201212008 六、發明說明： 【發明所屬之技術領域】 [0001] 本發明屬於語音通信領域，尤其涉及一種語音背景 雜訊的消除方法和裝置（一）。 【先前技術】 [0002] 在語音通信領域，由於通話者所處的環境較為複雜 ，導致音頻設備採集的音頻信號除了包括通話者的語音 信號外，還包括周圍環境其他聲源發出的雜訊信號，如 其他人的說話聲、交通工具發出的聲音等，這些雜訊信 I’ 號嚴重的干擾了語音通信的品質，因此在保持高效去噪 的同時，使用較少的器件去除語音信號中混雜的雜訊信 號的技術已成為眾多公司關注的焦點。 【發明内容】 [0003] 本發明的目的在於提供一種語音背景雜訊的消除方 法，旨在解決混雜在語音信號中的雜訊信號嚴重影響語 音通信品質的問題。 〇 [0004] 本發明是這樣實現的，一種語音背景雜訊的消除方 法，所述方法包括下述步驟： [0005] A、對接收的音頻信號進行有效值檢測，得到用於反 映音頻信號的平均功率的輸出信號； [0006] B、將用於反映音頻信號的平均功率的輸出信號與第 一閾值進行比較，產生消噪控制信號，所述第一閾值為 語音信號的平均功率與雜訊信號的平均功率的臨界值； [0007] C、在所述消噪控制信號的控制下，對音頻信號中的 099129997 表單編號A0101 第3頁/共44頁 0992052617-0 201212008 [0008] [0009] [0010] [0011] [0012] 雜訊信號進行消除處理，對語音信號進行放大處理。 本發明的另一目的在於提供一種語音背景雜訊的消 除裝置，所述裝置包括： 有效值檢測單元，用於對接收的音頻信號進行有效 值檢測，得到用於反映音頻信號的平均功率的輸出信號 ， 消噪控制信號產生單元，用於將所述有效值檢測單 元產生的用於反映音頻信號的平均功率的輸出信號與第 一閾值進行比較，產生消噪控制信號，所述第一閾值為 語音信號的平均功率與雜訊信號的平均功率的臨界值； 消噪放大單元，用於在所述消噪控制信號產生單元 產生的消嗓控制信號的控制下，對音頻信號中的雜訊信 號進行消除處理，對語音信號進行放大處理。 在本發明實施例中，通過對接收的音頻信號進行有 效值檢測，得到用於反映音頻信號的平均功率的輸出信 號，通過將該用於反映音頻信號的平均功率的輸出信號 與第一閾值進行比較，生成消噪控制信號，在消噪控制 信號的控制下對音頻信號中的雜訊信號進行消除處理， 對語音信號進行放大處理，從而達到消除語音背景雜訊 的目的。 [0013] 【實施方式】 為了使本發明的目的、技術方案及優點更加清楚明 白，以下結合附圖及實施例，對本發明進行進一步詳細 說明。應當理解，此處所描述的具體實施例僅僅用以解 099129997 表單編號A0101 第4頁/共44頁 0992052617-0 201212008 [0014] Ο [0015] [0016]201212008 VI. Description of the Invention: [Technical Field] The present invention relates to the field of voice communication, and in particular, to a method and apparatus for canceling speech background noise (1). [Previous Technology] [0002] In the field of voice communication, due to the complicated environment of the caller, the audio signal collected by the audio device includes the voice signal of the caller in addition to the voice signal of the caller, and the noise signal from other sources in the surrounding environment. Such as the voice of other people, the sound of the vehicle, etc., these noise signals I's seriously interfere with the quality of voice communication, so while maintaining efficient denoising, less devices are used to remove the mixed signal. The technology of noise signals has become the focus of many companies. SUMMARY OF THE INVENTION [0003] It is an object of the present invention to provide a method for eliminating speech background noise, which aims to solve the problem that the noise signal mixed in the speech signal seriously affects the quality of speech communication. 0004 [0004] The present invention is implemented as a method for eliminating speech background noise, the method comprising the following steps: [0005] A, performing RMS detection on the received audio signal to obtain an audio signal for reflecting Output signal of average power; [0006] B. Comparing an output signal for reflecting an average power of the audio signal with a first threshold to generate a noise cancellation control signal, the first threshold being an average power and noise of the voice signal The critical value of the average power of the signal; [0007] C. Under the control of the noise canceling control signal, 099129997 in the audio signal Form No. A0101 Page 3 / Total 44 Page 0992052617-0 201212008 [0008] [0009] [0012] [0012] The noise signal is subjected to cancellation processing, and the voice signal is amplified. Another object of the present invention is to provide a device for canceling speech background noise, the device comprising: an effective value detecting unit for performing RMS detection on a received audio signal to obtain an output for reflecting an average power of the audio signal a signal, a noise reduction control signal generating unit, configured to compare an output signal generated by the effective value detecting unit for reflecting an average power of the audio signal with a first threshold to generate a noise canceling control signal, where the first threshold is a threshold value of the average power of the voice signal and the average power of the noise signal; a noise cancellation amplifying unit configured to control the noise signal in the audio signal under the control of the cancellation control signal generated by the noise cancellation control signal generating unit The elimination process is performed to amplify the speech signal. In the embodiment of the present invention, an output signal for reflecting the average power of the audio signal is obtained by performing RMS detection on the received audio signal, by performing the output signal for reflecting the average power of the audio signal with the first threshold. Comparing, the denoising control signal is generated, and the noise signal in the audio signal is cancelled under the control of the denoising control signal, and the speech signal is amplified, thereby eliminating the background noise of the speech. [0013] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to solve the problem. 099129997 Form No. A0101 Page 4 / Total 44 Page 0992052617-0 201212008 [0014] [0016]
[0017] G 釋本發明，並不用於限定本發明。 在本發明實施例中，通過對接收的音頻信號進行有 效值檢測，得到用於反映音頻信號的平均功率的輸出信 號，通過將該用於反映音頻信號的平均功率的輸出信號 與第一閾值進行比較，生成消噪控制信號，該第一閾值 是語音信號的平均功率與雜訊信號的平均功率的臨界值 ，在消噪控制信號的控制下對音頻信號中的雜訊信號進 行消除處理，對語音信號進行放大處理，從而實現語音 背景雜訊的消除。 第一圖示出了本發明實施例提供的語音背景雜訊的 消除方法的實現流程，詳述如下： 在步驟S101中，對接收的音頻信號進行有效值（也 稱為均方根值，Root Mean Square，RMS)檢測，得到 用於反映音頻信號的平均功率的輸出信號。 在本發明實施例中，接收的音頻信號可以為採集的 音頻信號，也可以為從其他設備傳輸過來的音頻信號。 由於說話者所處的環境一般較為複雜，存在除說話者以 外的多種發聲源，如周圍其他說話者的聲音、周圍交通 工具的聲音等，因此，接收的音頻信號中不僅包括說話 者發出的語音信號，也包括說話者周圍環境其他聲源發 出的雜訊信號。由於接收的音頻信號中包括語音信號和 雜訊信號，而語音信號產生的電平幅度一般與雜訊信號 產生的電平幅度不同，通過對接收的音頻信號進行有效 值檢測，根據音頻信號的幅度變化，可以產生用於反映 099129997 表單編號A0101 第5頁/共44頁 0992052617-0 201212008 音頻信號的平均功率的輸出信號，即用於反映音頻信號 的平均功率的輸出信號反映了音頻信號的幅度變化。通 過根據語音信號的平均功率與雜訊信號的平均功率之間 的差別，設置語音信號的平均功率與雜訊信號的平均功 率的臨界值，作為第一閾值，即可通過將該用於反映音 頻信號的平均功率的輸出信號與該第一閾值進行比較， 判斷音頻信號為語音信號還是雜訊信號，從而產生消噪 控制信號。 [0018] 由於音頻信號一般是連續的，而對接收的音頻信號 進行有效值檢測可能存在延時，為了避免由於對接收的 音頻信號進行有效值檢測存在的延時而降低語音消噪的 精確度，在本發明另一實施例中，對接收的音頻信號進 行有效值檢測，得到用於反映音頻信號的平均功率的輸 出信號的步驟具體包括： [0019] A、判斷接收的音頻信號是否為語音信號，並根據判 斷結果與前一次產生的消噪控制信號產生有效值檢測的 觸發資訊，當該有效值檢測的觸發信號指示需要對音頻 信號進行有效值檢測時，執行步驟B，否則，執行步驟C 。其具體過程如下： [0020] A1、將接收的音頻信號與第二閾值進行比較，判斷 音頻信號是否為語音信號，得到正向比較結果；其中第 二閾值是正向語音信號的最小值，其具體的設置要依靠 麥克風的靈敏度，及該靈敏度的情況下對不同聲壓雜訊 的採集情況，具體的設定工作要結合聲學理論和實驗統 計。在將接收的音頻信號與第二閾值進行比較時，當音 099129997 表單編號A0101 第6頁/共44頁 0992052617-0 201212008 頻信號大於第二閾值，則判定音頻信號為語音信號，得 到正向比較結果為邏輯真；當音頻信號小於第二閾值時 ，則判定音頻信號為雜訊信號，得到正向比較結果為邏 輯假。 [0021] Α2、將接收的音頻信號與預設的第三閾值進行比較 ，判斷音頻信號是否為語音信號，得到反向比較結果； 其中第三閾值是反向語音信號的最大值，其具體要依靠 麥克風的靈敏度，及該靈敏度的情況下對不同聲壓雜訊 D 的採集情況而定，具體的設定工作要結合聲學理論和實 驗統計。在將接收的音頻信號與預設的第三閾值進行比 較時，如果音頻信號小於第三閾值，則判定音頻信號為 :'獨|霸綱曜_ 語音信號，得到反向比較結果為邏輯真；如果音頻信號 大於第三閾值時，則判定音頻信號為雜訊信號，得到反 向比較結果為邏輯假。 [0022] A 3、將正向比較結果和反向比較結果造行邏輯處理 ，判斷音頻信號是否為語音信號。 Ο [0023] 在本發明實施例中，通過將正向比較結果和反向比 較結果進行邏輯或處理，判斷接收的音頻信號是否為語 音信號。當正向比較結果或者反向比較結果中的一個為 邏輯真時，則邏輯或處理後的結果為邏輯真，判定音頻 信號為語音信號，輸出邏輯真；當正向比較結果和反向 比較結果均為假時，則邏輯或處理後的結果為邏輯假， 判定音頻信號為雜訊信號，輸出邏輯假。 [0024] A4、將該邏輯處理結果與前一次產生的消噪控制信 099129997 表單編號A0101 第7頁/共44頁 0992052617-0 201212008 [0025] [0026] [0027] [0028] [0029] [0030] 號進行邏輯處理，產生有效值檢測的觸發信號。 在本發明實施例中，通過將該邏輯處理結果與前一 次產生的消噪控制信號進行邏輯或處理，當該邏輯處理 結果或者前一次產生的消噪控制信號中的任意一個為邏 輯真時，則判定需要對音頻信號進行有效值檢測，輸出 的有效值檢測的觸發信號為邏輯真；當該邏輯處理結果 和前一次產生的消噪控制信號均為邏輯假時，則判定不 需要對音頻信號進行有效值檢測，輸出的有效值檢測的 觸發信號為邏輯假。 為了在音頻信號的幅度較小時，提高對音頻信號的 處理精度並減小誤差，在本發明另一實施例中，在判斷 接收的音頻信號是否為語音信號之前，該方法還包括下 述步驟： 對接收的音頻信號進行選擇性的預放大處理。 B、 對音頻信號進行有效值檢測，產生用於反映音頻 信號的平均功率的輸出信號。由於對音頻信號進行有效 值檢測的具體實現過程屬於現有技術，在此不再贅述。 C、 不對音頻信號進行有效值檢測，維持一固有初值 的輸出信號。其中固有初值為用於反映音頻信號中雜訊 信號的平均功率。 在本發明實施例中，通過對音頻信號中的語音信號 進行有效值檢測，在雜訊段内，保持一固有初值的輸出 ，從而在具體的電路設計中，一方面可以防止有效值檢 測部分在檢測語音信號平均功率開始時的延遲情況，即 099129997 表單編號A0101 第8頁/共44頁 0992052617-0 201212008 有有效值檢測的輸出信號較真實值偏小，另一方面可以 使其輸出在有雜訊的情況下保持恒定值，可以節省功耗 [0031] 在步驟S102中，將用於反映音頻信號的平均功率的 輸出信號與第一閾值進行比較，產生消噪控制信號。[0017] The invention is not intended to limit the invention. In the embodiment of the present invention, an output signal for reflecting the average power of the audio signal is obtained by performing RMS detection on the received audio signal, by performing the output signal for reflecting the average power of the audio signal with the first threshold. Comparing, generating a noise cancellation control signal, wherein the first threshold is a threshold value of an average power of the voice signal and an average power of the noise signal, and the noise signal in the audio signal is cancelled under the control of the noise cancellation control signal, The speech signal is amplified to achieve the elimination of speech background noise. The first figure shows the implementation flow of the method for canceling the voice background noise provided by the embodiment of the present invention, which is described in detail as follows: In step S101, the received audio signal is subjected to a valid value (also referred to as a root mean square value, Root). Mean Square, RMS) detects an output signal that is used to reflect the average power of the audio signal. In the embodiment of the present invention, the received audio signal may be an collected audio signal, or may be an audio signal transmitted from another device. Since the environment in which the speaker is located is generally complicated, there are various sources of sound other than the speaker, such as the voices of other speakers around, the sounds of surrounding vehicles, etc., so that the received audio signal includes not only the voice of the speaker. The signal also includes noise signals from other sources in the speaker's surroundings. Since the received audio signal includes a voice signal and a noise signal, the amplitude level generated by the voice signal is generally different from the level of the noise generated by the noise signal, and the RMS value is detected by the received audio signal according to the amplitude of the audio