200916813 九、發明說明： 【發明所屬之技術領域】 本發明係提供-種使祕轉克風_之音财位辨識器， 尤雜-種使用_雑麥克風_所構成的矩形_，用:降 低環境雜訊影響，提高聲音接收品s與聽覺纽辨識率。 【先前技術】 按，移動體（例如機器人）的聽覺功能多利用取音褒置（如 麥克風）接受聲波後，轉換成電塵來進行語音及語意之辨識，這 如同人類的耳雜㈣聲音，轉換成峨繼大腦一般。 聲波主要透過空氣進行傳遞，而在—個密閉的環境下(例如： 展覽館、辦公室等),，空氣中同時有其他的雜訊(例如：灰塵的折 射、空間的反射等)’在游離，這些干擾會嚴重降低語音辨識的效 能。習知麟f提㈣賴絲克絲取倾絲立絲克風， 來解決因麥克風與語者距離較遠所產生雜訊接收關題。然而使 用頭戴式麥克風會造成語者的不便，且長細&戴也會有不舒適的 感覺’因此改用免持式麥克風陣列來取代有線式麥克風的語音辨 識系統，可克服環境噪音和回音對語音訊號的影響，還原出較乾 淨的語音。同時此-技術並非針對特定澡音環境，可適用於任何 澡音環境下’得到令人滿意的效果。因此_有許多研究機構投 入此一領域進行相關的研究。 人的眼睛位於頭部的前方，因此視覺系統是針對前方的物體 去做判斷，搭配上頸部的轉動可以看到一定範圍的物體。然而機 200916813 器人不可能不停的旋轉頸部，這樣在與人類的互動上，是不太禮 貌的表現。然而有部分研究提出使用全方位鏡頭搭配CCD攝影機 作為視覺纽的元件，賴可以克服可視範_問題，但是這樣 的設計跟人類眼睛的器官構造與功能是有所不同的本發明希望 能達到人類與機器人互動，有如兩個人在互動_樣自然。所以必 須配合聽聽統能夠全方位接收訊⑽特性，触到有語者與機 器人在互動溝通，藉由這些訊號進行分析與判斷，得到語者方位 與距離，再驅動機器人轉向到語者方向，面對面的溝通:是符合 人機互動的良好關係。 近幾年麥克風陣列的發展，可以分成立體式和平面式。立體 式的優點在柯以多村慮說財的方位，销_度可以更加 精確，是在計算部份，必須用到三維的快速傅力業轉換，增加 了運算上的複雜性與花費時間。 另外’習知相關之專利前案，諸如胡竹生等人所創作之「個 人電腦之麥克風陣列系統」創作第Ι2βδ477號及許天明所創作之 「結合定位技術之麥克風陣列收音方法及其系統」創作第㈣卿 號。該些專利前案揭示有相關之麥克風陣列相職術，在此併入 本文，以供參考。 【發明内容】 …本伽之目的储供-種使用矩形麥克風_之音訊方位辨 識益係使用線性（平面式)麥克風陣列，用以迅速的讓機器人做 出反應’減少域式麥克風_計算上的繁雜。 200916813 為達致上述目的，本發明之語音方位辨識器，包括：複數個 平面，每一平面設置有複數顆各相距一定距離的麥克風，以形成 方形陣列；一個二階放大電路，用以接收前述麥克風之聲音訊號 後，將訊號進行放大；一帶通濾波器，把前述經二階放大電路放 大後之訊號，而將不必要的頻率雜訊給濾除掉。藉此，有效降低 系統複雜性，用以迅速的讓機器人做出反應，減少立體式麥克風 陣列計算上的繁雜。 為了讓本發明之上述目的、特徵、優點能更明顯，下文特舉 本發明較佳實施例，並配合所附圖示，作詳細說明如下。 【實施方式】 請參閱第-圖，其係根據本發明實施例所綠製之麥克風電路 圖。包括有：-麥克風部份2;—㈣路4，與該麥克風部份2 電性^接；及-電源部份6，用以提供前述配置動作所需之能源。 立明參閱第―®，其係根據本發明實施例崎製之麥克風陣列 丁忍圖包括有複數個平面，每一平面8設置有複數顆（例如四 顆）各她-定距離（例如5cm)的麥克風1〇，以形成方形陣列 (最好疋正方料列）的擺放方式，肋魏不同四個面的 作判定。 本發明實施靖_之料麥姐1G，細電容式麥克風來 現，由於電容式麥克風具有諸多優點（例如 成電能訊號、魏『肛* a h。日直接轉換 原曰重現』的特性及具有極為寬廣的頻率響 …，轉容式麥克朗對於聲音的反應極為靈敏，以致，也同 200916813 時把細微雜訊也接收進來，並且輸出電壓也很小，不容易去觀察 聲音的變化與判斷此音段為有聲段(speech segment)還是無聲段 (silence segment)或雜訊(noise)。因此在電路設計方面，前述 之麥克風10在接收到聲音訊號之後，係經過一個二階放大電路12 (如第三圖所示），用以先將訊號進行放大。由於訊號經放大後， 雜訊也一同放大，因此，本發明實施例以一高通濾波器（Hight Pass Filter ’ HPF)與一低通遽波器（Low Pass Filter，LPF)戶斤構成 的帶通濾波器14(如第四圖所示），以把不必要的頻率雜訊給濾除 掉。本發明實施例經試驗後，認為麥克風前置二級放大迴路增益 (Gain)之範圍應介於1〇~1〇〇〇〇之間為宜，如公式(一)及(二)式所計 算，本發明實施例經過實驗測試與調整後，將增益值設定為1〇〇。200916813 IX. Description of the invention: [Technical field to which the invention pertains] The present invention provides a rectangular _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ The influence of environmental noise, improve the sound receiving product s and the hearing recognition rate. [Prior Art] According to the sound function of a moving body (such as a robot), the sound is received by a sound pickup device (such as a microphone), and then converted into electric dust to perform voice and semantic recognition, which is like a human ear (four) sound. Converted into a sputum following the brain. Sound waves are mainly transmitted through the air, and in a closed environment (for example, exhibition halls, offices, etc.), other noises (such as: dust refraction, space reflection, etc.) are simultaneously in the air, These interferences can seriously degrade the performance of speech recognition. Xi Zhilin f mentions (4) Lai Si Kesi takes the silk to the wind, to solve the problem of noise reception caused by the distance between the microphone and the speaker. However, the use of a head-mounted microphone can cause inconvenience to the