201015539 九、發明說明： 【發明所屬之技術領域】 本發明係有關於一種語音辨識功能啟動系統及方法， 且特別有關於-種可以依據聲音之變異度決定是碰動語 音辨識功能之系統及方法。 【先前技術】 &年來，電子裝置’如電腦與可攜式裝置變得越來越 罾高階且變得更多功能化。由於這些裝置與應用的便利，也 使得這些裝置逐漸成為人們的生活必需品之一。 為了提供更便利的輸入與操作方式，部分電子裝置可 以提供語音辨識系統。使用者可以利用語音完成輸入與操 作電子裝置。另外，當使用者處於不適合利用手動輸入與 操作的環境中，如在開車的時候，語音辨識系統亦提供使 用者更便捷與安全的輸入與操作方式。雖然透過語音可以 輸人鋪作電子裝置與域汽衫統。然而，如何啟動語音 辨識系統成為設計者的重要關鍵。 由於環境中充滿各式各樣的聲音，如果讓語音辨識系 統=續地辨識聲音，常常會產生許多錯誤的辨識。因此， 通常會額外設計-個按鈕來啟動語音辨識系統。由於使用 者必須手動按下此按鈕來啟動語音辨識系統，因此，對於 使用者而言係不便的，且在特殊狀況下，如行車中，此行 為係缺乏安全性的。 =為了克服前述問題，一種習知技術開發來啟動語音辨 識系統。在此習知技術中，系統會持續偵測聲音中是否包 HTC097〇5〇-〇™746.A41763^raft_Finai 5 201015539 括 以進力，Γ則啟動語音辨識系統， 須手動按二:=在此習知技術中，使用者無 系_ 係持 r常::rr的環境中，可=:的二音 術=作::品:用關鍵字來啟動語音辨識系統的技 #【發明内容】 有鑑於此，本發明提供語音辨識功能啟動系統及方法。 立。本發明實施例之一種語音辨識功能啟動系統包括一收 音單元與一處理模組。處理模組取得收音單元偵測得到之 一第一期間之一第一聲音，且計算第一期間内第一聲音之 一第一變異度。處理模組判斷第一變異度是否小於一第一 設定值。當第一變異度小於第一設定值時，處理模組取得 收音單元偵測得到之一第二期間之一第二聲音，且判斷第 •二聲音中是否包括一關鍵字。當第二聲音中包括關鍵字 時，處理模組啟動一語音辨識功能。當語音辨識功能啟動 時’收音單元偵測得到之一第三聲音中之每一文字將被偵 測。 本發明實施例之一種語音辨識功能啟動方法。首先’ 取得一第一期間之一第一聲音，j_計算第一期間内第一聲 音之一第一變異度。判斷第一變異度是否小於一第一設定 值。當第一變異度小於第一設定值時，取得一第二期間之 一第二聲音。判斷第二聲音中是否包括一關鍵字。當第二 HTC097050-0-TW/0746-A41763-TW/Draft-Final 6 201015539 聲音中包括關鍵字時，啟動一語音辨識功能。當語音辨識 功能啟動時，一第三聲音中之每一文字將被偵測。 本發明上述方法可以透過程式碼方式存在。當程式碼 被機器載入且執行時，機器變成用以實行本發明之裝置。 為使本發明之上述目的、特徵和優點能更明顯易懂， 下文特舉實施例，並配合所附圖示，詳細說明如下。 【實施方式】 第1圖顯示依據本發明實施例之語音辨識功能啟動系 統。 語音辨識功能啟動系統100可以是一電子裝置，如電 腩糸統、車糸統、與可攜式裝置，如多媒體播放器、個 人數位助理、全球衛星定位裝置、觸控式手機、智慧型手 機或行動電話等之手持式裝置。語音辨識功能啟動系統 100包括一收音單元11〇、一顯示單元12〇與一處理模組 130。收音單元11〇可以是一麥克風用以接收環境中之聲 Φ 音。顯示單元12〇可以是一螢幕或是一燈號，用以顯示關 鍵字偵測圖示。處理模組130係依據收音單元110接收的 聲音執行本案之語音辨識功能啟動方法，其細節將於後說 明。 第2圖顯示依據本發明實施例之語音辨識功能啟動方 法。 如步驟S202，透過收音單元110接收一期間之聲音， 且如步驟S204,計算期間内聲音之變異度(Variance)。值得 注意的是’計算變異度的方法係數值分析熟習之技術，在 HTC097050-0-TW/0746-A41763-TW/Draft-Final 201015539 此不再贅述。如步驟S206，判斷此期間聲音的變異度是否 小於一第一設定值，且維持一既定時間。注意的是，第一 設定值與既定時間可以依據不同需求彈性設計。當此期間 聲音的變異度並未小於第一設定值或持續既定時間時（步 驟S206的否），流程回到步驟S202。當此期間聲音的變異 度小於第一設定值且持續既定時間時(步驟S206的是），如 步驟S208，透過顯示單元120顯示一關鍵字偵測圖示。關 鍵字偵測圖示之顯示可以提示使用者進行關鍵字之輸入。 值得注意的是，步驟S206中判斷變異度是否小於第一設定 值既定時間係用以避免瞬間聲音變化與/或不同聲音源造 成的誤判。然而，在一些實施例中，步驟S206亦可僅判斷 變異度是否小於第一設定值即可。 如步驟S210，透過收音單元110持續接收另一期間之 聲音，且如步驟S212，計算此期間内聲音之變異度。如步 驟S214，判斷此期間聲音的變異度是否大於一第二設定 值。當此期間聲音的變異度並未大於第二設定值時（步驟 S214的否），流程回到步驟S210。當此期間聲音的變異度 大於第二設定值時（步驟S214的是），如步驟S216,判斷聲 音中是否包括一内定之關鍵字。類似地，步驟S212與S214 中計算與判斷此期間聲音的變異度是否大於第二設定值係 用以避免瞬間聲音變化與/或不同聲音源造成的誤判。然 而，在一些實施例中，步驟S212與S214可以省略，而直 接進行步驟S216的判斷。若聲音中並未包括内定之關鍵字 (步驟S216的否），如步驟S218，取消在顯示單元120中相 HTC097050-0-TW/0746-A41763-TW/Draft-Final 8 201015539 應關，子偵測圖示之顯示，並回到步驟％⑽。若聲音中包 $内定之關鍵字（步騍S216的是），如步驟幻2〇,啟動一語 曰辨識功能。注意的是，當語音辨識功能啟動時，接收之 聲曰中母一文字都將會被偵測。 e因此，本案之語音辨識功能啟動系統及方法可以依據 %〇兄中聲音的變異度自動啟動語音辨識魏。當期間内聲 音的變異度小於設定值時，啟動關鍵字侧，且在偵 關鍵字之後自較動語音觸魏，㈣在钱與安全= 的考量下，啟動語音辨識功能。 本發明之方法，或特定型態或其部份，可以以程 的型態存在。程柄可以包含於實體媒體，如軟碟、^ 片硬碟、或疋任何其他機器可讀取(如電腦可讀取々、 媒體’亦或祕於外在形式之電腦㈣產品，其巾，=存 式碼被機器，如電腦載入且執行時，此機器變成用以 本發明之裝h程式碼也可以透過—些傳送媒體，如雷、 或電纜、光纖、或是任何傳輸型態進行傳送，其中♦線 式碼被機器’如電腦接收、載人且執行時，此機器變^ 以參與本發明之裝置。當在—般料處科元實作^用 式碼結合處料元提供—操伽⑽應用 “程 獨特裝置。 您弭1：路之 雖然本發明已以較佳實施例揭露如上然其並 限定本發明，任何熟悉此魏藝者，在殘離树明^ 神和範圍内，當可做些許更動與潤飾，因此本發 精 範圍當視後附之申請專利範圍所界定者為準。 呆護 HTC097050-0-TW/0746-A41763-TW/Draft-Final 201015539 【圖式簡單說明】 第1圖為一示意圖係顯示依據本發明實施例之語音辨 識功能啟動系統。 第2圖為一流程圖係顯示依據本發明實施例之語音辨 識功能啟動方法。 【主要元件符號說明】 100〜語音辨識功能啟動系統； 110〜收音單元； 120〜顯示單元； 130〜處理模組； S202、S204、…、S220〜步驟。