200912894 九、發明說明： 【發明所屬之技術領域】 本發明涉及一種發聲設備測試裝置及其測試 方法，特別涉及一種對發聲設備各項參數進行測試 時減小噪音影響的發聲設備測試裝置及測試方法。 【先前技術】 目前，對發聲設備如喇U八、音箱等音質進行測 試的參數主要有訊號/雜訊比測試、動態範圍測 試、頻率響應測試、總諧波失真測試以及聲道分離 測試等。目前的測試方法大致分為兩種：一種是人 ，檢測，主要依靠作業員耳朵聽發聲設備所發出的 聲音來判別發聲設備的好壞，這種方法雖然簡單， 但受作業員主觀影響大，而且沒有考慮環境以及發 f設備㈣音影響’結果不準確，檢測品質低； 時作業員的聽力也會受到損害。另—種是採用專用 曰頻:則立里儀器如Audi0Precisi0n (Ap)公司的奶7⑻ 系列等音頻分析儀進行測試，該音頻測試儀器測試結 / 4準確但測試儀器的價格昂貴，而且測試時， 須配合AP公司的標準麥克風和校準器，使之成為 一套專業的電聲測試系統，更增加了用戶的成本: 曰力了用戶的不方便，當麥克風和校準器損壞 用戶還而去購買專門的標準麥克風和校準器。 【發明内容】 200912894 :鑒於此’提供一種能減小噪音影響的發聲設 頻二置及測試方法’對待測發聲設備發出的音 ==先進行消噪處理再進行各項參數的測試，從 乂小的成本提高了對發聲設備參數測 確性，以解決現有技術中存在的問題。 早 :減小噪音影響的發聲設備測試裝一 ':採集裝置、一音頻處理裝置、—記憶體 ::Γ其中：聲音採集裝置用於採集-發聲設： 將料4=::采::號？音頻處理裝置用於 傅立葉變換模組、—運算模組以及 換成時域數位音頻信號轉 組、-傅…3處理早70包括—錄入模 測試模 其中’該錄入模組用於將咅考 時域數位音頻作泸砝,、、^衣置轉換的 組用於調用::=心憶體中，傅立葉變換模 立：；；Γ 音頻信號，對該預設長度的數位 雄中：弟—次快速傅立葉變換，記錄所得頻 -中母個頻率的頻率幅值於記憶體中，再 時域數位音頻信號並截取兩倍預:長： 號進行，對該兩倍預設長度的數位音頻信 ===傅立葉變換，記錄第二次傅立葉 每 :u—次傅立葉變換所得頻譜中 子目同頻率的前—Jig φ ΑΛ 月】頻率的頻率幅值即每一頻率 200912894 的噪音成分幅值於記憶體巾。該運算模 中記錄的第一次傅立葉變換的每個頻率φ/己憶體 第二次傅立葉變換後所記錄的第—次傅田值減去 的頻率幅值對應的嗓音成分幅值::= —次傅立葉變換後的頻率相 ^ 第 Μ葉變換模組還對該運算模組 k號進行反傅立葉變換，得到一 \:曰’員 測試模組對經過反傅立葉變換後的時曰:二:虎淮該 相關參數測試。 $戍4唬進订 下牛音影響的發聲設備測試方法包括以 味广^聲没備所輸出之時域類比音頻信 存“儲頻信號轉換成時域數位音頻信號； =:數位音頻信號;對所存儲的時域數位音 取—預設長度的時域數位音頻信號，㈣ 苹辦^長度數位音頻信號進行第—次快速傅立 己錄所得晴每個頻率的頻率幅值；對 斤存：的時域數位音頻信號截取兩倍預設長度的 =㈣號，對該兩倍預設長度的數位音頻信號 施丁快逮傅立葉變換;記錄第二次傅立葉變 、所仔頻譜中與第—次傅立葉變換所得頻譜令、一 立目同頻率的前一頻率的頻率幅值即為該頻率的啐 =分幅值；用第-次傅立葉變換的每個頻率幅值 也π第二次傅立葉變換後所記錄的噪音成分幅 ’得到-與第—次傳立葉變換後的頻率相同的頻 200912894 域信號；對該_音頻錢進行反傅立葉變換，得 =一 ^域音齡號；對經過㈣立輕換後的時域 曰頻彳§號進行相關參數測試。 通過本發明的發聲 ▼…今叹拥閃碼衷置和戈 >、’用較小的成本獲得了發聲設備各項參數較準 的測試結果。 【實施方式】 ，為減小噪音影響的發聲設備測試 糸，，先的木構圖。該發聲設備測試系統包括—讲 備測試裝置1及—第—發聲設備2 a及/或—^二ς 聲設備2b。該第—發聲設備2a可為具有心八^ 播放自身記憶體中的音頻槽的設備200912894 IX. The invention relates to a sounding device testing device and a testing method thereof, and particularly to a sounding device testing device and a testing method for reducing noise impact when testing various parameters of a sounding device . [Prior Art] At present, the parameters for testing the sound quality of sounding equipment such as RaU-8, speakers, etc. mainly include signal/noise ratio test, dynamic range test, frequency response test, total harmonic distortion test and channel separation test. The current test methods are roughly divided into two types: one is human, and the test mainly relies on the sound of the sounding device of the operator's ear to discriminate the sounding device. Although this method is simple, it is subjectively influenced by the operator. Moreover, the environmental impact and the influence of the equipment (four) sounds are not considered, and the results are inaccurate and the quality of the test is low; the hearing of the operator is also impaired. The other is a special frequency: the Lili instrument is tested by an audio analyzer such as Audi0 Precisi0n (Ap)'s milk 7 (8) series. The audio test instrument is accurate / but the test instrument is expensive, and when tested, It must cooperate with AP's standard microphone and calibrator to make it a professional electroacoustic test system, which increases the cost of the user: It is inconvenient for the user, when the microphone and calibrator are damaged, the user also goes to buy special Standard microphone and calibrator. [Description of the Invention] 200912894: In view of this, "provide a sound-frequency setting and test method that can reduce the influence of noise." The sound emitted by the sound-emitting device to be tested == De-noise processing is performed first, and