉_語的例子大同小異，故在 參閱「=進=者:餘相同處將不再贅述。請先 於文字以謂巾鍵入咖敝糊待翻文^， 201009613 如伊騎冊」，接著於語言選擇元件一扣選擇相應 」待翻文子的類型(例如：閩南語），以及於語言選擇元件 一312選擇欲翻成的類型(例如··普通話）。 “如第6圖」所示意，當鍵入待翻文字(例如·「伊例 »賣冊」）且完成轉選擇條件⑽如··以數值4代表間南語 . ，成普縣)賴定後，便麟通話制南語的 處理流程载入相應的方言語音(即普通話的語音）。特別要 ❹ 說明的是’由於普通話並無特殊腔調，因此調節參數可為 數值0，即代表不需進行聲調調節。 …另外，請參閱「第7圖」，「第7圖」為應用本發明進 订進階翻澤之示意圖。其翻譯的流程與前述普通話轉間南 語之說明雷同，其差異僅在於輸入待翻文字的方式不同， 以及除了可播放語音外’更可顯示相應的方言文字。當使 用者要將普通話翻譯成其它方言時，可透過進階翻譯視窗 伽點選語音接收元件⑽丨以便透過收音裝置如·麥克 • 風’用以接收使用者的語音並進行語音轉文字的處理，其 . ?吾音轉文字的處理為習知技術，故在此不再作贅述。當語 音轉為文字後’可顯示於文字顯示區塊一 4〇2，並且將翻 澤成閩南語的文字顯示於文字顯示區塊二4〇3中以提供 使用者參考之用。 " 綜上所述，可知本發明與先前技術之間的差異在於透 過筛選規則選取正柄方言語音並進行聲調調節的技術 手段，藉由此一技術手段可以準確選擇方言語音並根據不 同方言的聲調進行適當的調節以解決先前技術所存在的 201009613 問題，進而達成提高方言語音的準確性及擬真程度之技術 功效。 雖然本發明所揭露之實施方式如上，惟所述之内容並 非用以直接限定本發明之專利保護範圍。任何本發明所屬 技領域中具有通常知識者，在不脫離本發明所揭露之精 神和範圍的前提下，可以在實施的形式上及細節上作些許^ Over the mouse click on the language selection component two 312 to choose to translate the text to be translated into Minnan or Hakka. When the user types in the text input box 3〇1 through the keyboard (for example, “He is at school”), he can choose “Large” as “No. Component - 311, select the type of text to be flipped (for example: Mandarin), and click on the language selection component 2 312 to select the translation to be a silk · r_w such as _ 语), miscellaneous (mandarin translated into Min Nan) Recorded in the translation selection conditions (for example, the value i represents Mandarin translated into Minnan). At this time, the receiving module 102 receives the to-be-turned text (for example, "he is at school") and the translation selection condition (10), for example, the value 1), and loads the vocabulary corresponding to the translation selection condition through the lexical comparison module 10"3. The comparison table (that is, the default Mandarin and Minnan dialects), and the term ""he corresponds to "I"; "Yes" corresponds to "pieces" and pairs, "school" and pending text After the comparison, the words of the Putonghua vocabulary, "He", "Yes" and "School") and the corresponding dialect vocabulary (for example, "I", "", "σ" and "School"). 12 201009613 "The vocabulary screening module 104 then determines the part of speech (eg, subject, verb, and cheat) of each of the Putonghua vocabulary (eg, Nine He, "at" and "School") and selects the appropriate sieve through translation selection criteria. Selection rules (for example, the rule liquid of the compositional sentence combination between Pu'ernanese, according to the record of the screening one, such as: "The verb must be a verb between the subject and the noun" to Xue Xuanfang ^ sink, for example, will ride the village The vocabulary stands for "咧". ❹ ❹ Next, the voice loading module 105 selects the following orders according to the screening rules: "I", "Zoo" and "School", and loads the corresponding words from the voice_ __ After the dialect speech, the sound adjustment module 106 selects the corresponding adjustment parameter (for example, the sound data of the Sino-Speaking) to adjust (for example, processing the sound wave) after the tone of the Fangdali The output 'its output mode can be directly multiplexed or even both can be performed simultaneously. In addition, = (Example 1^3 is more dependent on group 107 for adding third-language voice 丨, 'Sichuan dialect voice') in voice database 101, and the following photos and updated translation options Conditions for the expansion of other dialect maps to read "5th" and "6th", "5th" and "6th invention" for the translation of Mandarin to Mandarin. The examples of the language are similar, so I will refer to the words "=============================================================================================== A button selects the type of the text to be flipped (for example, Min Nan), and the language selection component 312 selects the type (for example, Mandarin) to be converted. "As shown in Figure 6", when you type the words to be flipped (for example, "Issue»sales") and complete the selection criteria (10) such as ··· with the value 4 representing the Minnan language, Chengpu County) The processing flow of the lyric call system is loaded into the corresponding dialect voice (ie, the voice of Mandarin). In particular, it is stated that 'because there is no special accent in Mandarin, the adjustment parameter can be a value of 0, which means no tone adjustment is required. ... In addition, please refer to "Fig. 7", and "Fig. 7" is a schematic diagram of the application of the present invention for advanced grading. The translation process is similar to that of the aforementioned Mandarin dialect. The difference is only that the way to input the text to be flipped is different, and the corresponding dialect text can be displayed in addition to the playable voice. When the user wants to translate the Mandarin into other dialects, the voice receiving component (10) can be selected through the advanced translation window gamma to receive the user's voice and perform voice-to-text processing through the radio device such as · Mike Wind. The processing of the voice-to-text is a conventional technique, so it will not be repeated here. When the voice is converted to text, it can be displayed in the text display block 4〇2, and the text that will be translated into Minnan is displayed in the text display block 2〇3〇3 for user reference. In summary, it can be seen that the difference between the present invention and the prior art lies in the technical means of selecting the voice of the front handle dialect and adjusting the tone through the screening rule, by which the dialect voice can be accurately selected and according to different dialects. The tone is appropriately adjusted to solve the 201009613 problem existing in the prior art, thereby achieving the technical effect of improving the accuracy and immersiveness of the dialect speech. While the embodiments of the present invention are as described above, the above description is not intended to directly limit the scope of the invention. Any person skilled in the art to which the invention pertains may make some form and detail in the implementation without departing from the spirit and scope of the invention.