signal. The change may generate an output signal for reflecting the average power of the audio signal of 099129997 Form No. A0101 Page 5 / Total 44 Page 0992052617-0 201212008, that is, an output signal for reflecting the average power of the audio signal reflects the amplitude variation of the audio signal . By setting a threshold value of the average power of the voice signal and the average power of the noise signal according to the difference between the average power of the voice signal and the average power of the noise signal, as the first threshold, it can be used to reflect the audio. The output signal of the average power of the signal is compared with the first threshold to determine whether the audio signal is a speech signal or a noise signal, thereby generating a noise cancellation control signal. [0018] Since the audio signal is generally continuous, there may be a delay in performing RMS detection on the received audio signal, in order to avoid the accuracy of the speech denoising due to the delay of detecting the RMS value of the received audio signal, In another embodiment of the present invention, the step of performing RMS detection on the received audio signal to obtain an output signal for reflecting the average power of the audio signal specifically includes: [0019] A. determining whether the received audio signal is a voice signal, And generating trigger information of the effective value detection according to the judgment result and the previously generated noise cancellation control signal. When the trigger signal of the valid value detection indicates that the effective detection of the audio signal is required, step B is performed; otherwise, step C is performed. The specific process is as follows: [0020] A1, comparing the received audio signal with a second threshold, determining whether the audio signal is a voice signal, and obtaining a forward comparison result; wherein the second threshold is a minimum value of the forward voice signal, and the specific The setting depends on the sensitivity of the microphone, and the collection of different sound pressure noise in the case of the sensitivity. The specific setting work should be combined with acoustic theory and experimental statistics. When the received audio signal is compared with the second threshold, when the tone 099129997 form number A0101 page 6/44 page 0992052617-0 201212008 frequency signal is greater than the second threshold, it is determined that the audio signal is a voice signal, and a positive comparison is obtained. The result is logically true; when the audio signal is less than the second threshold, it is determined that the audio signal is a noise signal, and the forward comparison result is logically false. [0021] Α2, comparing the received audio signal with a preset third threshold, determining whether the audio signal is a voice signal, and obtaining a reverse comparison result; wherein the third threshold is a maximum value of the reverse voice signal, and the specific Depending on the sensitivity of the microphone and the sensitivity of the different sound pressure noise D, the specific setting work should be combined with acoustic theory and experimental statistics. When the received audio signal is compared with the preset third threshold, if the audio signal is less than the third threshold, it is determined that the audio signal is: 'alone|Big 曜 _ _ voice signal, and the reverse comparison result is logically true; If the audio signal is greater than the third threshold, it is determined that the audio signal is a noise signal, and the reverse comparison result is logically false. [0022] A3, the forward comparison result and the reverse comparison result are logically processed to determine whether the audio signal is a voice signal. [0023] In the embodiment of the present invention, whether the received audio signal is a speech signal is determined by logically processing the forward comparison result and the reverse comparison result. When one of the forward comparison result or the reverse comparison result is logically true, the logical or processed result is logically true, the audio signal is determined to be a speech signal, and the output logic is true; when the forward comparison result and the reverse comparison result are If they are all false, the logical or processed result is logically false, and the audio signal is determined to be a noise signal, and the output logic is false. [0024] A4, the logic processing result and the previous generation of the noise reduction control signal 099129997 Form No. A0101 Page 7 / Total 44 Page 0992052617-0 201212008 [0025] [0026] [0028] [0029] [0029] The 0030] number is logically processed to generate a trigger signal for rms detection. In the embodiment of the present invention, by logically processing the logical processing result and the previously generated denoising control signal, when any one of the logical processing result or the previous generated denoising control signal is logically true, Then, it is determined that the RMS value of the audio signal needs to be detected, and the trigger signal of the output RMS detection is logically true; when the logic processing result and the previously generated denoising control signal are both logically false, it is determined that the audio signal is not required. The rms detection is performed, and the trigger signal of the output RMS detection is logically false. In order to improve the processing precision of the audio signal and reduce the error when the amplitude of the audio signal is small, in another embodiment of the present invention, before determining whether the received audio signal is a voice signal, the method further includes the following steps. : Selective pre-amplification of the received audio signal. B. Performing an RMS detection on the audio signal to generate an output signal for reflecting the average power of the audio signal. The specific implementation process of performing the effective value detection on the audio signal belongs to the prior art, and details are not described herein again. C. The rms detection of the audio signal is not performed, and an output signal of an inherent initial value is maintained. The inherent initial value is used to reflect the average power of the noise signal in the audio signal. In the embodiment of the present invention, by detecting the effective value of the voice signal in the audio signal, an output of the inherent initial value is maintained in the noise segment, so that in the specific circuit design, the effective value detecting portion can be prevented on the one hand. The delay at the start of detecting the average power of the speech signal, ie 099129997 Form No. A0101 Page 8 / Total 44 Page 0992052617-0 201212008 The output signal with rms detection is smaller than the actual value, on the other hand, its output can be In the case of noise, a constant value is maintained, and power consumption can be saved. [0031] In step S102, an output signal for reflecting the average power of the audio signal is compared with a first threshold to generate a noise canceling control signal.
[0032] 在本發明實施例中，根據語音信號的平均功率和雜 訊信號的平均功率之間的差別，設置語音信號的平均功 率與雜訊信號的平均功率之間的臨界值，將該臨界值作 為第一閾值，通過將用於反映音頻信號的平均功率的輸 出信號與第一閾值進行比較，即可區分出音頻信號中的 語音信號和雜訊信號，從而產生用於控制消噪處理的消 噪控制信號。 [0033] 為了達到更舒適的語音信號，在本發明另一實施例 中，將用於反映音頻信號的平均功率的輸出信號與第一 閾值_進行比較，產生消噪控制信號的步驟i體包括：[0032] In the embodiment of the present invention, according to the difference between the average power of the voice signal and the average power of the noise signal, a critical value between the average power of the voice signal and the average power of the noise signal is set, and the threshold is set. The value is used as a first threshold, and by comparing the output signal for reflecting the average power of the audio signal with the first threshold, the voice signal and the noise signal in the audio signal can be distinguished, thereby generating a control for the noise canceling process. Denoising control signal. [0033] In order to achieve a more comfortable speech signal, in another embodiment of the invention, the output signal for reflecting the average power of the audio signal is compared with the first threshold _, and the step i of generating the denoising control signal includes :
