200930003 W 26236twf.doc/p 九、發明說明: 【發明所屬之技術領域】 本發明是有關於一種語音辨識系統,且特別是有關於 可於待機時降低語音辨識功能消耗功率的手持裝置與其語 音辨識方法。 【先前技術】 隨著行動通訊技術的進步,行動電話已經成為人們生 活中不可或缺的電子產品。因為現今生活步調加速,通 ® 訊方式更加講求時效,於是行動電話取代以往傳統電 話,憑藉其可攜帶性及便利性,被視為人與人之間最 便利、最快速的通訊工具,人與人之間可透過行動電 話做即時通訊而不受時間、地點的限制。 為使行動電話的使用更為便利,行動電話便整 合語音辨識系統,使用者可直接以聲音來操作行動電 話,例如撥號、接聽電話等,讓行動電話可以達到” 免持”的效果。然而,為降低功率消耗,行動電話在 G 待機狀態下,其語音辨識功能通常關閉的,使用者須 要經由按鍵操作才能啟動語音辨識功能。在某些情況 下,例如開車,以按鍵來操作行動電話的方式,對使 用者而言是不夠便利的,甚至是危險的。 此外,若是持續開啟語音辨識功能,則會產生大 量的功率消耗,而且行動電話也無法進入待機狀態, 對於行動電話的電源管理而言是不切實際的,這會嚴 重降低行動電話的待機時間。 5 200930003 iV 26236twf.doc/p 【發明内容】 本發明提供-種手持裝置與其語音辨識方法,其中手 持裝置中配置另-低功率的語音辨識晶片,讓手持裝置在 進入待機狀_健可錢行語音觸,祕增加使用者 的便利性並同時達到降低辨消耗的設計需求。 仰本發明提出-種手持裝置,包括主處理器、語音辨識 ΙΓί及聲音接收器。聲音接收器,用以將聲音轉換為- ❹ e 傳送至主處理器或語音辨識單元,其中語音辨 算單元,可用以辨識一預設音頻信號。當 態時,語音辨識單元比較音頻信號 能 ,若音麟合音齡號時則致 主處理器以進行語音辨識。 一實施例中,上述手持裝置更包括切換單 於聲;類比數位轉換器,其中切換單元耗接 ^聲曰接收讀#音辨識單元之間,並受控於 〜ίΐ 將音頻信號轉換為數位信號。1中, 接收器與前置放大器。 、連接聲曰 語音料财,謂料觸單低功率 在本發明一實施例中,上述聲音接收器為—麥克風。 放ϋ,減於切換單元與類比數位轉換器之間° 200930003 26236twf.doc/p 在本發明一實施例中,上述手 】時丰,另一提出一種語音崎法為 置’手持裝置包括-低功率語音觸 用於手持裝 方法包括下列步驟:首先,等待主4語音辨識 然後’等待預設音頻信號;接下來處 號時,致能主處理器以進行語音辨識:接收到預設音頻信 ❹ ❹ $發^因在手躲置中增設低功率 片’因此當手縣置進人待機 =識晶 此增加使用者的使較利餘;其他功能,藉 為讓本發明之上述特徵;==二率消耗。 ,魏文待 的手持裝置以開啟複雜的語音喚醒待機狀態下 =====之手持裝置的連接關係 隹收單元m、主包 ....... TO 120以及链在她如、 【實施方式】 盖一實施例 圖 圖。本實;, 括聲音接收單元明、’行動電話100包 'Γ:收單元⑽崎:處理器 器戰語音崎元送至主處理 麥克風。 耷^接收早疋η〇則例如為 當用戶超過預定時間未輪出 100進入待機狀離脖,士也 野订動電話 賴絲時’域理器⑽也會以待機狀態^ 7 200930003 26236twf.doc/p 降低行動電話100的功率消耗。此時,行動電話100會致 旎語音辨識單元130以進行語音辨識。語音辨識單元13〇 可辨2至少一預設音頻信號,此一預設音頻信號可由使用 者自行設定,或是選擇行動電話1〇〇内部的預設值。由於 ^實施例中之語音辨識單元13〇使用低功率語音辨識晶 片’因此不會消耗過多的電池電量。 田聲音接收器11Q所接收到的聲音符合預設音頻信號 _例如開機”或”撥號”等聲音,語音辨識單元130會喚 f處理€ 12G以進行複雜的語音觸魏或其他指令。 辨^本實施例於行動電話削中增設—低功率的語音 /此當主處理器12G處於待機狀態時,依然1 易的語音辨識,好可直接以聲音來喚醒待機狀 i择作r動電話觸或進行簡易的聲控功能,不須經由按 鍵刼作即可啟動語音辨識功能。 按 開啟術中,主處理器必須處於工作狀態才能進行 Ο =而:若長時間啟動主處理器會消耗大量的 本實施例在:處】:寺機時間。相較於習知技術, 後依然可以進行m機狀態(或稱為省電模式) ΐ:致能低功率的語音辨識單·來進行 ί:ίΞ=:;=:;=二戈 W 26236twf.doc/p ❹ ❹ 200930003 能的效益。 第二實施例 昭二下圖來手持農置的電路設計細節,請參 照圖2圖2為根據本發明第二實施例之 關係圖。本實施例同樣以行動 d 、 話200包括聲音接收單元為例作為說明,行動電 214、類比數位轉1器== 、前置放大器 _ 〇 _ 主處理器220以及語音辨識單 7G 230。切換單元212轉接於聲立接 开230以;耸接收态1〇、語音辨識單 π 23二以及别置放大盗214之間,並受控於 類比數位賴n 216 _於前狀Α|| 21盗= 220之間,用以將音頻作祙Δς;姑认〇 土處理盗 主處理器220。 號綱換為數位信號,並輸出至 ,主處理器220進入待機模式時,語音 會致^切換單元212會將聲音接收料2iq所== 立〇 頻^號AS傳送至語音辨識單元謂。當語音辨識單元^ 所收到的音頻信號As符合預設音頻信號 .或,,撥號,,等聲音,語音辨識單元13〇會喚醒主 以進行複雜的語音辨識功能或其他指令。 盗120 押此時’切換單元212會將音頻信號AS傳送 大器214 ’經由前置放大器214放大後,再經由類^ 轉換器^轉換為數位信號。主處理器2㈣根據 位轉換《21G所輸出的數位信號進行語音辨識與 的指令。本實施财之語音賴單元⑽與上述圖 皆為低功率之語音觸“,可在行域話2⑽處於相待同機 9 W 26236twf.doc/p 200930003 模式下進行語音辨識,藉此同時達到省電與方便的功效。 此外,在本發明另一實施例中,切換單元、前置放大 器214以及類比數位轉換器216亦可依照設計需求整合於 主處理器220之中,但並非用以限制本本發明之範圍。 茗三實施例 另-方面’本發明提出一種語音辨識方法,適用於具 有另-低功率語音辨識晶片與一主處理器之手持裝置。請 ❹ 3 ’圖3綠縣發明第三實麵魏之語音辨識 程圖。本實施例以行動電話為例作為說明,本語音 =方法包括下列步驟。首先,在步驟咖中,行動電 驟狀態’也就是正常運作的模式下。然後,在步 ’㈣電話是否進人待機模式,通 電,在-預設時間内未接收到用戶指令時會自 幻崎低功率雜。此外,用戶也可以主動 滑蓋手機的;ίζ機;式,例如將貝殼機合起來時或是將 e ° 行動電話進人待機模 易的語音功率語音辨識晶片以開啟簡 二的音頻信號是否符:二= -低功率語“晶戶,實施例所使用之另 因此,低功率語音辨 =二辨識功能。 器’可減少語音辨識功能所需斤;=力電力遠低於主處理 W 26236twf.doc/p ❹ ❹ 200930003 接下來,在步驟S350令,主處 以降低功率消耗。然後,進入步= 號(聲音‘日^號’當接收到行動電話所預設的音頻信 進入步驟S360 ,致能主處理器。然後,谁 =步驟S370,主處理器會啟動的語音辨識功能羑 令來執行相關功能。上述語音辨識方法的 :;他參照上述第-、二實施例之說明,在此不加累 _設立二:意的是’低功率語音辨識晶片所能辨識 定種類及數量,可依設計需求增加。 =識晶片,因此在待駐健可轉行語音 戶不須使用按鍵即可隨時使用聲控功機 到全時段聲控的效果,此,本發明不僅保=式$ 的用更大幅增加用戶在使用聲控功能時的方便性。 限定本ΪΪ發二已以較佳實施例揭露如上,然其並非用以 Ff疋本發月,任何所屬技術領域中具有通常知 和範圍内,當可作些許之更動與潤飾, 為準。 護_當視無之巾請專鄉_界定者 【圖式簡單說明】 圖。圖1為根據本發明第一實施例之手持裝置的連接關係 圖。圖2為根據本發明第二實施例之手持裝置的連接關係 11 N 26236twf.doc/p 200930003 圖3為根據本發明第三實施例所述之語音辨識方法流 程圖。 