201015539 IX. Description of the Invention: [Technical Field] The present invention relates to a voice recognition function activation system and method, and particularly to a system and method for determining a motion recognition function based on the variability of sound . [Prior Art] & Years, electronic devices such as computers and portable devices have become more sophisticated and more functional. Due to the convenience of these devices and applications, these devices have gradually become one of the necessities of life. In order to provide a more convenient input and operation mode, some electronic devices can provide a speech recognition system. The user can use voice to complete the input and operating electronics. In addition, when the user is in an environment that is not suitable for manual input and operation, such as when driving, the voice recognition system also provides a more convenient and safe way for the user to input and operate. Although voice can be used to convert people into electronic devices and domain sweaters. However, how to activate the speech recognition system is an important key to the designer. Since the environment is full of various sounds, if the speech recognition system = continuously recognizes the sound, many misidentifications are often generated. Therefore, an extra button is usually designed to activate the speech recognition system. Since the user must manually press this button to activate the speech recognition system, it is inconvenient for the user, and in special circumstances, such as driving, this behavior is insecure. In order to overcome the aforementioned problems, a conventional technique has been developed to activate a speech recognition system. In this prior art, the system will continuously detect whether the sound is included in the sound of HTC097〇5〇-〇TM746.A41763^raft_Finai 5 201015539, and then start the voice recognition system, you must manually press two: = here In the prior art, in the environment where the user does not have a system _ system r:: rr, the second sound can be =: =:: product: use the keyword to start the voice recognition system technology # [Summary] In view of this, the present invention provides a speech recognition function activation system and method. Standing. A voice recognition function starting system according to an embodiment of the present invention includes a sound receiving unit and a processing module. The processing module obtains a first sound of one of the first periods detected by the sounding unit, and calculates a first variability of the first sound in the first period. The processing module determines whether the first variability is less than a first set value. When the first variability is less than the first set value, the processing module obtains a second sound of one of the second periods, and determines whether a keyword is included in the second sound. When the second sound includes a keyword, the processing module activates a voice recognition function. When the speech recognition function is activated, the radio unit detects that each of the third sounds will be detected. A method for starting a voice recognition function according to an embodiment of the present invention. First, a first sound of one of the first periods is obtained, and j_ calculates a first variability of one of the first sounds in the first period. It is determined whether the first variability is less than a first set value. When the first variability is less than the first set value, a second sound of a second period is obtained. It is determined whether a keyword is included in the second sound. When the second HTC097050-0-TW/0746-A41763-TW/Draft-Final 6 201015539 