speaker, and the length of the & wear will also have an uncomfortable feeling. Therefore, the use of a hands-free microphone array instead of a wired microphone voice recognition system can overcome environmental noise and The effect of echo on the voice signal restores a cleaner voice. At the same time, this technology is not designed for a specific bathing environment and can be applied to any bathing environment to achieve satisfactory results. Therefore, there are many research institutions that have invested in this field to conduct related research. The human eye is located in front of the head, so the visual system is judged against the object in front, and a certain range of objects can be seen with the rotation of the upper neck. However, the machine 200916813 is not able to rotate the neck constantly, so it is not polite to interact with humans. However, some studies have proposed that using a omnidirectional lens with a CCD camera as a component of the visual link can overcome the problem of visual paradigm, but this design is different from the organ structure and function of the human eye. Robot interaction, like two people in interaction _ like nature. Therefore, it is necessary to cooperate with the listening system to receive the information (10) in all directions, and to reach the interactive communication between the speaker and the robot. The signals are analyzed and judged to obtain the orientation and distance of the speaker, and then the robot is turned to the speaker direction. Face-to-face communication: a good relationship that is in line with human-computer interaction. In recent years, the development of microphone arrays can be divided into three-dimensional and flat. The advantage of the three-dimensional one is that the degree of sales can be more precise in the case of Ke Yiduo, and in the calculation part, the three-dimensional fast and heavy-duty conversion must be used, which increases the computational complexity and time. In addition, the 'patent-related patents, such as the "Microphone Array System for Personal Computers" created by Hu Zhusheng and others, create the "Microphone Array Radio Collection Method and System with Combined Positioning Technology" created by Dijon 2βδ477 and Xu Tianming. The fourth (Q) Qing. The prior patents disclose related microphone arrays, which are incorporated herein by reference. [Summary of the Invention] ... The purpose of this gamma is to use a rectangular microphone _ the audio azimuth identification system uses a linear (planar) microphone array to quickly allow the robot to react 'reduced domain microphone _ on the calculation Complex. 200916813 In order to achieve the above object, the speech orientation recognizer of the present invention comprises: a plurality of planes, each plane is provided with a plurality of microphones each separated by a certain distance to form a square array; and a second-order amplification circuit for receiving the aforementioned microphone After the sound signal, the signal is amplified; a band pass filter is used to filter out the signal amplified by the second-order amplifying circuit, and the unnecessary frequency noise is filtered out. In