then the parameters are tested. The small cost improves the accuracy of the sounding device parameters to solve the problems in the prior art. Early: Reduce the noise impact of the sound equipment test installation ': collection device, an audio processing device, - memory:: Γ where: sound collection device for acquisition - sound settings: material 4 =:: mining:: ? The audio processing device is used for the Fourier transform module, the operation module, and the time domain digital audio signal grouping, the -fu...3 processing early 70 includes - the input mode test mode, wherein the input module is used for the test time The field digital audio is used to call, and the set of clothing conversion is used to call::= heart recall body, Fourier transform mode:;; 音频 audio signal, the number of the preset length of the male: brother - times Fast Fourier transform, recording the frequency amplitude of the obtained frequency-medium frequency in the memory, and then the digital audio signal in the time domain and intercepting twice the pre-: long: number, the digital audio signal of the preset length twice == Fourier transform, record the second Fourier per: u-secondary Fourier transform spectrum in the spectrum of the sub-frequency of the same frequency - Jig φ ΑΛ month frequency frequency amplitude, that is, the frequency component amplitude of each frequency 200912894 in the memory towel. The amplitude of the arpeggio component corresponding to the frequency amplitude subtracted by the first-order Four-field value recorded after the second Fourier transform of the first Fourier transform recorded in the operation mode::= - The frequency phase after the Fourier transform ^ The third leaf transform module also performs an inverse Fourier transform on the k number of the operation module to obtain a time after the inverse Fourier transform of the \:曰's test module: Huhuai related parameter test. $戍4唬 The sounding device test method that influences the influence of the cow sound includes the time domain analog audio signal outputted by the wide-ranging sound sound. “The frequency-locked signal is converted into a time-domain digital audio signal; =: the digital audio signal; For the stored time domain digital sounds, the preset length of the time domain digital audio signal, (4) the length of the digital audio signal for the first time, the fast frequency of the frequency is obtained, and the frequency amplitude of each frequency is obtained; The time domain digital audio signal intercepts twice the preset length of the = (four) number, and applies the fast Fourier transform to the digital audio signal of twice the preset length; records the second Fourier transform, the first spectrum and the first Fourier Transforming the obtained spectrum, the frequency amplitude of the previous frequency of the same frequency is the 啐=segment value of the frequency; the amplitude of each frequency using the first-time Fourier transform is also π after the second Fourier transform The recorded noise component amplitude 'gets-the frequency of the frequency of the 200912894 domain after the first-time sub-leaf transformation; the inverse Fourier transform of the _ audio money, the ==^ domain age number; Time domain曰 彳 彳 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行 进行Mode], the sounding equipment test to reduce the noise impact, the first wood composition. The sounding equipment test system includes - the test equipment 1 and - the first sounding equipment 2 a and / or - 2 two sound equipment 2b The first sounding device 2a may be a device having a heartbox to play an audio slot in its own memory