【主要元件符號說明】 100 :行動電話 110 :聲音接收單元 120 :主處理器 130 :語音辨識單元 200 :行動電話 210 :聲音接收單元 212 :切換單元 214 :前置放大器 216 :類比數位轉換器 220 :主處理器 230 :語音辨識單元 AS :音頻信號 S310〜S370 :步驟 12200930003 W 26236twf.doc/p IX. Description of the Invention: [Technical Field] The present invention relates to a speech recognition system, and more particularly to a handheld device capable of reducing the power consumption of a speech recognition function during standby and its speech recognition method. [Prior Art] With the advancement of mobile communication technology, mobile phones have become an indispensable electronic product in people's lives. Because today's pace of life is accelerating, the way of communication is more time-sensitive, so mobile phones replace traditional telephones. By virtue of their portability and convenience, they are regarded as the most convenient and fastest communication tool between people. People can make instant messages via mobile phones without being restricted by time or place. In order to make the use of mobile phones more convenient, the mobile phone integrates the voice recognition system, and the user can directly operate the mobile phone by voice, such as dialing, answering the phone, etc., so that the mobile phone can achieve the effect of "free". However, in order to reduce the power consumption, the voice recognition function of the mobile phone is normally turned off in the G standby state, and the user needs to perform the voice recognition function through the button operation. In some cases, such as driving, the way to operate a mobile phone with a button is not convenient or even dangerous for the user. In addition, if the voice recognition function is continuously turned on, a large amount of power consumption is generated, and the mobile phone cannot enter the standby state, which is impractical for the power management of the mobile phone, which seriously reduces the standby time of the mobile phone. 5 200930003 iV 26236twf.doc/p [Invention] The present invention provides a handheld device and a voice recognition method thereof, wherein a handheld device is configured with a low-power voice recognition chip, so that the handheld device enters a standby mode. Voice touch, the secret increases the user's convenience and at the same time achieves the design requirements of reducing the consumption. The present invention proposes a handheld device including a main processor, a voice recognition device, and a sound receiver. The sound receiver is configured to convert the sound into - ❹ e to the main processor or the voice recognition unit, wherein the voice recognition unit can be used to recognize a preset audio signal. In the state, the speech recognition unit compares the audio signal energy, and if the sound is in the same age, the main processor is used for speech recognition. In one embodiment, the handheld device further includes a switching single sound; an analog digital converter, wherein the switching unit is connected between the sound receiving and the reading, and controlled by the audio signal to be converted into a digital signal. . 1, the receiver and the preamplifier. In the embodiment of the invention, the sound receiver is a microphone. ϋ ϋ 减 减 切换 切换 切换 2009 2009 2009 2009 2009 2009 2009 2009 2009 2009 2009 2009 2009 2009 2009 2009 2009 ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° ° The power voice touch method for the handheld device includes the following steps: First, waiting for the main 4 voice recognition and then 'waiting for the preset audio signal; when the number is followed, enabling the main processor for voice recognition: receiving the preset audio signal ❹ $发^ Because the low-power film is added in the hand to hide, so when the hand county is placed in the standby mode = the crystal is added to increase the user's advantage; other functions, by the above features of the present invention; == Second rate consumption. Wei Wen's handheld device to open the complex voice wake-up standby state ===== the connection relationship of the handheld device, the receiving unit m, the main package .... TO 120 and the chain in her, [ Embodiments A cover is shown in an embodiment. This is the case; including the sound receiving unit, 'Mobile phone 100 pack' Γ: Receive unit (10) Saki: The processor warfare is sent to the main processing microphone.