[0034] A、將用於反映音頻信號的平均功率的輸出信號與第 四閾值進行比較，提取有效語音信號的控制信號； [0035] 其中第四閾值主要用於檢測有效的語音信號，即設 定雜訊門限。其設定具體要依靠麥克風的靈敏度，及該 靈敏度的情況下對不同聲壓雜訊的採集情況而定。具體 的設定工作要結合聲學理論和實驗統計。在本發明實施 例中，當檢測到有效語音信號時，輸出邏輯真，未檢測 到有效語音信號時，輸出邏輯假。 [0036] B、對提取的有效語音信號的控制信號進行補償，得 099129997 表單編號A0101 第9頁/共44頁 0992052617-0 201212008 到補償處理後的有效語音信號的控制信號； [0037] C、將用於反映音頻信號的平均功率的輸出信號與第 五閾值進行比較，提取超大信號的控制信號；其中第五 閾值主要用於檢測有效的語音信號，即設定雜訊門限。 其設定具體要依靠麥克風的靈敏度，及該靈敏度的情況 下對不同聲壓雜訊的採集情況而定。具體的設定工作要 結合聲學理論和實驗統計。在本發明實施例中，當檢測 到超大語音信號時，輸出邏輯真，未檢測到超大語音信 號時，輸出邏輯假。 [0038] D、將補償處理後的有效語音信號的控制信號和提取 的超大信號的控制信號進行異或處理，產生消噪控制信 號。 [0039] 在本發明實施例中，在將補償處理後的有效語音信 號的控制信號和提取的超大信號的控制信號進行異或處 理時，如果有效語音信號的控制信號和超大信號的控制 信號的邏輯值不相同時，輸出邏輯真，相同時，輸出邏 輯假。如當有效語音信號的控制信號和提取的超大信號 的控制信號均為邏輯真時，則表示檢測到有效語音信號 ，同時檢測到超大語音信號，異或處理後，產生的消噪 控制信號為邏輯假；當有效語音信號的控制信號和提取 的超大信號的控制信號均為邏輯假時，則表示未檢測到 有效語音信號，同時未檢測到超大語音信號，異或處理 後，產生的消噪控制信號仍然為邏輯假；當有效語音信 號的控制信號為邏輯真，提取的超大信號的控制信號均 為邏輯假時，則表示檢測到有效語音信號，同時未檢測 099129997 表單編號A0101 第10頁/共44頁 09920526]7-0 201212008 到超大語音信號，異或處理後，產生的消噪控制信號為 邏輯真；當有效語音信號的控制信號為邏輯假，提取的 超大信號的控制信號均為邏輯真時，則表示未檢測到有 效語音信號，同時未檢測到超大語音信號，異或處理後 ，產生的消噪控制信號為邏輯真。 [0040] 在步驟S103中，在消噪控制信號的控制下，對音頻 信號中的雜訊信號進行消除處理，對語音信號進行放大 處理。 Q [0041] 在本發明實施例中，當消噪控制信號為邏輯真時， 對音頻信號進行放大處理，當消噪控制信號為邏輯假時 ，對音頻信號進行消除處理，從而消除語音背景雜訊， 同時放大語音信號，實現語音背景雜訊的消除。其中在 消噪控制信號的控制下，對音頻信號中的雜訊信號進行 消除處理，對語音信號進行放大處理的方式有多種，如 在消噪控制信號的控制下，直接將音頻信號放大預設倍 ，或者禁止輸出音頻信號；也可以在消噪控制信號的控 制下，對音頻信號進行同相或者反相處理，以消除雜訊 信號，放大語音信號。當然還可以根據該消噪控制信號 ，採用其他方式對音頻信號進行放大處理，對音頻信號 進行消除處理，在此不再--舉例說明。 [0042] 由於接收的音頻信號中不可避免的會出現音頻過大 的情況，如果不對過大的音頻信號進行處理，則該音頻 信號後續被增強時，將降低聽者的舒適度，為了提高聽 者的舒適度，在本發明另一實施例中，在步驟S103之前 ，該方法還包括下述步驟： 099129997 表單編號A0101 第11頁/共44頁 0992052617-0 201212008 [0043] [0044] [0045] [0046] 將用於反映音頻信制平均功率的輸出信號與第六 閾值進行比較，產生衰減控制信號。該衰減控制信號用 於控制對接收的音頻信號進行不同比例的反相放大，以 適當的縮小過大的音頻信號。 其中第六閾值預先設置的可允許的音齡號的最大 功率。在本發明實施例中，當用於反映音頻信號的平均 功率的輸出信號大於第六閾值時，輸出的衰減控制信號 為邏輯真，該衰減控制信號用於控制對接收的音頻信號 進行一定比例的衰減，以適當的縮小過大的音頻信號； 當用於反映音頻信號的平均功率的輸出信號小於或者等 於第六閾值時’則輸出的衰減控制信號為邏輯假，該衰 減控制信號用於對音頻信號進行等幅的反相放大，即不 對音頻信號的幅度進行處理。 在衰減控制信號的控制下，對接收的音頻信號進行 不同比例的反相放大，以適當的縮小過大的音頻信號。 在本發明實施例中，奮音頻信號:的音量在正常範圍 時，衰減控制信號控制對音頻信號進行等幅的反相放大 ，即不對音頻信號的幅度進行處理，得到反相放大後的 信號；當音頻信號的音量超出正常範圍時，衰減控制信 號控制對過大的音頻信號進行一定比例的衰減，從而適 當的縮小過大的音頻信號，得到反相放大後的信號。 此時’在消噪控制信號的控制下，對反相放大後的 音頻信號進行消除或者放大處理，以消除反相放大後的 音頻信號中的雜訊信號，放大反相放大後的音頻信號中 099129997 表單編號A0101 第12頁/共44頁 0992052617-0 [0047] 201212008 [0048] [0049] Ο [0050] [0051]❹ [0052] 的語音信號。 第二圖示出了本發明實施例提供的語音背景雜訊消 除裝置的結構，為了便於說明，僅示出了與本發明實施 例相關的部分。其中： 有效值檢測單元（2)對接收的音頻信號進行有效 值檢測，得到用於反映音頻信號的平均功率的輸出信號 。其中有效值檢測單元（2)包括語音信號判斷電路（ 2 2)、有效值檢測控制電路（2 3)和有效值檢測電 路（24)。其中： 語音信號判斷電路（2 :2 )判斷接收的音頻信號是 否為語音信號。在本發明實施例中’為了提高判斷的準 確性，該語音信號判斷電路（2 2)包括正向語音信號 判斷模組（2 2 1 )、反向語音信號判斷模組（2 ^ )和第一邏輯處理模組（2 2 3)。 其中正向語音信號判斷模組（2 2將接收的音 頻㈣與第二閾值進行比較,，判斷音難號是否為語音 信號，得到正向比較結果； 其中第二閾值是正向語音信號的最小值其具體的 »又置要依靠麥克風的靈敏度，及該靈敏度的情況下對不 同聲Μ雜訊的採集情況’具體的設^作要結合聲學理 論和實驗統計。在將接收的音頻信號與第二間值進行比 較時，當音頻信號大於第二閾值，則判定音頻信號為語 音信號’得到正向比較結果為邏輯真；當音頻信號小: 第二閾值時’關定音頻信號為雜訊信號，得到正向比 099129997 表單編號A0101 第13頁/共44頁 0992052617-0 201212008 較結果為邏輯假。 [0053] 反向語音信號判斷模組（2 2 2 )將接收的音頻信 號與預設的第三閾值進行比較，判斷音頻信號是否為語 音信號，得到反向比較結果； [0054] 其中第三閾值是反向語音信號的最大值，其具體要 依靠麥克風的靈敏度，及該靈敏度的情況下對不同聲壓 雜訊的採集情況而定，具體的設定工作要結合聲學理論 和實驗統計。在將接收的音頻信號與預設的第三閾值進 行比較時，如果音頻信號小於第三閾值，則判定音頻信 號為語音信號，得到反向比較結果為邏輯真；如果音頻 信號大於第三閾值時，則判定音頻信號為雜訊信號，得 到反向比較結果為邏輯假。 [0055] 第一邏輯處理模組（2 2 3 )將正向比較結果和反 向比較結果進行邏輯處理，判斷接收的音頻信號是否為 語音信號。 [0056] 在本發明實施例中，通過將正向比較結果和反向比 較結果進行邏輯或處理，判斷接收的音頻信號是否為語 音信號。當正向比較結果或者反向比較結果中的一個為 邏輯真時，則邏輯或處理後的結果為邏輯真，判定音頻 信號為語音信號，輸出邏輯真；當正向比較結果和反向 比較結果均為假時，則邏輯或處理後的結果為邏輯假， 判定音頻信號為雜訊信號，輸出邏輯假。 有效值檢測控制電路（2 3)將第一邏輯處理模組 (2 2 3 )輸出的邏輯處理結果與消噪控制信號產生單 099129997 表單編號A0101 第14頁/共44頁 0992052617-0 [0057] 201212008 元（3)產生的前一次消噪控制信號進行邏輯處理，產 生有效值檢測的觸發信號，以控制是否對音頻信號進行 有效值檢測。 [0058] 在本發明實施例中，通過將該邏輯處理結果與消噪 控制信號產生單元（3 )產生的前一次消噪控制信號進 行邏輯或處理，當該邏輯處理結果或者前一次消噪控制 信號中的任意一個為邏輯真時，則判定需要對音頻信號 進行有效值檢測；當該邏輯處理結果和前一次消噪控制 信號均為邏輯假時，則判定不需要對音頻信號進行有效 〇 值檢測。 [0059] 有效值檢測電路（2 4 )在有效值檢測控制電路（ 2 3)的控制下，對音頻信號進行有效值檢測，產生用 於反映音頻信號的平均功率的輸出信號，或者不對音頻 信號進行有效值檢測，維持一固有初值的輸出信號。其 中固有初值為用於反映音頻信號中雜訊信號的平均功率 〇 〇 [0060] 在本發明實施例中，通過對音頻信號中的語音信號 進行有效值檢測，在雜訊段内，保持一固有初值的輸出 ，從而在具體的電路設計中，一方面可以防止有效值檢 測部分在檢測語音信號平均功率開始時的延遲情況，即 有有效值檢測的輸出信號較真實值偏小，另一方面可以 使其輸出在有雜訊的情況下保持恒定值，可以節省功耗 〇 [0061] 為了在音頻信號的幅度較小時，提高對音頻信號的 099129997 表單編號A0101 第15頁/共44頁 0992052617-0 201212008 處理精度並減小誤差，在本發明另—實施例中，該裝置 還包括信號預放大電路（2 1 )。該信號預放大電路（ 2 1 )的輸人端與音頻採集單元（丄）連接，其輸出端 分別與正向語音信號判斷模組（2 2工）、反相語音信 號判斷模組（2 2 2 )以及有效值檢測電路（2 4)連 接’該信號預放大電路（2 1 )對接㈣音頻信號進行 選擇性的預放大處理。此時，正向語音信號判斷模組（ 2 2 1)將信號預放大電路（2 1)輸出的音頻信號與 第二閾值騎比較，朗音頻㈣是否為語音信號，得 到正向比較結果；反向語音信號科斷模組（2 2 2)將 仏號預放大電路（2 1)輸出的音頻信號與預設的第三 閾值進行比較’判斷音頻信蚊Μ語音㈣，得到反 向比較結果’·有效值檢測電路（2 4)在有效值檢測控 制電路（2 3)的控制下’對信號預放大電路（2工） 輸出的音頻信號進行有效值檢測，產生用於反映音頻信 號的平均功率的輸出信號。在本發明另一實施例中在 有效值檢測控制電路（2 3)的控制下，不對音頻信號 進行有效值檢測，維持一固有初值的輸出信號。 [0062] 由於說話者所處的環境一般較為複雜，存在除說話 者以外的多種發聲源，如周圍其他說話者的聲音、周圍 父通工具的聲音等，因此，接收的音頻信號中不僅包括 說話者發出的語音信號，也包括說話者周圍環境其他聲 源發出的雜訊信號。 在本發明實施例中，接收的音頻信號可以為採集的 音頻信號，也可以為從其他設備傳輸過來的音頻信號。 099129997 表單編號Α0101 第16頁/共44頁 0992052617-0 [0063] 201212008 當接收的音頻信號為採集的音頻信號時，該裝置還包括 音頻採集單元（1 )。該音頻採集單元（1 )採集音頻 其中θ頻彳5號包括语音信號和雜訊信號。該音頻 採集單元（1 )的具體電路結構請參閱第三圖。在此不 再詳述β [0064] Ο 〇 [0065] 消噪控制信號產生單元（3)將有效值檢測單元（ 3)產生的用於反映音頻信號的平均功率的輸出信號與 第一閾值進行比較，產生消噪控制信號。該消噪控制信 號用於控制消噪放大單元對音頻信號進行消噪處理。其 中第一閾值是指預設的語音信號的平均功率與雜訊信號 的平均功率之間的臨界值。通過將用於反映音頻信號的 平均功率的輸出信號與第一間值進行比較，即可區分出 音頻信號中的語音信號和雜訊信號，從而產生用於控制 4嗓放大單元對音頻信號進行消噪處理的消噪控制信號 。該消噪控制信號產生單元（3)包括有效語音信號比 較模組（3 1)、超大語音信號比較稹組（3 2)、控 制信號補償模組（3 3)以及第二邏輯處理模組（3 4 )。其中： 有效s吾音彳§號比較模組（3 1)將有效值檢測單元 (2)產生的用於反映音頻信號的平均功率的輸出信號 與第四閾值進行比較，提取有效語音信號的控制信號； [0066] 099129997 其中第四閾值主要用於檢測有效的語音信號，即設 疋雜sfl門限。其設定具體要依靠麥克風的靈敏度，及該 靈敏度的情況下對不同聲壓雜訊的採集情況而定。具體 的設定工作要結合聲學理論和實驗統計。在本發明實施 表單編號A0101 第17頁/共44頁 0992052617-0 201212008 [0067] [0068] [0069] [0070] 例中，當檢測到有效語音信號時，輸出邏輯真，未檢測 到有效語音信號時，輸出邏輯假。 *控制信號補償模組（3 3 )對有效語音信號比較模 ”且（3 1 )提取的有效語音信號的控制信號進行補償， 得到補償處理後的有效語音信號的控制信號； 超大語音信觀較模組（3 2 )將有效值檢測單元 (2 )產生的用於反映音頻信號的平均功率的輪出信號 與第五閾錢行比較，提取超大«的控制信號； —其中第五閾值主要用於檢測有效的語音信號，即設 定雜訊Η限。其設定具體要依靠麥克風的靈敏度，及該 靈=度的情況下對不同聲Μ雜訊的採集情況:較。具體 的叹疋工作要結合聲學理論和實驗統計。在本發明實施 幻中*檢測到超大語音信號時，輸出邏_真未檢測 到超大語音信號時，輸出邏輯假。 第二邏輯處職組（34)將補魏理後的有效語 音信號的控制信號和提取的超大信號的控制信號進行異 或處理’產生消噪控制信號。 [0071] 省噪放大單元（4)在消噪控制信號產生單元（3 ▲產生的4噪控制信號的控制下，對音頻信號中的雜訊 信號進行消除處理，對語音㈣進行放大處理。 [0072] 099129997 立4噪放大早兀（4)在消噪控制信號的控制下對 音頻信號中的㈣信號進行祕處理，對語音信號進行 放大處理的方式有多種’如在消噪控制信號的控制下， 直：音頻信號放大預設倍，或者禁止輸出音頻信號； 編號A0101 第18頁/共44頁 0992052617-0 201212008 也可以在消噪控制信號的控制下，對音頻信號進行同相 :者反相處理，以消除雜訊信號’放大語音信號。當然 s以根據該消噪控制信號，採用其他方式對音頻信號 進订放大處理’對音頻信號進行祕處理，在此僅认出 -種消噪放大單元（4)的具體結構，其包括可控:相 。器（41)、第—反相跟隨器（42)、第二反相跟隨 盗（4 3 )、反相加法器（4 4)和低通濾波器（4 5 〇 100731 纟於純的音趣财不可㈣的會ώ現音頻過大 的情況，如果獨過大的音驗號㈣處理，則該音頻 信號後續被增強時’將降低聽者的舒適度V，為了提高聽 者的舒適度，在本發㈣―實施射，該裝置還包括衰 減控制信號產生單元（5)和音頻可控賴單元（6) 。其中： [0074] 〇 衰減控制信號產生單元（5)將有效值檢測單元（ 2)產生的用於反映音頻信_平均功率的輸出信號與 第六閾值進行比較’產生衰減控龍號。該衰減控制信 號用於控制音頻可控衰減單元（6)對接收的音頻信號 進行不同比例的反減大，以適#_小過大的音頻信 號。 [0075] 099129997 其中第六閾值預先設置的可允許的音頻信號的最大 功率。在本發明實施财，當用於反映音頻信號的不均 功率的輸出信號大於第六閣值時，輸出的衰減控制信號 為邏輯真；當用於反映音頻信號的平均功率的輸出信號 J於或者科第六’時’則輪出的衰減控制信號為邏 表單塢珑A0101 第19頁/共44頁 〇£ 201212008 輯假。 [0076] [0077] [0078] [0079] 音頻可控衰減單元（6)在衰減控制信號的控制下 ，對接收的音頻信號進行不同比例的反相放大，以適當 的縮小過大的音頻信號。 在本發明實施例中，當音頻信號的音量在正常範圍 時，衰減控制信號控制對音頻信號進行等幅的反相放大 ，即不對音頻信號的幅度進行處理，得到反相放大後的 信號；當音頻信號的音量超出正常範圍時，衰減控制信 號控制對過大的音頻信號進行一定比例的衰減，從而適 當的縮小過大的音頻信號，得到反相放大後的信號。 此時，消噪放大單元（2 )在消噪控制·信號的控制 下，對音頻可控衰減單元（60輸出的音頻信號進行消 除或者放大處理，以消除反相放大後的音頻信號中的雜 訊信號，放大反相放大後的音頻信號中的語音信號。 第三圖示出了本發明實施例提供的音頻採集單元（ 1 )的電路原理，該音頻採集單元（1 )包括由分壓電 路、隔直電路和反相跟隨器’其中分壓電路由麥克風^ 和電阻R1組成，隔直電路由電容C1構成，反相跟隨器由 運放Μ、電阻R2和電咖組成。其中麥克風為全指向麥 克風，如手機科減料㈣規麥W，祕感應外 界音頻信號，並將聲音信號轉化為電信^通過由第一 電容Π構隔直電路對麥克風得㈣電錢進行隔直 處理’將電信號轉«交流電信號，再經反相跟隨器輸 出信號①。 099129997 表單編號Α0101 第20頁/共料頁 201212008 [0080] 第四圖示出了本發明實施例提供的信號預放大電路 (2 1 )的電路原理。該信號預放大電路（2 1 )包括 運放U7，與運放U7的其中一輸入端連接的電阻R16和跨接 在運放的輸入端和輸出端的電阻R17。信號預放大電路（ 2 1 )根據系統實際需要的處理精度，通過調整電阻R16 和電阻R17的比例來實現對信號採集單元（1 )的輸出信 號①進行選擇性的預放大處理，得到預放大處理後的信 號⑦。 ^ [0081] 第五圖示出了本發明實施例提供的正向語音信號判 斷模組（2 2 1 )和反向語音信號判斷模組（2 2 2 ) 的電路原理。 [0082] 該正向語音信號判斷模組（2 2 1 )和反向語音信 號判斷模組（2 2 2 )均包括比較器U8和施密特反相器 (U9、U10)，比較器U8將信號預放大電路（2 1 )處理 後的音頻信號⑦與預設的第一閾值進行比較，再通過施 密特反相器（U9、U10 )對雜訊進行整形處理後，得到比 〇 較結果信號⑧；將信號預放大電路（2 1 )處理後的音頻 信號⑦與預設的第二閾值進行比較，再通過施密特反相 器（U9、U10)對雜訊進行整形處理後，得到比較結果信 號④。 [0083] 第六圖示出了本發明實施例提供的第一邏輯處理模 組（2 2 3 )的電路原理。 [0084] 該第一邏輯處理模組（2 2 3 )由邏輯或閘U11構成 ，邏輯或閘U11的兩個輸入信號分別為正向語音信號判斷 099129997 表單編號A0101 第21頁/共44頁 0992052617-0 201212008 模組（p 9 1、. 丄）的輸出信號⑧和反向語音信號判斷模組（ )的輪出k號④，當正向語音信號判斷模組（2 2 •的輪出6號⑧和反向語音信號判斷模組（2 2 2 )的 ' 號④中的任意一個為真時，即當正向語音信號判斷 矣符-f ^ c\ z 2 1 )和反向語音信號判斷模組（2 2 2 )中 •、個別疋音頻彳§號為語音信號時，經該邏輯或閘1 1 處理後’其輸出信號⑩為真’即經邏輯或閘un處理後， 判疋音頻信號為語音信號’否則，判定音頻信號為雜訊 信號。 [0085] [0086] 第七圖不出了本發明實施例提供的有效值檢測控制 電路（2 3)和有效值檢測電路（2 4)的電路原理。 有效值檢測控制電路（2 3 )包括邏輯或閘U13以及 二選一選擇開關S3。其中邏輯或閘U13的兩個輸入端分別 與第一邏輯處理模組（2 2 3)的輸出端連接，其另一 輸入端與消噪控制信號產生單元（3)包括的第二邏輯 處理模組（3 4)的輸出端連接，邏輯或閘μ3的輸出信 號作為二選一選擇開關S3的控，制像號，通過二選一選擇 開關S3控制有效值檢測電路（2 4 )對音頻信號進行有 效值檢測’產生用於反映音頻信號的平均功率的輸出信 號’或者不對音頻信號進行有效值檢測，而直接維持_ 固有初值V1的輸出。該電路工作原理如下： 邏輯或閘U13將第一邏輯處理模組（2 2 3 )的輸出 4吕说⑩與第一邏輯處理模組（3 4 )的輸出信號⑲進行 或處理，得到作為二選一選擇開關S3的控制信號⑪，於 099129997 表單編號A0101 第22頁/共44頁 0992052617-0 [0087] 201212008 [0088] ❹ [0089] Ο [0090] 099129997 制信號⑪控制二選一選擇開關在信號預放大電路（2工 )的輸出信號⑦和預設的固有初值V1之間進行切換，得 到輸出信號⑫，通過該輸出信號⑫控制有效值檢測電路 (2 4)對信號預放大電路（2丄）的輸出信號⑦進行 有效值檢測，並產生用於反映音頻信號的平均功率的輸 出信號DD ’或者不對音頻信號進行有效值檢測，而直接 維持一固有初值VI的輸出。 第八圖示出了本發明實施例提供的有效語音信號比 較模組（3 1 )和超大語音信號此較模組（3 2)的電 路原理。 該有效語音信號比較模組（3 1 )包括丨比較器ui5和 施密特反相器（U17、U18)」比較器U15將有效值檢測 電路（2 4 )的輸出信號DD與預設的第四閾值進行比較 ，以針對有效語音信號進行控制信號的提取，輸出有效 語音的控制信號⑭。其中第四閾值主要^檢測有效的 語音信號，即設定雜訊門限。其設定具'體要依靠麥克風 的靈敏度，及該靈敏度的情況下對不同聲壓雜訊的採集 情況而定，具體的設定工作要結合聲學理論和實驗統計 。在進行比較時，信號DD大於第四閾值時，有效語音信 號比較模組（3 1 )的輸出為邏輯真，代表語音信號。 小於第四閾值時’有效語音信號比較模組（31)的輸 出邏輯假，代表雜訊信號。 該超大語音信號比較模組（3 2 )包括比較器U16和 施密特反相器（U19、U20) ’比較器U16將有效值檢測 電路（2 4 )的輸出信號DD與預設的第五閾值進行比較 表單煸號A0101 第23頁/共44頁 0992052617-0 201212008 ，以針對超大信號進行控制信號的提取，輸出超大信號 的控制信號⑮。在通訊的過程中，難免通訊環境中會出 現風聲，突發的超大雜訊會給接收方（聽者）以刺耳的 聲音，在這種情況下，通過該超大語音信號比較模組（ 3 2 )針對超大信號進行控制信號的提取，從而使消噪 放大單元（4 )對超大信號進行消除處理，提高收聽者 的聽覺舒適度。 [0091] [0092] [0093] [0094] 第九圖示出了本發明實施例提供的控制信號補償模 組（3 3)的電路原理。 控制信號補償模組（3 3 )包括多諧振盪器U21和或 閘U22。多諧振盪器U21檢測有效語音信號比較模組（3 1 )的輸出脈衝的下降沿，一旦檢測到下降沿，多諧振 盪器就輸出一特有寬度的脈衝信號，經過或閘U22後可以 將有效語音信號比較模組（3 1 )的輸出脈衝補償加寬 這一特有脈衝寬度，得到補償處理後的有效語音的控制 信號⑱。這樣在語音結束端，能夠減少語音的損失，高 品質的還原語音。 第十圖示出了本發明實施例提供的第二邏輯處理模 組（3 4)的電路原理。 該第二邏輯處理模組（3 4 )由邏輯異或閘U23構成 。控制信號補償模組（3 3 )的輸出信號⑯和超大語音 信號比較模組（3 2 )的輸出信號⑬分別送到第二邏輯 處理模組（3 4)的輸入端，兩路信號經過異或邏輯形 成消噪控制信號⑰，該消噪控制信號⑰用於控制消噪放 099129997 表單編號A0101 第24頁/共44頁 0992052617-0 201212008 [0095] [0096] Ο Ο [0097] = 對音财控衰減單元（6)處理後的音頻 α就進行消嚼處理。 ^ J ^出了本發明實施例提供的衰減控制信號 產生早元（5)的電路原理。 =衰減控制信财生單元⑴由峨腸2構成， 父器U12的其巾-個輸人信號為有效值檢測電路（2 輪出_於反映音頻信_平均功率的輪出信號卯 iKt號為預設的第六閾值’比較器川將輸出 和預設的第六閣值進行比較，產生衰減控制信號 ⑱。該衰減㈣健_於㈣音射控衰減單元（6 )對音頻採集單元（1 )輪“音頻域進行不同比例 的反相放大。在本發明實施例卜，當用於反映音頻信號 的平均功率的輸出信號大於預設的第六^隹時，產生的 哀減控制信號⑱，控制音頻可控衰減箪元（6)對音頻 信號進行等幅的反相放大，即不對音頻信號的幅度進行 衰減處理；當用於反_音頻信轉的平均功率的輸出信號 小於或者等於預設的第六®'值秦，產生的衰減控制信號 ⑱控制音頻可控衰減單元（6)對過大的音頻信號進行 一定比例的衰減，以適當的縮小過大的音頻信號。 第十二圖示出了本發明實施例提供的音頻可控衰減 單元（6)的電路原理。 該音頻可控衰減單元（6 )包括運放U2、控制開關 S1、電阻R3、R4 、R5。其中電阻R3和電阻R4串連後與 運放U2的輸入端連接，控制開關S1並聯在電阻R3的兩端 099129997 表單編St A0101 第25頁/共44頁 0992052617-0 [0098] 201212008 ’電阻R5跨接在運放U2的輸入端和輸出端。控制開關SI 的控制信號來自衰減控制信號產生單元（5 )產生的衰 減控制信號⑱’衰減控制信號⑩控制該控制開關的導通 與斷開’從而實現對音頻採集單元（1 )輸出的音頻信 號①進行不同比例的反相放大，得到反相放大後的信號 ②。