之更動。本發明之專利保護範圍，仍須以所附之申請專利 範圍所界定者為準。 【圖式簡單說明】 第1 圖為本發明普通話與多種方言之間的語音翻譯系 統之方塊圖。 第2圖為本發明普通話與多種方言之間的語音 法之流程圖。 第3圖及第4 @為朗本發明進行普通話轉閩南語之Change. The scope of the invention is to be determined by the scope of the appended claims. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a block diagram of a speech translation system between Mandarin and a plurality of dialects of the present invention. Figure 2 is a flow chart of the speech method between the Mandarin and the various dialects of the present invention. Figure 3 and 4 @ for the invention of the present invention
第5圖至第6圖為應 示意圖。 用本發明進行閩南語轉普通話之 譯之示意圖。 第7圖為應用本發明進行進階翻 【主要元件符號說明】 101語音資料庫 102接收模組 103詞彙比對模組 104詞彙筛選模組 105語音載入模組 15 201009613 106聲調調節模組 107更新模組 300翻譯視窗 301文字輸入框 ' 302语音播放元件 311語言選擇元件一 312語言選擇元件二 400進階翻譯視窗 ® 4〇1語音接收元件 402文字顯示區塊一 403文字顯示區塊二 步驟201預先儲存至少—普通話語音及至少一方一 語音 δ 步驟202接收至少一待翻文字及-轉選擇條件 步驟2〇3根據該翻譯選擇條件載入相應的—詞棄對 • 照表，並將該詞彙對照表與該些待翻文字 進行比對後，生成至少一普通話詞彙及相 •應的至少一方言詞彙 步驟204判斷各該普通話詞彙或各該方言詞囊之詞 性，並透過該翻譯選擇條件選擇相應的一 篩選規則後，根據該筛選規則篩選該些方 言詞彙或該些普通話詞彙 步驟205根據該篩選規則所篩選的該些方言詞彙或 該些普通話詞彙’依序載入相應的該些方 16 201009613 言語音或該些普通話語音 步驟206根據該翻譯選擇條件選擇相應的一調節參 數，用以調節該些方言語音或該些普通話 語音的聲調後輸出 φ 參 17Figures 5 to 6 are schematic diagrams. The schematic diagram of translating the Minnan language into Mandarin is carried out by the present invention. Figure 7 is a schematic diagram of the application of the present invention. [Main component symbol description] 101 voice database 102 receiving module 103 vocabulary comparison module 104 vocabulary screening module 105 voice loading module 15 201009613 106 tone adjustment module 107 update module 300 translation window 301 text input box '302 voice playback element 311 language selection component 312 language selection component two 400 advanced translation window о 4 〇 1 voice receiving component 402 text display block 403 text display block two Step 201 pre-stores at least - Putonghua voice and at least one voice δ step 202 receives at least one to-be-turned text and-to-selection condition step 2〇3 loads the corresponding-word-discarded pair according to the translation selection condition, and After the vocabulary comparison table is compared with the to-be-turned words, at least one Mandarin vocabulary and at least one vocabulary vocabulary step 204 is generated to determine the vocabulary of each of the Mandarin vocabulary or each of the dialect vocabulary, and the translation is selected through the translation. After selecting a corresponding screening rule according to the condition, filtering the dialect vocabulary or the common vocabulary words according to the screening rule, step 205 according to the The dialect vocabulary or the vocabulary vocabulary selected by the selection rule is sequentially loaded into the corresponding party 16 201009613 speech or the mandarin speech step 206 according to the translation selection condition to select a corresponding adjustment parameter for adjusting The dialect voices or the tones of the Mandarin voices are output φ 17 17