器、手機等手持設備，該第二發聲設備加為^ 被動播放其他電子裝置中的音頻文件的設備心 相等。在對第二發聲設備2b進行測試時，該第二 備2b通過一音頻連接線連接至發聲設備測 =衣置ί或者其他電子裝置上。該發聲設備測試裝 置1包括一處理單元10、—音頻採集裝置加、一 曰頻處理裝置30及一記憶體4〇。 其中，該音頻採集裝置2〇用於採集第一發聲 設備2a或者第二發聲設備2b所發出的時域類：音 頻信號。在本實施方式中該音頻採集裴置扣採^ ”筒，由於這種超心型話筒對偏離方向的聲 音大量衰減，故應將該超心型話筒正對且儘量靠近 11 200912894 發聲設備’但是為了避免接觸產生新的噪音，話筒 不能接觸發聲設備。該音頻處理裝置3〇為普通音 效卡，用於將音頻採集裝置2〇採集到的時域類比 音頻信號轉換成時域數位音頻信號。本實施方式 中，通過一特定測試音頻文件對發聲設備進行測 試，當所測試的發聲設備為第—發聲設備2a時， 則測試音敎件存儲於第—發聲設備2a的記憶體 中、，當所測試的發聲設備為第二發聲設備汕時， 則測試音頻文件存儲於發聲設備測試裝置丄的記 憶體40中，該第一發聲設備“以及記憶體4〇中 的音頻槽可為任意頻率音頻信號。為了提高測試準 確性，本實施方式中，該測試音頻槽為1G00赫兹 / 1KHZ )頻率的標準參考音頻信號。同時第一發 ^設備及/或第二發聲設備所輸出的音頻信號的音 里應大於90分貝’這樣可減小環境噪音的幹擾。 處理單元10包括-錄入模組101、一播放模 組102、一傅立葉變換模組1〇3、一運算模組⑽ =及—測試模組105。其中錄入模組1〇1，用於將 =頻處理裝置轉換的時域數位音頻信號儲存於記 =體扣中。播放模組1〇2用於在所測試的發聲設 為第二發聲設備2b，且第二發聲設備处連接至 〜測試設備1時，調用儲存於記憶體40中的音頻 ^ ’並將該音頻文件傳輸至音頻處理裝置％轉 、為類比音頻信號後通過第二發聲設備孔輸出。 200912894 :傅：葉變換模組103調用存儲在記憶體 f立曰頻信號對其分別進行第-次及第二次快迷 ^ 立葉變換（FFT’FastF〇urierTransfGrm)，具體 、交換過程請參閱圖2及圖3。 :::圖2’為傅立葉變換模組對音頻信號進 丁 A、速傅立葉變換的頻譜圖。傅立葉變換模 、二 103調用儲存在記憶體仙中的時域數位音頻V 號後’I先對該數位音頻信號進行第_次FF丁^ 換’即截取-預設長度的時域數位音頻信號，為了 2不是整週期截取數位音頻信號所產生的截斷 效應’即減小頻譜㈣，可對該預設長度的時域數 位音頻信號加窗處理後進行FFT變換，得到圖2 ，不頻譜圖。該窗可為漢明（Hamming)窗或漢宜 (Hanning)窗等窗函數。 圖2所示頻譜圖橫坐標每一個點為逐漸均 增大的頻率值，‘縱坐標為每個頻率值的幅值（db)。 板坐標每個點㈣率可由下述傅立葉公式i確定. (〇A handheld device such as a mobile phone or the like, and the second sounding device is added to the same as the device for passively playing audio files in other electronic devices. When the second sounding device 2b is tested, the second device 2b is connected to the sounding device via an audio connection cable or other electronic device. The sounding device testing device 1 includes a processing unit 10, an audio collecting device plus, a frequency processing device 30, and a memory device. The audio collecting device 2 is configured to collect a time domain class: an audio signal sent by the first sounding device 2a or the second sounding device 2b. In the present embodiment, the audio collection device is labeled with a "tube". Since the supercardioid microphone attenuates a large amount of off-axis sound, the supercardioid microphone should be facing right and as close as possible to the 11 200912894 sounding device' In order to avoid contact to generate new noise, the microphone cannot contact the sounding device. The audio processing device 3 is a normal sound card for converting the time domain analog audio signal collected by the audio collecting device 2 into a time domain digital audio signal. In an embodiment, the sounding device is tested by a specific test audio file. When the tested sounding device is the first sounding device 2a, the test sound component is stored in the memory of the first sounding device 2a. When the tested sounding device is the second sounding device, the test audio file is stored in the memory 40 of the sounding device testing device, and the first sounding device "and the audio slot in the memory 4" can be any frequency audio signal. . In order to improve the accuracy of the test, in the present embodiment, the test audio slot is a standard reference audio signal of a frequency of 1 G00 Hz / 1 KHz. At the same time, the sound of the audio signal output by the first device and/or the second sounding device should be greater than 90 decibels to reduce the interference of