耷 ^ Receive early 疋 〇 〇 例如 例如 例如 例如 例如 例如 例如 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当 当/p Reduces the power consumption of the mobile phone 100. At this time, the mobile phone 100 causes the voice recognition unit 130 to perform voice recognition. The voice recognition unit 13 can discriminate at least one preset audio signal, which can be set by the user or select a preset value of the mobile phone. Since the speech recognition unit 13 in the embodiment uses the low power speech recognition chip', it does not consume excessive battery power. The sound received by the field sound receiver 11Q conforms to a preset audio signal such as "on" or "dial", and the voice recognition unit 130 will call f to process 12 12G for complex voice touch or other commands. The embodiment is added to the mobile phone cutting-low-power voice/this is still easy voice recognition when the main processor 12G is in the standby state, so that the standby mode can be directly invoked by the sound, and the mobile phone can be touched or operated. The simple voice control function can activate the voice recognition function without pressing the button. In the process of opening, the main processor must be in the working state to be able to perform Ο = and: if the main processor is started for a long time, it will consume a large amount of this embodiment. :处】: Temple time. Compared with the conventional technology, it can still be in the m state (or called the power saving mode). ΐ: Enable low-power speech recognition. · ί:ίΞ=:;= :;=二戈 W 26236twf.doc/p ❹ ❹ 200930003 Benefits of the second embodiment. The second embodiment shows the circuit design details of the handheld farm. Please refer to FIG. 2 and FIG. 2 for the relationship according to the second embodiment of the present invention. Figure. The embodiment also takes the action d, the speech 200 including the sound receiving unit as an example, the mobile device 214, the analog digital converter ==, the preamplifier_〇_ main processor 220, and the voice recognition list 7G 230. The switching unit 212 Transfer to the sound standoff 230; to receive the state 1〇, the voice recognition single π 23 2 and the other between the magnified thief 214, and controlled by the analog number 赖 n 216 _ in the front state | | | 21 thief = Between 220, the audio is used as 祙Δς; the 〇 处理 处理 盗 主 主 主 主 主 。 。 。 。 。 换 换 换 换 换 换 换 换 换 换 换 换 212 212 212 212 212 212 212 212 212 212 212 212 212 212 The sound receiving material 2iq == 〇 〇 frequency ^ number AS is transmitted to the voice recognition unit. When the voice signal As received by the voice recognition unit ^ conforms to the preset audio signal, or, dial, etc., voice, voice The identification unit 13 唤醒 wakes up the main character to perform a complicated speech recognition function or other instructions. The pirate 120 will now switch the audio signal AS to the amplifier 214 to be amplified by the preamplifier 214 and then converted via the class ^ The device ^ is converted to a digital signal. Device 2 (4) according to the bit conversion "21G output digital signal