this way, the complexity of the system is effectively reduced, and the robot can be quickly reacted to reduce the complexity of the calculation of the stereo microphone array. The above described objects, features, and advantages of the present invention will become more apparent from the description of the appended claims. [Embodiment] Please refer to Fig., which is a circuit diagram of a microphone made in green according to an embodiment of the present invention. The method includes: - a microphone portion 2; a (four) way 4, electrically connected to the microphone portion 2; and a power supply portion 6 for providing the energy required for the aforementioned configuration action. Referring to the first item, the microphone array according to the embodiment of the present invention includes a plurality of planes, each of which is provided with a plurality of (for example, four) each of a certain distance (for example, 5 cm). The microphone is 1 〇 to form a square array (preferably a square square), and the ribs are determined by four different faces. The invention implements Jing _ material Mai sister 1G, fine-capacitance microphone comes to the present, because the condenser microphone has many advantages (such as the characteristics of the electric energy signal, Wei "anal * ah. Japanese direct conversion original reproduction" and has extremely The wide frequency response... The transmissive McLaren is extremely sensitive to the sound, so that the fine noise is also received in the same way as 200916813, and the output voltage is also small. It is not easy to observe the change of the sound and judge the sound. The segment is a speech segment or a silence segment or noise. Therefore, in terms of circuit design, the aforementioned microphone 10 passes through a second-order amplification circuit 12 after receiving the audio signal (such as the third). The figure is used to first amplify the signal. Since the signal is amplified, the noise is also amplified together. Therefore, the embodiment of the present invention uses a high pass filter (HPF) and a low pass chopper. (Low Pass Filter, LPF) a bandpass filter 14 (shown in Figure 4) to filter out unnecessary frequency noise. After testing in the embodiment of the present invention It is considered that the range of the gain of the microphone preamplifier secondary amplification loop (Gain) should be between 1 〇 and 1 ,, and the embodiment of the present invention is experimentally tested according to the formulas (1) and (2). After the adjustment, set the gain value to 1〇〇.
Gain{min)Gain{min)
Gain(max) R2 R7(mix) 100K 10K 10K χϊοκ ' 〜min) 100^ 1½ x 丁Gain(max) R2 R7(mix) 100K 10K 10K χϊοκ ' ~min) 100^ 11⁄2 x D
10 100000 (二) 在真實的系統上，做出理想的低通濾波器是不易實現的，一 個真實的低通濾波器，其頻率響應在截止頻率處通常會有衰減。 另外，帶通濾波器的設計上也相當複雜，因此，本發明實施例之 200916813 ;慮波器為_低通;慮波器與南通滤波器結合而成，取樣頻率會比 信號最大的頻寬的兩倍再多一點，一般是再多出約1〇%，本發明 實施例所設定的取樣頻率為16kHz，如公式(三)及(四)計算。 高通濾波器之截止頻率為 hp 2nR3Cx 2s-x80xl〇〇//F= 2〇Hz (二) 低通濾波器之截止頻率為 fl广 (四) 另外，為了整合音訊接㈣制電源部份，本發明實施例以 -電源整合電路16，來分配給四㈣路雜（即配置前述之方形 麥克風陣列）所需的電力（如第五圖所示），這樣的設計可以減少 線材的耗費财雜’可叫低過多配、_起的電磁波干擾與線 阻的損失，也讓本纽在整理收納的時候比較方便。 常用來接收音賴元件為麥克風，市面上搭配麥克風來搁取 音訊的農置’通常都是使用音效卡，但是傳統的音效卡只有一個 輸入頻道，_高_音效卡也不過賴四個舰的接收，而且 價格非常的昂貴。