environmental noise. The processing unit 10 includes a recording module 101, a playing module 102, a Fourier transform module 1-3, a computing module (10) = and a test module 105. The input module 101 is used to store the time domain digital audio signal converted by the frequency processing device in the note body buckle. The play module 1 用于 2 is configured to call the audio stored in the memory 40 and transmit the audio when the tested utterance is set to the second utterance device 2b and the second utterance device is connected to the test device 1. The file is transmitted to the audio processing device, and is converted to an analog audio signal and then output through the second sounding device hole. 200912894: Fu: The leaf transform module 103 calls the first and second fast transforms (FFT'FastF〇urierTransfGrm) stored in the memory of the memory, and the details of the exchange process are shown in the figure. 2 and Figure 3. ::: Figure 2' is the spectrum of the Fourier transform module for the audio signal A and the fast Fourier transform. Fourier transform mode, two 103 call the time domain digital audio V number stored in the memory fairy, 'I first perform the first FF FF changing on the digital audio signal', that is, intercepting - the preset length of the time domain digital audio signal In order to eliminate the truncation effect caused by the interception of the digital audio signal by the whole period, that is, to reduce the frequency spectrum (4), the time-domain digital audio signal of the preset length may be windowed and subjected to FFT transformation to obtain FIG. 2 and no spectrogram. The window can be a window function such as a Hamming window or a Hanning window. Each point of the abscissa of the spectrogram shown in Fig. 2 is a gradually increasing frequency value, and the ordinate is the amplitude (db) of each frequency value. The point (four) rate of the plate coordinates can be determined by the following Fourier formula i.
N J 1 N * 2 〆 其中，i表示頻譜圖橫坐標第i個點，^為該點 的頻率’fs為採樣頻率，N為所截取的數位音頻传 號長度，由於G猶，可知共有N個縣示從咖 到fs/2的頻率。 在本實施方式中，採樣頻率fs設為96khz， 13 200912894 第人FFT邊換截取的數位音頻信號長度設為 4096個數位音頻信號點，即n=4_，由上述的傅 立葉公式1可知得到頻譜圖的頻率範圍在0ΗΖ到 48KHZ之間，有侧個點均勻地表*這些頻率印 有4096個頻率，記錄每個頻率對應的幅值即頻率 幅值圮於記憶體40中。 。、二二閱圖3,為傅立葉變換模組1〇3對音頻信 唬進打第二次FFT變換所得到的頻譜圖。在 人FFT交換中，傅立葉變換模組如再次調用記 :體：的時域數位音頻信號，然後截取長度為 弟一 變換職取的預設長度❼倍的時域 位曰齡號，對該兩倍預設長度的時域 信號加窗處理後進行第二次附變換，得到= 所不㈣圖。該頻譜圖的料範圍NJ 1 N * 2 〆 where i represents the i-th point of the abscissa of the spectrogram, ^ is the frequency of the point 'fs is the sampling frequency, and N is the length of the intercepted digit audio signal. Since G is still, it is known that there are N The frequency of the time from the coffee to the fs/2. In the present embodiment, the sampling frequency fs is set to 96khz, 13 200912894 The length of the digital audio signal intercepted by the first FFT side is set to 4096 digital audio signal points, that is, n=4_, and the spectrogram is obtained by the above-described Fourier formula 1. The frequency range is from 0ΗΖ to 48KHZ, with side points evenly surfaced. * These frequencies are printed with 4096 frequencies, and the amplitude corresponding to each frequency, that is, the frequency amplitude is recorded in the memory 40. . Figure 2 shows the spectrum of the second FFT transform of the audio signal into the Fourier transform module 1〇3. In the human FFT exchange, the Fourier transform module re-invokes the time domain digital audio signal of the body: and then intercepts the time domain digits of the preset length ❼ times the length of the change of the brother, the two The time-domain signal of the preset length is windowed and then subjected to the second sub-transformation to obtain a graph of = (four). Material range of the spectrogram