在本發明實施例中，當音頻信號的音量在正常範圍 時’衰減控制信號⑱控制音頻可控衰減單元（6)對音 頻信號進行等幅的反相放大，即不對音頻信號的幅度進 行處理’得到反相放大後的信號②;當音頻信號的音量 超出正常範圍時，衰咸控制信號⑱控制音頻可控衰減單 元（6)對過大的音頻信號也行一定比例的衰減，從而 適當的縮小過大的音頻信號，得到反相、玫大後的信號② [0099] [0100] 099129997 第十二圖示出了本發明實施例提供的可控反相器（ 4 1 )的電路原理。 該可控反相器(4 1 )的兩偶輪入端分別與音頻可 控哀減單元（6)以及第二邏輯處理模組（3 4)的輸 出端連接。該可控反相器（4 1 )包括運放U3、電阻R6 、R7和二選一選擇開關S2，其中電阻R6與運放U3的其中 一輸入端連接’電阻R7跨接於運放U3的輪入端和輸出端 ，二選一選擇開關S2在消噪控制信號⑬的控制下，對反 相放大後的信號①進行同相或者反相處理，得到信號③ 。當進行同相處理時，增強反相放大後的信號②，得到 信號③；當進行反相處理時，消除雜訊，得到信號③。 表單煸號A0101 第26頁/共44頁 0992052617-0 201212008 [0101] 第十四、十五圖示出了本發明實施例提供的第— 相跟隨器（4 2)和第二反相跟隨器（4 3)的電路原 理α [0102] 其中第一反相跟隨器（4 2)的輸入端與音頻可控 衰滅單元（6)的輪出端連接，其輸出端與反相加法器 (4 1)的輸入端連接。該第一反相跟隨器（4 2 )包 括運放U4，與運放U4的其中一個輸入端連接的電阻R8， 以及跨接在運放U4的輪入端和輸出端的電阻R9。其輸入 〇 信號為音頻可控衰減單元（6)的輸出信號①，輸出信 號④；第二反相跟隨器（4 3)的輸入端與可控反相器 (4 1)連接，其輸出端與反相加法器沾輸^^端連接。 該第二反相跟隨器（4 3 )包括運放115，.與運放U5的其 中一個輸入端連接的電阻R1〇，以及跨接在運放U5的輸入 端和輸出端的電阻R11。其輸入信號為反相加法器（4 1 )的輸出信號③，輸出信號⑤。 ^ [0103] 〇 第十六圖示出了本發明實.施例提供的反相加法器（ 4 4)的電路原理。 μ..... [0104] 該反相加法器（4 4)的兩個輸入端分別與第一反 相跟隨器（4 2)和第二反相跟隨器（4 3)的輸出端 連接，其包括運放U6，與運放U6的其中一輸入端連接的 電阻R12，與運放U6的另一輸入端連接的電阻R13，跨接 在運放U6的電阻R1 3所在的輸入端和運放U6的輸出端的電 阻R14。該反相加法器（4 4)的兩個輸入信號分別為第 一反相跟隨器（4 2 )的輸入信號④、第二反相跟隨器 099129997 表單編號A0101 第27頁/共44頁 0992052617-0 201212008 (4 3 )的輸出信號⑤，輸出信號④。 [0105] 第十七圖示出了本發明實施例提供的低通濾波器（ 4 5 )的電路原理。 [0106] 該低通濾波器的輸入端與反相加法器（4 4 )的輸 出端連接。其包括第電阻R15和電容C2。該低通濾波器（ 4 4)濾出裝置中額外產生的語音帶外的雜訊，保留語 音帶内的頻率成分。 [0107] 在本發明實施例中，通過對接收的音頻信號進行有 效值檢測，得到用於反映音頻信號的平均功率的輸出信 號，通過將該用於反映音頻信號的平均功率的輸出信號 與第一閾值進行比較，生成消噪控制信號，在消噪控制 信號的控制下對音頻信號中的雜訊信號進行消除處理， 對語音信號進行放大處理，從而實現語音背景雜訊的消 除。同時通過將用於反映音頻信號的平均功率的輸出信 號與第六閾值進行比較，產生衰減控制信號，在衰減控 制信號的控制下對接收的音頻信號進行不同比例的反相 放大，以適當的縮小過大的音頻信號，提高接收方的聽 覺舒適度。 [0108] 綜上所述，本創作確實可達到上述諸項功能及目的 ，故本創作應符合專利申請要件，爰依法提出申請。 【圖式簡單說明】 [0109] 第一圖係本發明實施例提供的語音背景雜訊的消除方法 的實現流程圖； [0110] 第二圖係本發明實施例提供的語音背景雜訊消除裝置的 099129997 表單編號A0101 第28頁/共44頁 0992052617-0 201212008 結構框圖； [0111] 第三圖係本發明實施例提供的音頻採集單元的電路原理 圖； [0112] 第四圖係本發明實施例提供的信號預放大電路的電路原 理圖； [0113] 第五圖係發明實施例提供的正向語音信號判斷模組和反 向語音信號判斷模組的電路原理圖； [0114] 第六圖係本發明實施例提供的第一邏輯處理模組的電路 原理圖； [0115] 第七圖係本發明實施例提供的有效值檢測控制電路和有 效值檢測電路的電路原理圖； [0116] 第八圖係本發明實施例提供的有效語音信號比較模組和 超大語音信號比較模組的電路原理圖； [0117] 第九圖係本發明實施例提供的控制信號補償模組的電路 原理圖； [0118] 第十圖係本發明實施例提供的第二邏輯處理模組的電路 原理圖； [0119] 第十一圖係本發明實施例提供的衰減控制信號產生單元 的電路原理圖； [0120] 第十二圖係本發明實施例提供的音頻可控衰減單元的電 路原理圖； [0121] 第十三圖係本發明實施例提供的可控反相器的電路原理 099129997 表單編號A0101 第29頁/共44頁 0992052617-0 201212008 圖； [0122] 第十四、十五圖係本發明實施例提供的第一反相跟隨器 和第二反相跟隨器的電路原理圖； [0123] 第十六圖分別係為本發明實施例提供的反相加法器的電 路原理圖； [0124] 第十七圖係本發明實施例提供的低通濾波器的電路原理 圖。 【主要元件符號說明】 [0125] ( 1 )輸入端與音頻採集單元 [0126] ( 2 )有效值檢測單元 [0127] ( 2 1 )信號預放大電路 [0128] ( 2 2 )語音信號判斷電路 [0129] ( 2 2 1 )正向語音信號判斷模組 [0130] ( 2 2 2 )反向語音信號判斷模組 [0131] ( 2 2 3 )第一邏輯處理模組 [0132] ( 2 3 )有效值檢測控制電路 [0133] ( 2 4 )有效值檢測電路 [0134] ( 3 )消嗓控制信號產生單元 [0135] ( 3 1 )有效語音信號比較模組 [0136] ( 3 2 )超大語音信號比較模組 [0137] ( 3 3 )控制信號補償模組 0992052617-0 099129997 表單編號A0101 第30頁/共44頁 201212008 [0138] [0139] [0140] [0141] [0142] [0143] [0144] [0145] [0146][0034] A, the output signal for reflecting the average power of the audio signal is compared with a fourth threshold, and the control signal of the effective voice signal is extracted; [0035] wherein the fourth threshold is mainly used for detecting a valid voice signal, that is, setting Noise threshold. The setting depends on the sensitivity of the microphone, and the sensitivity of the situation depends on the collection of different sound pressure noise. The specific setting work should be combined with acoustic theory and experimental statistics. In the embodiment of the present invention, when a valid speech signal is detected, the output logic is true, and when the valid speech signal is not detected, the logic false is output. [0036] B, the control signal of the extracted valid voice signal is compensated, and 099129997 Form No. A0101 Page 9 / Total 44 Page 0992052617-0 201212008 Control signal to the compensated effective voice signal; [0037] The output signal for reflecting the average power of the audio signal is compared with a fifth threshold to extract a control signal of the oversized signal; wherein the fifth threshold is mainly used for detecting a valid voice signal, that is, setting a noise threshold. The setting depends on the sensitivity of the microphone, and the sensitivity is determined by the collection of different sound pressure noise. The specific setting work should be combined with acoustic theory and experimental statistics. In the embodiment of the present invention, when the oversized speech signal is detected, the output logic is true, and when the oversized speech signal is not detected, the logic is false. [0038] D. XOR processing the control signal of the compensated effective speech signal and the extracted super large signal to generate a denoising control signal. [0039] In the embodiment of the present invention, when the control signal of the compensated effective speech signal and the control signal of the extracted oversized signal are XORed, if the control signal of the valid speech signal and the control signal of the oversized signal are When the logical values are different, the output logic is true. When the same, the output logic is false. For example, when the control signal of the effective speech signal and the control signal of the extracted super-large signal are both logically true, it means that the valid speech signal is detected, and the super-large speech signal is detected, and after the exclusive-OR processing, the generated de-noise control signal is logic. False; when the control signal of the effective speech signal and the control signal of the extracted super-large signal are both logically false, it means that no valid speech signal is detected, and no over-large speech signal is detected, and the de-noise control is generated after the exclusive-OR processing The signal is still logically false; when the control signal of the effective speech signal is logically true, and the control signal of the extracted super-large signal is logically false, it indicates that a valid speech signal is detected, and the detection is not detected. 099129997 Form No. A0101 Page 10 / Total 44 pages 09920526] 7-0 201212008 To the super large speech signal, after the XOR processing, the generated denoising control signal is logically true; when the control signal of the effective speech signal is logically false, the control signals of the extracted super large signal are all logically true. When it is, no valid speech signal is detected, and no excessive speech signal is detected, XOR After processing, the control signal generated noise canceling logically true. [0040] In step S103, under the control of the noise canceling control signal, the noise signal in the audio signal is subjected to cancellation processing, and the voice signal is amplified. [0041] In the embodiment of the present invention, when the noise cancellation control signal is logic true, the audio signal is amplified, and when the noise cancellation control signal is logic false, the audio signal is cancelled, thereby eliminating the background noise. At the same time, the voice signal is amplified at the same time to eliminate the noise of the background background. In the control of the noise-cancellation control signal, the noise signal in the audio signal is cancelled, and the voice signal is amplified. For example, under the control of the noise-cancellation control signal, the audio signal is directly amplified. Times, or prohibits the output of the audio signal; the audio signal may also be in-phase or inverted under the control of the noise-cancellation control signal to eliminate the noise signal and amplify the voice signal. Of course, according to the denoising control signal, the audio signal can be amplified by other methods, and the audio signal can be eliminated, which is not illustrated here. [0042] Since the audio is too large in the received audio signal, if the audio signal is not processed, the audio signal is subsequently enhanced, which reduces the comfort of the listener, in order to improve the listener's Comfort, in another embodiment of the present invention, before the step S103, the method further comprises the following steps: 099129997 Form No. A0101 Page 11 / Total 44 Page 0992052617-0 201212008 [0043] [0044] [0045] 0046] The output signal for reflecting the average power of the audio signal is compared to a sixth threshold to generate an attenuation control signal. The attenuation control signal is used to control different ratios of inverting amplification of the received audio signal to appropriately reduce the excessively large audio signal. The maximum power of the allowable sound age number in which the sixth threshold is preset. In the embodiment of the present invention, when the output signal for reflecting the average power of the audio signal is greater than the sixth threshold, the output attenuation control signal is logic true, and the attenuation control signal is used to control a certain proportion of the received audio signal. Attenuating to appropriately reduce the excessively large audio signal; when the output signal for reflecting the average power of the audio signal is less than or equal to the sixth threshold, then the output of the attenuation control signal is a logical false, and the attenuation control signal is used for the audio signal The inversion amplification of the equal amplitude is performed, that is, the amplitude of the audio signal is not processed. Under the control of the attenuation control signal, the received audio signals are inversely amplified in different proportions to appropriately