本系統之麥克風陣列是由十六個電容式麥克風 所組成’為了分別接收這十六個類比訊號進入_做運算，資料 200916813 的P刀在實施上係選用由Advantech研華公司所生產的 11負料榻取|置。由於考量到要達成即時處理並且運算分 析出使用者的方向，因此選用擷取速度更快可達纟】_ kS/s的 =SB2.0界φ裝置。除了提供十六個類比輸人頻道，另外此系統也 提供了一侧比輸出親，可用來軸馬達哺速控制機器人 朝向使用者所在的方向。另外各有八個數位輸入與輸出的頻道， 方便於日後在此系統上更多功能的與開發。 本發明實施細提出的使舰形麥克風陣狀音訊方位辨識 器，其結合了線性麥克風_、二階訊號放大難帶通渡波器， 在肩配放大贿濾、波$之間的最佳化之後，讓資料擷取系統得到 的訊號可以代表最多語者方向的條件。因為使用環境的因素本 發明選㈣積小同時也絲顧各方位聲音訊號的線性麥克風陣 列，捨棄體積大成本高的環狀麥克風陣列系統，與需要經過複雜 運算來達到三維空間聲音定位的3D麥克風陣列系統。在資料掏取 系統部分，免去傳統音效卡只能支援少數通道的訊號接收，使用 多通道轉換迅速的資料擷取器，不僅可以配合本發明十六通道的 訊號接收’也能提高系統的反應速度。 本發明所使用之線性(平面式)麥克風陣列，可以迅速的讓機器 人做出反應，減少立體式麥克風陣列計算上的繁雜。亦設計麥克 風陣列應使用的最佳麥克風數，以及麥克風間的擺放適當距離的 最佳方式係在單一平面由四顆電容式麥克風各相距5cm的麥克風 陣列，以四面也就是正方形擺放方式，來接收四個面的聲音作判 200916813 定為最佳方式。麥克風陣列接收訊號後經過一個二階放大電路， 先將訊號進行放大之後，再、㈣-辦通舰肋^必要的頻率 雜訊給遽除掉。本創作在實耻更以―低通濾波器搭配一高通遽 波器，截止頻率在20Ηζ〜16ΚΗζ之間，有效降録統複雜性。 使用麥克風陣列於聽覺系統是近年來主要的研發議題，雖然 環型麥克風_可以將聲音減晝分最詳細，每顆麥克風所得到' 的訊號資訊也㈣更確實，但是環性麥克風陣列必須考量圓形系 統與聲波(假設為平行波)的切面問題，使得王袁型系統相當複雜，陣 列也相當龐大。雖立_麥姐_突破了聲音源高度的問題， 使得聲音接收為三維空間更加全面’但是三維富立雜換的運算 相當複雜’若要使用關狀或監控纟鱗即時性工作，將降低 便利性與反應速度。 本發明之線性麥克風_，使用四職容式麥克風增加接收 訊號資訊的範圍，在組合成矩形_四方接收訊號更全面。工作 電路部份整合-麥克風接收電路、二階放大電路與帶通遽波器， 每個頻道（Channel)尺寸已縮小（約只有⑽啦），突破了過去 其他系統體積龐大_題。且在選紅作IC部分也多方#試得到 效能最理想敝合，崎紐A電路誠波器可能造成的訊號衰 減。 如前所述，本發明之線性(平面式)麥克贿列有下列之優點： (i)適用於任何環境。 (11)測試者事先不需要經過訓練即可使用 200916813 (iii) 體積小方便運用於機器人等設備上。 (iv) 系統元件成本低。 古W方便搭配聲音定位系統或是語者辨識系統，系統相容性 同。適用於各類聲音，不受限於任意發聲元件。 故本發明之提出，確實已可改善前逑習知之諸多缺失，應符 合專利產業上雖然前述 的描述及圖式已揭示本發明之較佳實細，惟此乃健實施例之 呈現，舉凡各種增添、修改和取代可能使用於本創作較佳實施例， 仍應屬落入本發明之申請專利範圍所界定之範圍内。因此，本文 於此所揭不的實施例所有觀點，應被視為用以說明本發明，而非 用以限制本發明。本發明之範圍應由後附之申請專利範圍所界 疋，並涵蓋其合法均等物，並不限於先前之描述。 11 200916813 【圖式簡單說明】 第一圖係本發明實施例之麥克風電路圖。 第二圖為本發明實施例之麥克風陣列示意圖。 第三圖係本發明實施例之麥克風陣列二階放大電路圖。 第四圖係本發明實施例HPF與LPF構成的帶通濾波器。 第五圖係本發明實施例之電源整合電路圖。 【主要元件符號說明】 2 麥克風部份 4 RC電路 6 電源部份 8 平面 10 麥克風 12 二階放大電路 14 帶通濾、波器 16 電源整合電路 1210 100000 (b) On a real system, it is not easy to make an ideal low-pass filter. A true low-pass filter whose frequency response is usually attenuated at the cutoff frequency. In addition, the design of the band pass filter is also quite complicated. Therefore, in the embodiment of the present invention, the 200916813; the filter is _low-pass; the filter is combined with the Nantong filter, and the sampling frequency is greater than the maximum bandwidth of the signal. A little more than twice, generally about 1% more, the sampling frequency set in the embodiment of the present invention is 16 kHz, as calculated by equations (3) and (4). The cutoff frequency of the high-pass filter is hp 2nR3Cx 2s-x80xl〇〇//F= 2〇Hz (2) The cutoff frequency of the low-pass filter is wide (4) In addition, in order to integrate the power supply part of the audio (4) system, In the embodiment of the invention, the power supply integration circuit 16 is used to distribute the power required for the four (four) way (i.e., the square microphone array described above) (as shown in the fifth figure), such a design can reduce the cost of the wire. It can be called the low electromagnetic interference and the loss of the line resistance. It also makes it convenient to arrange and store. It is often used to receive the sound-receiving component as a microphone. The farmer's device that uses the microphone to hold the audio is usually using a sound card, but the traditional sound card