f , (〇 ⑽由傅立葉公式1可以推出傅立葉公式2:下 ^ 2f , (〇 (10) Fourier Formula 1 can be derived from Fourier Formula 2: Lower ^ 2
N ::第二次FFT變換的長度為第—次 的兩倍，則由傅立葉公式2可知， 4 、N: The length of the second FFT transform is twice the length of the first time, as shown by Fourier's formula 2, 4
變換後表示頻率fi的點為第；點， =FFTAfter the transformation, the point indicating the frequency fi is the first; point, =FFT
變換後表示頻率^的點將 ;mFT 人V木Z1點，gp力 一 :人FFT變換後，表示相同頻率值的點數為第7 FFT變換後的兩倍。即在本實施方式中人 次FFT變換中代表每個頻率 如果第一 3.....4096, 14 200912894 =*!、二:欠FFT變換後代表同一頻率的點則變為 M 2 2、3 ......第4096*2點等偶數點， ...........第8191等4096個奇數胃占 為新插入的點，即為第一次FFT變換的各頻率八 離出來的噪音成分，即第二次FFT尚、^_ '刀 點為第一次FFT變換的第n點八^、弟Μ 分（HM 4096 )，例如^ 2刀離出來的噪音成 的第1 _相玄’# 1點為弟—次FFT變換 ，弟1點的頻率分離出的噪音成分，第3點為第一 -人FFT變換的第2點的嚙至八 料頻率㈣^”音成 雜山 次附變換的第3點的頻率分 離出來嶋成分，...，第8191個點為第一 :換：二4096點的頻率分離出的噪音成分。記錄 可，點的幅值即噪音成分幅值⑽於記憶體 在得到消嗔前信號頻率幅值 ，後’運算模組⑽用頻率幅值心= °呆曰成分幅值Na，即，节篦Λ 點的幅值減去第_ Λ 變換的第1 第-變換的第1點的幅值， 變換的第Λ弟2個點的幅值減去第二次fft 個點的幅值等等。則可消去大部分噪 ‘甚去了噪音成分的有4096個點的頻域俨 反傅立葉變換得到一時域了的;;:= < 的時域音頻錢發送給賴模組⑽。 15 200912894 測试模組105根據消噪後的時域音頻進行發聲設 備的各項參數如總諧波失真的測試，由⑨各項灸數 的測=為現有技術，因此在本發明中不多加描述。 請參閱圖4，為減小噪音影響的發聲設備測試 方法流程圖。首先，發聲設備輪出一段時域類比音 ，信號（S10! ) ’·音頻採集裝置2 〇採集該時域類二 音頻信號並通過聲音處理裝置3〇將該時域類比音 頻仏號轉換成時域數位音頻信號（sl〇2);該時域 數位音頻信號通過錄入模、址1〇1被儲存在記憶體 4〇中（S1〇3);傅立葉變換模組1〇3調用記憶體4〇 :的％•域數位音頻信號，並截取—預設長度的數位 曰，信號’在對該預設長度的數位音頻信號加窗處 2進打第-次快速傅立葉㈣（_);記錄頻 ::中田值Fa ( S1G5) ’傅立葉變換模组1()3再次調取 =憶體40中的時域數位音頻信號，並截取2倍預 ，長度的時域數位音頻錢，對該2倍預設長度的 蚪域數位音頻信號加窗處理後進行第二次快 =變：_;記錄第二次快速傅立葉變換所 于頻禮中與第—次快速傅立葉變換所得頻譜 二前;頻率的頻率幅值即為該每個 鴻旱的本日成分幅值Na(sl()7);用第—次 到的頻率幅值Fa減去第二次快速傅立 = = 分幅值⑽，得到-消噪的頻 1頻k唬（Sl〇8);傅立葉變換模組ι〇3對消噪 16 200912894 的頻域音頻信號進行反傅立 。 域音頻信號然後傳送該消噪的時域：頻;4Γ寺 號對總諸波失真等各項參數上: 17 200912894 【圖式簡單說明】 圖1 =減小％音影響的發聲設備測試系統的架構圖。 立葉 疋傅立葉變換模組對音頻信號進行第—次快速傅 ’交換的頻譜圖。 立葉變換的頻譜圓 圓4是減小噪音影響的發聲設備測 【主要元件符號說明】 去的流程圖 處理單元 音頻採集裝置 10 音頻處理裝置 20 記憶體 30 手持設備 40 Jt. Λ-Ατ 曰相 2a 錄入模組 2b 播放模組 101 傅立葉變換模組 102 運算模組 103 測試模組 104 105 18After the transformation, the point indicating the frequency ^ will be; mFT person V wood Z1 point, gp force one: After the human FFT transformation, the number of points indicating the same frequency value is twice that after the 7th FFT transformation. That is, in the present embodiment, in the human FFT conversion, each frequency represents the first 3.....4096, 14 200912894 =*!, 2: the point representing the same frequency after the FFT conversion becomes M 2 2, 3 ...the 4096*2 point and other even points, ........... 819 odd-numbered stomach accounts for the newly inserted point, which is the frequency of the first FFT transformation Eight out of the noise component, that is, the second FFT is still, ^_ 'the tool point is the first FFT transform of the nth point 八 ^, Μ Μ ( (HM 4096), such as ^ 2 knife out of the noise into The first _ phase Xuan '# 1 point is the brother-secondary FFT transform, the noise component separated by the frequency of the 1st point, and the third point is the bite to the eighth frequency of the first-person FFT transform (four)^" The frequency of the third point of the sound-mixing mountain sub-transformation is separated into the 嶋 component, ..., the 8191th point is the first: the noise component separated by the frequency of two 4096 points. The record can be, the