reduce the excessively large audio signals. In the embodiment of the present invention, when the volume of the audio signal is in the normal range, the attenuation control signal controls to perform equal-amplitude inversion amplification on the audio signal, that is, the amplitude of the audio signal is not processed, and the inversely amplified signal is obtained; When the volume of the audio signal is out of the normal range, the attenuation control signal controls a certain proportion of attenuation of the excessive audio signal, thereby appropriately reducing the excessively large audio signal to obtain an inverted amplified signal. At this time, under the control of the noise canceling control signal, the inverted and amplified audio signal is cancelled or amplified to eliminate the noise signal in the inverted amplified audio signal, and the amplified and inverted amplified audio signal is 099129997 Form No. A0101 Page 12 of 44 0992052617-0 [0047] 201212008 [0048] [0050] [0051] The speech signal of [0052]. The second figure shows the structure of the speech background noise canceling apparatus provided by the embodiment of the present invention. For the convenience of description, only the parts related to the embodiment of the present invention are shown. Wherein: the RMS detecting unit (2) performs RMS detection on the received audio signal to obtain an output signal for reflecting the average power of the audio signal. The effective value detecting unit (2) includes a speech signal judging circuit (22), an effective value detecting control circuit (23), and an effective value detecting circuit (24). Wherein: the speech signal judging circuit (2:2) determines whether the received audio signal is a speech signal. In the embodiment of the present invention, in order to improve the accuracy of the judgment, the speech signal judging circuit (2 2) includes a forward speech signal judging module (2 2 1 ), a reverse speech signal judging module (2^), and a A logic processing module (2 2 3). The forward speech signal judging module (2 2 compares the received audio (4) with the second threshold, determines whether the sound difficulty number is a speech signal, and obtains a positive comparison result; wherein the second threshold is a minimum value of the forward speech signal The specific » depends on the sensitivity of the microphone, and the acquisition of different sonar noise in the case of the sensitivity. The specific settings should be combined with acoustic theory and experimental statistics. The received audio signal and the second When the inter-value is compared, when the audio signal is greater than the second threshold, it is determined that the audio signal is a speech signal 'to obtain a positive comparison result is logically true; when the audio signal is small: the second threshold is to 'close the audio signal as a noise signal, Get the positive ratio 099129997 Form No. A0101 Page 13 / Total 44 Page 0992052617-0 201212008 The result is logically false. [0053] The reverse speech signal judgment module (2 2 2) will receive the audio signal with the preset number Comparing the three thresholds to determine whether the audio signal is a speech signal, and obtaining a reverse comparison result; wherein the third threshold is a maximum value of the reverse speech signal, It depends on the sensitivity of the microphone, and the sensitivity of the situation, depending on the collection of different sound pressure noise, the specific setting work should be combined with acoustic theory and experimental statistics. The received audio signal is compared with the preset third threshold. In comparison, if the audio signal is less than the third threshold, it is determined that the audio signal is a voice signal, and the reverse comparison result is logically true; if the audio signal is greater than the third threshold, the audio signal is determined to be a noise signal, and a reverse comparison is obtained. The result is logical false. [0055] The first logic processing module (2 2 3 ) logically processes the forward comparison result and the reverse comparison result to determine whether the received audio signal is a voice signal. [0056] In the implementation of the present invention In the example, whether the received audio signal is a voice signal is determined by logically processing the forward comparison result and the reverse comparison result. When one of the forward comparison result or the reverse comparison result is logically true, the logical OR The processed result is logically true, and the audio signal is determined to be a voice signal, and the output logic is true; when the forward comparison result is When the comparison result is false, the logical or processed result is logically false, and the audio signal is determined to be a noise signal, and the output logic is false. The valid value detection control circuit (23) will be the first logic processing module (2 2) 3) Output logic processing result and noise reduction control signal generation unit 099129997 Form No. A0101 Page 14 / Total 44 Page 0992052617-0 [0057] 201212008 Element (3) The previous denoising control signal is logically processed to produce effective The value detection trigger signal controls whether the audible value detection is performed on the audio signal. [0058] In the embodiment of the present invention, the previous denoising control generated by the logic processing result and the noise cancellation control signal generating unit (3) The signal is logically ORed, and when any one of the logic processing result or the previous denoising control signal is logically true, it is determined that an RMS value detection is required for the audio signal; when the logic processing result and the previous denoising control signal are When both are logically false, it is determined that there is no need to perform effective threshold detection on the audio signal. [0059] The RMS detection circuit (24) performs RMS detection on the audio signal under the control of the RMS detection control circuit (23) to generate an output signal for reflecting the average power of the audio signal, or an audio signal Perform rms detection to maintain an output signal of an inherent initial value. The inherent initial value is used to reflect the average power of the noise signal in the audio signal. [0060] In the embodiment of the present invention, by detecting the effective value of the voice signal in the audio signal, the noise is maintained in the noise segment. The output of the inherent initial value, so that in the specific circuit design, on the one hand, the delay of the RMS detection part at the start of detecting the average power of the speech signal can be prevented, that is, the output signal with the RMS detection is smaller than the true value, and the other It can save its output with a constant value in the presence of noise, which can save power. [0061] In order to increase the amplitude of the audio signal, increase the number of the audio signal 099129997 Form No. A0101 Page 15 of 44 0992052617-0 201212008 Processing accuracy and reducing error, in another embodiment of the invention, the apparatus further comprises a signal preamplifying circuit (2 1 ). The input end of the signal pre-amplifying circuit (2 1 ) is connected to the audio collecting unit (丄), and the output end thereof is respectively combined with a forward speech signal judging module (2 2 working) and an inverted speech signal judging module (2 2 2) and the rms detection circuit (24) is connected to the signal preamplifier circuit (2 1 ) to perform selective pre-amplification processing on the (4) audio signal. At this time, the forward speech signal judging module (2 2 1) compares the audio signal outputted by the signal preamplifying circuit (2 1) with the second threshold riding, and whether the Lang audio (4) is a speech signal, and obtains a positive comparison result; To the voice signal section module (2 2 2), the audio signal outputted by the apostrophe preamplifier circuit (2 1) is compared with a preset third threshold value [judge the audio signal mosquito Μ voice (4), and the reverse comparison result is obtained] The RMS detection circuit (24) performs RMS detection on the audio signal output from the signal preamplification circuit (2) under the control of the RMS detection control circuit (23) to generate an average power for reflecting the audio signal. Output signal. In another embodiment of the present invention, under the control of the rms detection control circuit (23), the rms value is not detected for the audio signal, and an output signal of an inherent initial value is maintained. [0062] Since the environment in which the speaker is located is generally complicated, there are multiple sources of sound other than the speaker, such as the voices of other speakers around, the voice of the surrounding parent tools, etc., so the received audio signal includes not only the voice. The voice signal sent by the person also includes the noise signal from other sources in the speaker's surroundings. In the embodiment of the present invention, the received audio signal may be an collected audio signal, or may be an audio signal transmitted from another device. 