has only one input channel, and the _ high-sound card is not only the four ships. Received, and the price is very expensive. The microphone array of this system is composed of sixteen condenser microphones. In order to receive these sixteen analog signals respectively, the operation of the P-knife of 200916813 is implemented by using Advantech Advantech's 11 negative materials. Tat take | set. Since it is considered to achieve immediate processing and the operation analyzes the user's direction, the selection speed is faster than = k_/s = SB2.0 boundary φ device. In addition to providing sixteen analog input channels, the system also provides a side-to-output master that can be used to control the direction of the user in the direction of the user. In addition, there are eight digital input and output channels, which are convenient for more functions and development on this system in the future. The utility model provides a ship-shaped microphone array audio azimuth identifier, which combines a linear microphone _, a second-order signal amplifying undone band-passing wave device, and after optimization between the shoulder and the amplification bribe filter and the wave $, The signal obtained by the data capture system can represent the condition of the most speaker direction. Because of the use of environmental factors, the present invention selects (4) a small-sized microphone array that also takes care of various sound signals, discards a large-volume and high-cost ring microphone array system, and a 3D microphone that requires complicated operations to achieve three-dimensional spatial sound localization. Array system. In the data acquisition system, the traditional sound card can only support the signal reception of a few channels, and the data channel extractor with multi-channel conversion can not only meet the 16-channel signal reception of the present invention, but also improve the system response. speed. The linear (planar) microphone array used in the present invention can quickly react to the robot and reduce the complexity of the calculation of the stereo microphone array. The best way to design the optimal number of microphones for the microphone array and the proper distance between the microphones is to arrange a microphone array with 5 cm each of the four condenser microphones in a single plane, in a four-sided, square arrangement. It is best to judge the sound of four faces to judge 200916813. After receiving the signal, the microphone array passes through a second-order amplifying circuit, and then the signal is amplified first, and then (4) - the necessary frequency of the ship ribs is removed. This creation is based on a low-pass filter with a high-pass chopper, and the cut-off frequency is between 20Ηζ16ΚΗζ16, which effectively reduces the complexity of the recording. The use of a microphone array in the auditory system is a major research and development issue in recent years. Although the ring microphone _ can reduce the sound to the most detailed, the signal information obtained by each microphone is also more accurate, but the ring microphone array must consider the circle. The problem of the shape of the system and the acoustic wave (assumed to be a parallel wave) makes the Wang Yuan type system quite complex and the array is quite large. Although the _ Mai Jie _ breaks through the problem of the height of the sound source, making the sound reception more comprehensive in the three-dimensional space 'but the operation of the three-dimensional rich and miscellaneous exchange is quite complicated. 