amplitude of the point That is, the amplitude of the noise component (10) is obtained by the amplitude of the signal frequency before the memory is obtained, and then the 'operational module (10) uses the frequency amplitude heart = ° the component amplitude Na, that is, the amplitude of the throttling point minus The first _ 变换 transformation of the first - The amplitude of the first point of the transformation, the amplitude of the 2 points of the transformed second brother minus the amplitude of the second fft point, etc., then the majority of the noise can be eliminated, and even 4096 of the noise components are removed. The frequency domain of the point is inversed by the Fourier transform to obtain a time domain;;: == The time domain audio money is sent to the Lai module (10). 15 200912894 The test module 105 performs the sounding device according to the denoised time domain audio. The test of various parameters such as total harmonic distortion is determined by the test of 9 moxibustion numbers, so it is not described in the present invention. Please refer to Figure 4, the flow chart of the test method for sounding equipment to reduce the influence of noise. First, the sounding device rotates a time domain analog sound, and the signal (S10!) '·the audio collecting device 2 collects the time domain type second audio signal and converts the time domain analog audio number into a sound processing device 3 Time domain digital audio signal (sl〇2); the time domain digital audio signal is stored in the memory 4〇 through the input mode, the address 1〇1 (S1〇3); the Fourier transform module 1〇3 calls the memory 4 〇: %•domain digital audio signal, and intercepted—the number of preset lengths曰, the signal 'in the window of the preset length of the digital audio signal 2 into the first fast Fourier (four) (_); recording frequency:: Nakata value Fa (S1G5) 'Fourier transform module 1 () 3 again Retrieve the time domain digital audio signal in the memory 40, and intercept the time-domain digital audio money of 2 times the pre-length, and then process the second-time preset digital audio signal to the second time. = variable: _; record the second fast Fourier transform in the frequency and the first fast Fourier transform obtained spectrum two; the frequency amplitude of the frequency is the daily component amplitude of each of the drought Na (sl ( 7); subtract the second fast Fourier = = framing value (10) from the first-order frequency amplitude Fa, and obtain the --noise frequency 1 frequency k 唬 (Sl 〇 8); Fourier transform module ι 〇3 inversely modulates the frequency domain audio signal of the denoising 16 200912894. The domain audio signal then transmits the time domain of the noise cancellation: frequency; 4Γ Temple number to the total wave distortion and other parameters: 17 200912894 [Simple diagram of the diagram] Figure 1 = The sound equipment test system that reduces the influence of the % sound Architecture diagram. The Fourier transform Fourier transform module performs a first-time fast Four's exchange of the audio signal. The spectral circle 4 of the Fourier transform is a sounding device that reduces the influence of noise. [Main component symbol description] Flowchart processing unit Audio acquisition device 10 Audio processing device 20 Memory 30 Handheld device 40 Jt. Λ-Ατ 曰相2a Input module 2b Playback module 101 Fourier transform module 102 Operation module 103 Test module 104 105 18