for speech recognition and instructions. The voice of the implementation of the unit (10) and the above picture are low-power voice touch", can be in the field 2 (10) in the same machine 9 W 26236twf.doc/p 200930003 mode for voice recognition, which at the same time achieves power saving and convenience. In addition, in another embodiment of the present invention, the switching unit, preamplifier 214, and analog digital converter 216 may also be integrated into the main processor 220 according to design requirements, but are not intended to limit the scope of the present invention. Third Embodiment Another aspect of the present invention provides a speech recognition method suitable for a handheld device having another low-power speech recognition chip and a main processor. Please ❹ 3 ’ Figure 3 Green County’s invention of the third real face Wei’s voice recognition process map. This embodiment takes a mobile phone as an example. The voice = method includes the following steps. First, in the step coffee, the action state is also in the normal operating mode. Then, in the step ' (4) whether the phone enters the standby mode, the power is turned on, and the user will not receive the user's command within the preset time. In addition, the user can also actively slide the phone; the type, for example, when the shell machine is put together or the e ° mobile phone enters the standby mode of the voice power speech recognition chip to open the simple two audio signal : Two = - Low power language "Crystal, the other one used in the embodiment, therefore, low power speech recognition = two identification function. The device can reduce the need for speech recognition function; = power is much lower than the main processing W 26236twf. Doc/p ❹ ❹ 200930003 Next, in step S350, the main unit is to reduce the power consumption. Then, enter the step = number (sound 'day ^ number' when receiving the audio signal preset by the mobile phone proceeds to step S360, enabling The main processor. Then, who = step S370, the main processor will activate the speech recognition function to execute the relevant function. The above speech recognition method: he refers to the description of the above-mentioned second and second embodiments, and does not add here Tired_Setup 2: It means that the 'low-power speech recognition chip can identify the type and quantity, which can be increased according to the design requirements. = Identify the chip, so you can always use the button without waiting for the key. The effect of the voice control machine to the voice control in the whole time period, the invention not only ensures that the user's use of the voice control function is greatly increased, but the user has already disclosed the above as the preferred embodiment, but It is not intended to be used in the Ff 发 发 , , , , , , , , , , , 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何 任何1 is a connection diagram of a handheld device according to a first embodiment of the present invention. FIG. 2 is a connection relationship of a handheld device according to a second embodiment of the present invention. 11 N 26236 twf.doc/p 200930003 FIG. 3 is based on Flowchart of the speech recognition method according to the third embodiment of the present invention. [Description of main component symbols] 100: mobile phone 110: sound receiving unit 120: main processor 130: voice recognition unit 200: mobile phone 210: sound receiving unit 212 : switching unit 214: preamplifier 216: analog digital converter 220: main processor 230: speech recognition unit AS: audio signal S310~S370: step 12