099129997 Form number Α0101 Page 16 of 44 0992052617-0 [0063] 201212008 When the received audio signal is the acquired audio signal, the device also includes an audio collection unit (1). The audio collecting unit (1) collects audio, wherein the θ frequency 彳 5 includes a voice signal and a noise signal. Please refer to the third figure for the specific circuit structure of the audio acquisition unit (1). The denoising control signal generating unit (3) performs the output signal generated by the effective value detecting unit (3) for reflecting the average power of the audio signal with the first threshold value. [0065] In comparison, a noise cancellation control signal is generated. The noise canceling control signal is used to control the noise canceling amplifying unit to perform noise canceling processing on the audio signal. The first threshold is a threshold between the average power of the preset speech signal and the average power of the noise signal. By comparing the output signal for reflecting the average power of the audio signal with the first value, the voice signal and the noise signal in the audio signal can be distinguished, thereby generating a control for the 4 嗓 amplification unit to cancel the audio signal. Noise canceling control signal for noise processing. The noise reduction control signal generating unit (3) comprises an effective voice signal comparison module (31), a large voice signal comparison group (32), a control signal compensation module (33), and a second logic processing module ( 3 4 ). Wherein: the effective s 彳 彳 § number comparison module (3 1) compares the output signal generated by the effective value detecting unit (2) for reflecting the average power of the audio signal with the fourth threshold, and extracts the control of the effective speech signal Signal [0066] 099129997 The fourth threshold is mainly used for detecting a valid speech signal, that is, setting a noisy sfl threshold. The setting depends on the sensitivity of the microphone, and the sensitivity of the situation depends on the collection of different sound pressure noise. The specific setting work should be combined with acoustic theory and experimental statistics. In the present invention, the form number A0101, page 17 / page 44 0992052617-0 201212008 [0067] [0070] In the example, when a valid voice signal is detected, the output logic is true, and no valid voice is detected. When the signal is signaled, the output logic is false. * The control signal compensation module (3 3 ) compensates the control signal of the effective speech signal extracted by the effective speech signal comparison mode and (3 1 ), and obtains the control signal of the valid speech signal after the compensation processing; The module (32) compares the round-out signal generated by the effective value detecting unit (2) for reflecting the average power of the audio signal with the fifth threshold line, and extracts the control signal of the oversized «; wherein the fifth threshold is mainly used In order to detect the effective speech signal, the noise limit is set. The setting depends on the sensitivity of the microphone, and the collection of different sonar noise in the case of the Ling = degree: the specific sigh work should be combined Acoustic theory and experimental statistics. In the implementation of the invention, when the super-large speech signal is detected, the output logic _ true does not detect the super-large speech signal, the output logic is false. The second logical service group (34) will make up the Wei Li The control signal of the effective speech signal and the extracted control signal of the oversized signal are XORed to generate a denoising control signal. [0071] The noise reduction amplifying unit (4) produces the denoising control signal The unit (3 ▲ generated by the 4-noise control signal, the noise signal in the audio signal is eliminated, and the voice (4) is amplified. [0072] 099129997 Vertical 4 noise amplification early (4) in the noise cancellation control Under the control of the signal, the (four) signal in the audio signal is secretly processed, and there are various ways to amplify the speech signal. For example, under the control of the denoising control signal, the audio signal is amplified by a preset multiple, or the audio signal is prohibited from being output. ; No. A0101 Page 18 of 44 0992052617-0 201212008 It is also possible to perform in-phase: inverting processing of the audio signal under the control of the noise-cancellation control signal to eliminate the noise signal 'enlarge the voice signal. Of course s The denoising control signal adopts other methods for ordering and amplifying the audio signal 'the secret processing of the audio signal, and only recognizes the specific structure of the denoising and amplifying unit (4), which includes controllable phase: (41), the first-inverting follower (42), the second inverting follower (4 3 ), the inverting adder (4 4), and the low-pass filter (4 5 〇100731 纟 pure pure fun Not (four) If the audio is too large, if the audio signal is subsequently enhanced, it will reduce the comfort of the listener. In order to improve the comfort of the listener, in this issue (4) The apparatus further includes an attenuation control signal generating unit (5) and an audio controllable unit (6), wherein: [0074] the 〇 attenuation control signal generating unit (5) generates the RMS value detecting unit (2) The output signal reflecting the audio signal _average power is compared with a sixth threshold value to generate an attenuation control dragon number. The attenuation control signal is used to control the audio controllable attenuation unit (6) to perform different proportions of the received audio signal. To the #_ small oversized audio signal. [0075] 099129997 wherein the sixth threshold is preset with a maximum power of the allowable audio signal. In the implementation of the present invention, when the output signal for reflecting the uneven power of the audio signal is greater than the sixth value, the output of the attenuation control signal is logically true; when the output signal J for reflecting the average power of the audio signal is or The sixth time 'time' is the attenuation control signal for the logical form docking A0101 page 19 / a total of 44 pages 2012 £ 201212008. [0079] [0079] The audio controllable attenuation unit (6) performs different proportions of inversion amplification on the received audio signal under the control of the attenuation control signal to appropriately reduce the excessively large audio signal. In the embodiment of the present invention, when the volume of the audio signal is in the normal range, the attenuation control signal controls to perform equal-amplitude inversion amplification on the audio signal, that is, the amplitude of the audio signal is not processed, and the inversely amplified signal is obtained; When the volume of the audio signal exceeds the normal range, the attenuation control signal controls a certain proportion of attenuation of the excessive audio signal, thereby appropriately reducing the excessively large audio signal to obtain an inverted amplified signal. At this time, the noise canceling amplifying unit (2) cancels or amplifies the audio signal outputted by the audio controllable attenuation unit (60) under the control of the noise canceling control signal to eliminate the mismatch in the inverted amplified audio signal. The signal is amplified to amplify the voice signal in the amplified audio signal. The third figure shows the circuit principle of the audio collecting unit (1) provided by the embodiment of the present invention, and the audio collecting unit (1) includes a piezoelectric component The circuit, the DC-blocking circuit and the inverting follower are composed of a piezoelectric routing microphone ^ and a resistor R1. The blocking circuit is composed of a capacitor C1, and the inverter follower is composed of an operational amplifier, a resistor R2 and an electric coffee. Omnidirectional microphone, such as mobile phone section reduction (four) gauge wheat W, secret sensing external audio signal, and convert the sound signal into telecommunications ^ through the first capacitor 隔 隔 直 对 对 对 对 麦克风 麦克风 麦克风 麦克风 麦克风 四 四The electrical signal is converted to an "alternating current signal, and then an inverted follower output signal is 1. 099129997 Form number Α 0101 page 20 / common page 201212008 [0080] The fourth figure shows an embodiment of the present invention The circuit principle of the preamplifier circuit (2 1 ). The signal preamplifier circuit (2 1 ) includes an operational amplifier U7, a resistor R16 connected to one of the input terminals of the operational amplifier U7, and an input terminal and output across the operational amplifier The resistor R17 of the terminal. The signal preamplifier circuit (2 1 ) realizes the selective preamplification processing of the output signal 1 of the signal acquisition unit (1) by adjusting the ratio of the resistor R16 and the resistor R17 according to the processing precision required by the system. The pre-amplified signal 7 is obtained. [0081] The fifth figure shows the forward speech signal judging module (2 2 1 ) and the reverse speech signal judging module (2 2 2 ) provided by the embodiment of the present invention. The circuit principle is as follows: [0082] The forward speech signal judging module (2 2 1 ) and the reverse speech signal judging module (2 2 2 ) both include a comparator U8 and a Schmitt inverter (U9, U10). The comparator U8 compares the audio signal 7 processed by the signal preamplifier circuit (2 1 ) with a preset first threshold value, and then shapes the noise by the Schmitt inverter (U9, U10). Obtaining a comparison result signal 8; pre-amplifying the signal 2 1) The processed audio signal 7 is compared with a preset second threshold, and the noise is shaped by the Schmitt inverter (U9, U10) to obtain a comparison result signal 4. [0083] The figure shows the circuit principle of the first logic processing module (2 2 3 ) provided by the embodiment of the present invention. [0084] The first logic processing module (2 2 3 ) is composed of logic or gate U11, logical OR The two input signals of the gate U11 are the forward speech signal judgment 099129997 Form No. A0101 Page 21 / Total 44 Page 0992052617-0 201212008 Module (p 9 1, . 