'If you want to use the close-up or monitor the scales, it will reduce the convenience. Sex and reaction rate. The linear microphone _ of the present invention uses a four-capacity microphone to increase the range of the received signal information, and is combined into a rectangular _ square receiving signal is more comprehensive. The working circuit is partially integrated - the microphone receiving circuit, the second-order amplifying circuit and the band-pass chopper, and the size of each channel has been reduced (about (10)), breaking through the bulk of other systems in the past. And in the selection of the red IC part is also a multi-party # try to get the best performance, the signal can be reduced by the Shuangxin A circuit. As previously stated, the linear (planar) MacBee of the present invention has the following advantages: (i) Applicable to any environment. (11) The tester can use the training without prior training. 200916813 (iii) The small size is convenient for use on equipment such as robots. (iv) System components are low cost. The ancient W is convenient to match the sound localization system or the speaker recognition system, and the system compatibility is the same. Applicable to all kinds of sounds, not limited to any sounding components. Therefore, the present invention has indeed been able to improve many of the shortcomings of the prior art, and should be consistent with the patent industry. Although the foregoing description and drawings have disclosed the preferred embodiments of the present invention, Additions, modifications, and substitutions may be used in the preferred embodiments of the present invention, and are still within the scope defined by the scope of the invention. Therefore, the opinions of the embodiments disclosed herein are not to be construed as limiting the invention. The scope of the present invention is defined by the scope of the appended claims, and the legal equivalents thereof are not limited to the foregoing description. 11 200916813 [Simplified description of the drawings] The first figure is a microphone circuit diagram of an embodiment of the present invention. The second figure is a schematic diagram of a microphone array according to an embodiment of the present invention. The third figure is a second-order amplifying circuit diagram of the microphone array of the embodiment of the present invention. The fourth figure is a band pass filter composed of HPF and LPF according to an embodiment of the present invention. The fifth figure is a power integration circuit diagram of an embodiment of the present invention. [Main component symbol description] 2 Microphone section 4 RC circuit 6 Power supply section 8 Plane 10 Microphone 12 Second-order amplifier circuit 14 Bandpass filter, wave 16 Power integration circuit 12