201042985 ':六、發明說明： 【發明所屬之技術領域】 特別是指其噪音消 本發明係涉及一種語音通訊裝置 除結構及方法之創新設計者。 【先前技術】201042985 ': Sixth, invention description: [Technical field to which the invention pertains] In particular, the invention relates to a noise communication device. The invention relates to an innovative designer of a voice communication device in addition to the structure and method. [Prior Art]
按，最常見的傳統噪音 音材料，以被動方式吸收噪 狀態時，聲音的波長遠大於 率為100Hz的聲音，在常溫、 ，因此，除非使用非常厚重 聲音的傳遞，而對於一般的 音的技術，對於低頻噪音的 實務應用上，被動式噪音消 抑制技術’多半是利用各種吸 音的能量來達成。然而在低頻 ~~般吸音材料的厚度，例如頻 海平面上的波長約為3~4公尺 的吸音器材才可有效阻絕低頻 應用而言，這種被動式消除噪 隔絕能力並不理想 '所以，在 除法往往必需在低頻噪音的有 效消除與吸音器材可能極佔空間、體積龐大及價格昂貴的 Ο 矛盾問題中取得平衡。 為了此夠有效克服如述問題，一種使用外加聲源以消 除噪音之主動式°喿音控制（active noise control; ANC)理論，多 年來，已經引起廣泛且深入的研究與發展。主動式α喿音控 制系統的基本原理，主要的是利用波形合成原理，在噪音 源附近產生一個與原噪音波形相同、大小相等、相位差18 0度的反噪音聲波，使其對嗓音源產生破壞性干涉而達到 噪音音量降低之目的。如第1圖所示，係為嚼音（n〇ise)、皮 形（L1 )、反噪音（anti-noise)波形（L2 )、以及°喿音與反 3 201042985 -:α呆音波形疊加後的殘餘D柴音（residual noise)波形（L3 此可知，主動式噪音消除技術的成敗，將取決於反 形、大小與相位的估測與控制的正確性。 ' 實務上’我們希望最終傳遞到人耳耳鼓中的空 不再含有噪音成份或至少噪音成份可被壓抑到極限 以主動式嗓音控制技術而言’這個目標可由兩種方 成，第一種方式是在噪音的來源處裝設反制噪音（ 訊號波形相同、大小相等、相位相反）之反噪音源 望°喿音在源頭就能被抵消而無法向外擴散，然而， 〇些特別的設施如汽車排氣尾管、中央空調系統之冷 官，其聲音傳播的主要方向是與管路平行之外，在 活中所感受的噪音’其聲波通常是向四面八方傳播 欲消除每一個方向的噪音訊號，必須考慮到音源及 徑上所有相關的三維聲場以及其複雜度；在多重噪 傳播路徑可能非常複雜的環境中，這種方法使得主 音控制系統過於龐大且複雜，因而不符經濟效益。 方式則是放棄製造產生一個大的無噪音空間，而僅 〇州在耳鼓外之噪音訊號，並據以在耳鼓附近產生一 波形相同、大小相等、相位相反之噪音訊號，以在 2產生一個很小的無噪音空間。而這種技術由於所 :2波形估測與反噪音波形之產生與控制都來得相 午夕，因此市面上已有多種降噪耳機以配戴耳機的 2用耳機内的揚聲器輸出反臂音波形來降低耳鼓附 曰 〇 次查’帛來產生反相噪音的控制系統主要分為 /、適應性兩種。其中固定參數的控制，是將 本曰訊號直接經過反相電路後產生反相噪音，然而 )。由' 噪音波 氣波動 狀態， 式來達 與噪音 ，並期 除了某 氣輸送 曰常生 的，若 傳播路 音源及 動式噪 第二種 止於估 個與其 耳鼓附 牽涉的 對簡單 方式， 近的嗓 固定參 輸入的 這種方 4 201042985 •.法所產生的反相噪音會因為反相電路的不理想之大小與相 位響應，使得噪音無法被完全消除，為了能夠補償控制系 統本身的相位延遲並能夠因應時變的噪音，必需採用能约 -自動調整參數的適應性控制系統’目前主要應用於耳機带 置的適應性主動式噪音控制系統是屬於回授式，其作法是 在耳機内部配置一個微型麥克風來取得耳内噪音，經由適 應性控制系統估測出反噪音後經由耳機内的揚聲器輸出， 藉以達到嗓音消除之目的。然而，該回授式主動噪音控制 糸統的主要限制在於當揚聲器輸出反脅音後，耳内d桑音與 Ο反嗶音結合後，耳機内部的麥克風取得的是殘餘噪音，而 適應！生控制系統所需要的噪音參考訊號並無法直接取得， 只能使用合成的技術來產生。當所合成的噪音參考訊號不 準確日寸，會使得所估測出的反噪音亦隨之不準確，因而大 幅降低噪音消除的性能。 疋以，針對上述習知噪音抑制技術設計上所存在之問 題點，如何研發出一種嘴音消除效果更佳、更具理想實用 a 土 ’斤°又汁，實有待相關業界再加以思索突破之目標及According to the most common traditional noise sound material, when the noise state is passively absorbed, the wavelength of the sound is much larger than the sound of 100Hz, at room temperature, therefore, unless the transmission of very heavy sound is used, and for the general sound technology For the practical application of low-frequency noise, passive noise suppression technology is mostly achieved by using various sound-absorbing energy. However, in the low-frequency sound-absorbing material thickness, for example, a sound-absorbing device with a wavelength of about 3 to 4 meters on a frequency sea level can effectively block low-frequency applications, this passive noise-cancellation capability is not ideal. In division, it is often necessary to strike a balance between the effective elimination of low-frequency noise and the contradiction that sound-absorbing equipment may be extremely space-consuming, bulky, and expensive. In order to effectively overcome the problems described above, an active noise control (ANC) theory using an external sound source to eliminate noise has been widely and intensively researched and