丄) The output signal 8 and the reverse speech signal judging module ( ) take the k number 4, when the forward speech signal judging module (2 2 • the round 6 and the reverse speech signal judging module) 2 2 2 ) When any of '4 is true, that is, when the forward speech signal is judged by the sign -f ^ c\ z 2 1 ) and the reverse speech signal judging module (2 2 2 ) When the individual 疋 audio 彳 § is a voice signal, after the logic or gate 1 1 processing, its output signal 10 is true, that is, after the logic or gate un processing, the audio signal is judged as a voice signal. Otherwise, the audio signal is determined. For noise signals. [0086] The seventh diagram illustrates the circuit principle of the RMS detection control circuit (23) and the RMS detection circuit (24) provided by the embodiment of the present invention. The rms detection control circuit (23) includes a logic OR gate U13 and a second selection switch S3. Wherein the two input ends of the logic OR gate U13 are respectively connected to the output end of the first logic processing module (2 2 3), and the other input end thereof and the second logic processing mode included in the noise canceling control signal generating unit (3) The output of the group (3 4) is connected, and the output signal of the logic or gate μ3 is used as the control of the selection switch S3. The image number is controlled by the selection switch S3 to control the effective value detection circuit (2 4 ) to the audio signal. The RMS detection 'generates an output signal for reflecting the average power of the audio signal' or does not perform rms detection on the audio signal, but directly maintains the output of the _ inherent initial value V1. The working principle of the circuit is as follows: The logic or gate U13 performs or processes the output 4 of the first logic processing module (2 2 3 ) and the output signal 19 of the first logic processing module (3 4 ) to obtain two Select a control signal 11 of the switch S3, at 099129997 Form No. A0101 Page 22 / Total 44 Page 0992052617-0 [0087] 201212008 [0088] ❹ [0089] Ο [0090] 099129997 Signal 11 controls two selection switches Switching between the output signal 7 of the signal preamplifier circuit and the preset natural initial value V1, the output signal 12 is obtained, by which the effective value detecting circuit (24) is controlled to the signal preamplifying circuit. The output signal 7 of (2丄) performs rms detection and generates an output signal DD' for reflecting the average power of the audio signal or does not perform rms detection on the audio signal, but directly maintains the output of an inherent initial value VI. The eighth figure shows the circuit principle of the effective voice signal comparison module (31) and the oversized voice signal module (32) provided by the embodiment of the present invention. The effective speech signal comparison module (3 1 ) includes a 丨 comparator ui5 and a Schmitt inverter (U17, U18). The comparator U15 sets the output signal DD of the rms detection circuit (2 4 ) with a preset number The four thresholds are compared to extract the control signal for the effective speech signal and output a control signal 14 for the active speech. The fourth threshold mainly detects an effective speech signal, that is, sets a noise threshold. The setting has a body that depends on the sensitivity of the microphone, and the sensitivity is determined by the collection of different sound pressure noise. The specific setting work should be combined with acoustic theory and experimental statistics. When the comparison is made, when the signal DD is greater than the fourth threshold, the output of the active speech signal comparison module (31) is logically true, representing the speech signal. When less than the fourth threshold, the output logic of the effective speech signal comparison module (31) is false, representing a noise signal. The oversized speech signal comparison module (32) includes a comparator U16 and a Schmitt inverter (U19, U20) 'the comparator U16 sets the output signal DD of the rms detection circuit (24) to a preset fifth The threshold is compared with the form nickname A0101, page 23/total 44 page 0992052617-0 201212008, for the extraction of the control signal for the oversized signal, and the control signal 15 of the oversized signal is output. In the process of communication, it is inevitable that there will be wind noise in the communication environment, and the sudden large noise will give the receiver (listener) a harsh sound, in this case, through the oversized speech signal comparison module (3 2 The control signal is extracted for the super-large signal, so that the noise-canceling amplifying unit (4) cancels the over-large signal and improves the hearing comfort of the listener. [0094] The ninth diagram shows the circuit principle of the control signal compensation module (33) provided by the embodiment of the present invention. The control signal compensation module (3 3 ) includes a multivibrator U21 and or a gate U22. The multivibrator U21 detects the falling edge of the output pulse of the effective speech signal comparison module (3 1 ). Once the falling edge is detected, the multivibrator outputs a pulse signal of a specific width, which can be valid after passing through the gate U22. The output pulse of the speech signal comparison module (3 1 ) compensates for the unique pulse width, and a control signal 18 for compensating the processed effective speech is obtained. In this way, at the end of the voice, the loss of voice can be reduced, and the voice can be restored with high quality. The tenth figure shows the circuit principle of the second logic processing module (34) provided by the embodiment of the present invention. The second logic processing module (34) is composed of a logical exclusive OR gate U23. The output signal 16 of the control signal compensation module (3 3 ) and the output signal 13 of the super large speech signal comparison module (32) are respectively sent to the input end of the second logic processing module (34), and the two signals are different. Or logic forms a denoising control signal 17, which is used to control the denoising amplifier 099129997 Form No. A0101 Page 24 / Total 44 Page 0992052617-0 201212008 [0095] [0096] Ο Ο [0097] = yin The audio α processed by the financial control attenuation unit (6) is subjected to a chewing process. ^ J ^ The circuit principle of generating the early element (5) of the attenuation control signal provided by the embodiment of the present invention. = attenuation control information wealth unit (1) is composed of calf 2, the parent-U12's towel-input signal is the effective value detection circuit (2 round-out_reflecting the audio signal_average power of the round-out signal 卯iKt number is The preset sixth threshold 'comparator compares the output with the preset sixth value to generate an attenuation control signal 18. The attenuation (four) is _ (four) tone-controlled attenuation unit (6) for the audio collection unit (1) The "audio domain performs different scales of inverting amplification. In the embodiment of the present invention, when the output signal for reflecting the average power of the audio signal is greater than the preset sixth signal, the generated suffocation control signal 18, Controlling the audio controllable attenuation unit (6) performs equal-amplitude inversion amplification on the audio signal, that is, does not attenuate the amplitude of the audio signal; when the average power output signal for the anti-audio signal is less than or equal to the preset The sixth®' value, the resulting attenuation control signal 18 controls the audio controllable attenuation unit (6) to attenuate the excessively large audio signal to a suitable proportion to reduce the excessively large audio signal. Figure 12 shows This hair The circuit control principle of the audio controllable attenuation unit (6) provided by the embodiment. The audio controllable attenuation unit (6) comprises an operational amplifier U2, a control switch S1, resistors R3, R4, R5, wherein the resistor R3 and the resistor R4 are connected in series Connected to the input of the op amp U2, the control switch S1 is connected in parallel at both ends of the resistor R3. 