sound includes a keyword, a voice recognition function is activated. When the speech recognition function is activated, each of the third sounds will be detected. The above method of the present invention can exist in a coded manner. When the code is loaded and executed by the machine, the machine becomes the means for practicing the invention. The above described objects, features and advantages of the present invention will become more apparent from the description of the appended claims. [Embodiment] Fig. 1 shows a voice recognition function starting system according to an embodiment of the present invention. The voice recognition function activation system 100 can be an electronic device, such as a power system, a car system, and a portable device, such as a multimedia player, a personal digital assistant, a global satellite positioning device, a touch mobile phone, a smart phone. Or a handheld device such as a mobile phone. The voice recognition function activation system 100 includes a sound pickup unit 11A, a display unit 12A, and a processing module 130. The radio unit 11A can be a microphone for receiving sound Φ in the environment. The display unit 12A can be a screen or a light to display a keyword detection icon. The processing module 130 performs the voice recognition function activation method of the present invention according to the sound received by the sound pickup unit 110, the details of which will be described later. Fig. 2 shows a method of starting a voice recognition function in accordance with an embodiment of the present invention. In step S202, the sound of a period is received by the sound pickup unit 110, and as in step S204, the variability of the sound during the period is calculated. It is worth noting that the technique for calculating the coefficient value of the method for calculating the variability is familiar with HTC097050-0-TW/0746-A41763-TW/Draft-Final 201015539. In step S206, it is determined whether the variability of the sound during this period is less than a first set value and maintained for a predetermined time. Note that the first set value and the set time can be flexibly designed according to different needs. When the variability of the sound during this period is not less than the first set value or continues for a predetermined time (NO at step S206), the flow returns to step S202. When the variability of the sound during this period is less than the first set value and continues for a predetermined time (YES in step S206), in step S208, a keyword detection icon is displayed through the display unit 120. The display of the keyword detection icon can prompt the user to enter a keyword. It should be noted that it is determined in step S206 whether the variability is less than the first set value for a predetermined time to avoid an instantaneous sound change and/or a misjudgment caused by a different sound source. However, in some embodiments, step S206 may also only determine whether the variability is less than the first set value. In step S210, the sound of another period is continuously received by the sound pickup unit 110, and in step S212, the degree of variability of the sound during the period is calculated. In step S214, it is determined whether the variability of the sound during this period is greater than a second set value. When the variability of the sound during this period is not greater than the second set value (NO in step S214), the flow returns to step S210. When the variability of the sound during this period is greater than the second set value (YES in step S214), in step S216, it is judged whether or not a predetermined keyword is included in the sound. Similarly, steps S212 and S214 calculate and determine whether the variability of the sound during this period is greater than the second set value to avoid an instantaneous sound change and/or a misjudgment caused by a different sound source. However, in some embodiments, steps S212 and S214 may be omitted and the determination of step S216 is performed directly. If the default keyword is not included in the voice (No in step S216), in step S218, canceling the phase HTC097050-0-TW/0746-A41763-TW/Draft-Final 8 201015539 in the display unit 120, the sub-detection Test the display and return to step %(10). If the voice contains the default keyword (step S216 is), if the step is 2, the function is activated. Note that when the voice recognition function is activated, the parent text in the received voice will be detected. e Therefore, the voice recognition function starting system and method of the present invention can automatically start the voice recognition Wei according to the variability of the sound in the % brother. When the variability of the sound during the period is less than the set value, the keyword side is activated, and after the Detect keyword, the voice is touched by the constant voice, and (4) under the consideration of money and safety =, the voice recognition function is activated. The method of the invention, or a particular form or portion thereof, may exist in the form of a process. The handle can be included in physical media, such as a floppy disk, a hard disk, or any other machine readable (such as a computer readable 々, media 'or a secret external computer (4) product, its towel, If the stored code is loaded and executed by a machine, such as a computer, the machine becomes the code for the invention, and can also be transmitted through some medium, such as lightning, cable, fiber, or any transmission type. Transmission, wherein the ♦ line code is received by the machine, such as a computer, when loaded and executed, the machine is changed to participate in the apparatus of the present invention. When the unit is used in the general material, the unit is provided by the combination of the elements. - 伽 伽 (10) application "Cheng unique device. You 弭 1: Road Although the invention has been disclosed in the preferred embodiment as above and defines the invention, any person familiar with this Wei artist, in the remnant of the tree and the scope In the meantime, when a little change and retouching can be done, the scope of this priming is subject to the definition of the patent application scope attached. HTC097050-0-TW/0746-A41763-TW/Draft-Final 201015539 Brief Description] Figure 1 is a schematic diagram showing the basis The speech recognition function activation system of the embodiment is shown in Fig. 2. Fig. 2 is a flow chart showing the method for starting the speech recognition function according to the embodiment of the present invention. [Description of main component symbols] 100~ speech recognition function activation system; 110~ radio unit; 120 to display unit; 130 to processing module; S202, S204, ..., S220~ steps.
HTC097050-0-IW/0746-A41763-TW/Draft-FinalHTC097050-0-IW/0746-A41763-TW/Draft-Final