099129997 Form 编 St A0101 Page 25 / Total 44 Page 0992052617-0 [0098] 201212008 'Resistor R5 is connected across the input of the op amp U2 And the output terminal. The control signal of the control switch SI is derived from the attenuation control signal generated by the attenuation control signal generating unit (5). The attenuation control signal 10 controls the conduction and the opening of the control switch to implement the output of the audio collection unit (1). The audio signal 1 is subjected to different ratios of inverting amplification to obtain the inversely amplified signal 2. In the embodiment of the invention, the attenuation control signal 18 controls the audio controllable attenuation unit when the volume of the audio signal is in the normal range (6) The equal-amplitude inversion amplification of the audio signal, that is, the amplitude of the audio signal is not processed 'to obtain the inverted-amplified signal 2; when the volume of the audio signal exceeds the normal range During the encirclement, the fading control signal 18 controls the audio controllable attenuation unit (6) to attenuate a certain proportion of the excessively large audio signal, thereby appropriately reducing the excessively large audio signal to obtain the inverted, amplified signal 2 [0099 [0100] 099129997 The twelfth figure shows the circuit principle of the controllable inverter (4 1 ) provided by the embodiment of the present invention. The two-in-one wheel of the controllable inverter (4 1 ) is respectively connected with the audio The controllable mitigation unit (6) and the output end of the second logic processing module (34) are connected. The controllable inverter (4 1 ) includes an operational amplifier U3, resistors R6, R7, and an optional switch S2. The resistor R6 is connected to one of the input terminals of the operational amplifier U3. The resistor R7 is connected across the wheel input end and the output end of the operational amplifier U3. The second selection switch S2 is controlled by the noise canceling control signal 13 to reverse the phase. The amplified signal 1 is processed in phase or inversion to obtain a signal 3. When the in-phase processing is performed, the signal 2 after the inverting amplification is enhanced to obtain the signal 3; when the inversion processing is performed, the noise is eliminated to obtain the signal 3. Form nickname A0101 Page 26/44 page 0992052617-0 201212008 [0101] The fourteenth and fifteenth figures show the first phase follower (42) and the second reverse follower provided by the embodiment of the present invention. Circuit principle α of (4 3) [0102] wherein the input end of the first inverting follower (42) is connected to the round-out end of the audio controllable attenuation unit (6), and the output end thereof and the inverting adder ( 4 1) The input is connected. The first inverting follower (42) includes an operational amplifier U4, a resistor R8 coupled to one of the inputs of the operational amplifier U4, and a resistor R9 coupled across the intermset and output of the operational amplifier U4. The input chirp signal is the output signal 1 of the audio controllable attenuation unit (6), and the output signal 4; the input end of the second inverting follower (43) is connected to the controllable inverter (4 1), and the output end thereof Connected to the inverting adder. The second inverting follower (43) includes an operational amplifier 115. A resistor R1〇 connected to one of the inputs of the operational amplifier U5, and a resistor R11 connected across the input and output terminals of the operational amplifier U5. Its input signal is the output signal 3 of the inverting adder (4 1 ), and the output signal 5 is output. ^ [0103] 〇 The sixteenth figure shows the invention. The circuit principle of the inverting adder (4 4) provided by the embodiment. μ. . . . . [0104] The two inputs of the inverting adder (4 4) are respectively connected to the outputs of the first inverting follower (42) and the second inverting follower (43), which comprise an operational amplifier U6. A resistor R12 connected to one of the input terminals of the operational amplifier U6 and a resistor R13 connected to the other input terminal of the operational amplifier U6 are connected across the input end of the resistor R1 3 of the operational amplifier U6 and the output end of the operational amplifier U6. Resistor R14. The two input signals of the inverting adder (4 4) are the input signal 4 of the first inverting follower (4 2 ), the second inverting follower 099129997, the form number A0101, page 27 / total 44 pages 0992052617- 0 201212008 (4 3 ) Output signal 5, output signal 4. [0105] FIG. 17 shows a circuit principle of a low pass filter (45) provided by an embodiment of the present invention. [0106] The input of the low pass filter is connected to the output of the inverting adder (4 4 ). It includes a first resistor R15 and a capacitor C2. The low pass filter (4 4) filters out additional out-of-band noise in the device, preserving the frequency components within the speech band. [0107] In the embodiment of the present invention, an output signal for reflecting the average power of the audio signal is obtained by performing RMS detection on the received audio signal, and the output signal for reflecting the average power of the audio signal is A threshold is compared to generate a noise canceling control signal, and the noise signal in the audio signal is cancelled under the control of the noise canceling control signal, and the voice signal is amplified, thereby eliminating the background noise of the voice. At the same time, by comparing the output signal for reflecting the average power of the audio signal with the sixth threshold, an attenuation control signal is generated, and the received audio signal is inversely amplified in different proportions under the control of the attenuation control signal to appropriately reduce Excessive audio signal improves the hearing comfort of the receiver. [0108] In summary, the creation can achieve the above functions and purposes, so the creation should meet the requirements of the patent application, and apply in accordance with the law. BRIEF DESCRIPTION OF THE DRAWINGS [0109] The first figure is a flowchart for implementing a method for canceling speech background noise according to an embodiment of the present invention; [0110] The second figure is a speech background noise canceling apparatus provided by an embodiment of the present invention 099129997 Form No. A0101 Page 28 / Total 44 Page 0992052617-0 201212008 Block Diagram; [0111] The third figure is a circuit schematic diagram of the audio collection unit provided by the embodiment of the present invention; [0112] The fourth figure is the present invention The circuit schematic diagram of the signal pre-amplification circuit provided by the embodiment; [0113] The fifth figure is a circuit schematic diagram of the forward speech signal judging module and the reverse speech signal judging module provided by the embodiment of the invention; [0114] The circuit diagram of the first logic processing module provided by the embodiment of the present invention; [0115] The seventh diagram is a circuit schematic diagram of the RMS detection control circuit and the RMS detection circuit provided by the embodiment of the present invention; [0116] The eighth figure is a circuit schematic diagram of the effective voice signal comparison module and the oversized voice signal comparison module provided by the embodiment of the present invention; [0117] The ninth figure is a control letter provided by the embodiment of the present invention. The circuit diagram of the compensation module is shown in FIG. 10 is a circuit schematic diagram of the second logic processing module provided by the embodiment of the present invention; [0119] The eleventh figure is the attenuation control signal generated by the embodiment of the present invention. The circuit schematic diagram of the unit; [0120] The twelfth figure is a circuit schematic diagram of the audio controllable attenuation unit provided by the embodiment of the present invention; [0121] The thirteenth figure is a controllable inverter provided by the embodiment of the present invention Circuit Principles 099129997 Form No. A0101 Page 29/44 Page 0992052617-0 201212008 Figure; [0122] The fourteenth and fifteenth drawings are the first inverting follower and the second inverting follower provided by the embodiment of the present invention. FIG. 16 is a circuit schematic diagram of an inverting adder according to an embodiment of the present invention; [0124] FIG. 17 is a circuit diagram of a low pass filter provided by an embodiment of the present invention; Figure. [Key component symbol description] [0125] (1) Input terminal and audio acquisition unit [0126] (2) RMS detection unit [0127] (2 1 ) Signal preamplifier circuit [0128] (2 2 ) Voice signal judgment circuit [2129] (2 2 1 ) Forward Speech Signal Judgment Module [0130] (2 2 2 ) Reverse Speech Signal Judgment Module [0131] (2 2 3) First Logical Processing Module [0132] (2 3 RMS detection control circuit [0133] ( 2 4 ) rms detection circuit [0134] (3) annihilation control signal generation unit [0135] (3 1 ) effective speech signal comparison module [0136] (3 2 ) oversized Voice signal comparison module [0137] (3 3 ) Control signal compensation module 0992052617-0 099129997 Form number A0101 Page 30 / Total 44 page 201212008 [0138] [0140] [0141] [0143] [0146] [0146]
(3 4 )第二邏輯處理模組 (4) 消噪放大單元 (4 1 )可控反相器 (4 2)第一反相跟隨器 (4 3)第二反相跟隨器 (4 4)反相加法器 (4 5)低通遽波器 (5) 衰減控制信號產生單元 (6) 音頻可控衰減單元(3 4) second logic processing module (4) noise canceling amplifying unit (4 1 ) controllable inverter (42) first inverting follower (43) second inverting follower (4 4) Inverting adder (4 5) low pass chopper (5) attenuation control signal generating unit (6) audio controllable attenuation unit
099129997 表單編號A0101 第31頁/共44頁 0992052617-0099129997 Form No. A0101 Page 31 of 44 0992052617-0