developed over the years. The basic principle of the active alpha-sound control system is to use the principle of waveform synthesis to generate an anti-noise sound wave with the same magnitude, equal size and phase difference of 180 degrees near the noise source, so that it produces the sound source. Destructive interference to achieve the purpose of reducing the noise level. As shown in Fig. 1, it is a combination of chewing sound (n〇ise), skin shape (L1), anti-noise waveform (L2), and °喿 and anti 3 201042985 -:α dull waveform. Residual D residual noise waveform (L3) It can be seen that the success or failure of active noise cancellation technology will depend on the correctness of the estimation, control and control of the inverse shape, size and phase. 'Practical' we hope to finally pass The air in the eardrums no longer contains noise components or at least the noise components can be suppressed to the limit. In the case of active voice control technology, this goal can be achieved in two ways. The first way is to install at the source of the noise. The anti-noise source of counter-noise (the same signal waveform, equal size, opposite phase) can be offset at the source and cannot be spread out. However, these special facilities such as car exhaust pipe, central air conditioner The cold official of the system, the main direction of sound transmission is parallel to the pipeline, the noise that is felt in the life. The sound wave is usually transmitted in all directions to eliminate the noise signal in each direction. The sound source must be considered. All relevant three-dimensional sound fields and their complexity; in environments where multiple noise propagation paths can be very complex, this approach makes the lead control system too large and complex, and therefore not economical. The way is to abandon manufacturing to produce a big There is no noise space, and only the noise signal outside the eardrum of Luzhou, and a noise signal with the same waveform, equal size and opposite phase is generated near the eardrum to create a small noise-free space in 2. The technology is due to: 2 waveform estimation and anti-noise waveform generation and control come to the midnight, so there are a variety of noise-reducing headphones on the market to wear the headphones in the headset with 2 headphones to output the anti-arm sound waveform to reduce the eardrum The control system that produces the reverse noise is mainly divided into / adaptability. Among them, the control of the fixed parameter is to generate the anti-phase noise after passing the signal directly through the inverter circuit, however). From the 'noise wave fluctuation state, the style to reach the noise, and in addition to a gas transmission 曰 often, if the propagation road sound source and dynamic noise second stop to estimate the simple way involved with its ear drum, near这种The fixed-input input of this square 4 201042985 • The inverse noise generated by the method can cause the noise to be completely eliminated due to the undesired size and phase response of the inverter circuit, in order to compensate the phase delay of the control system itself. In order to cope with the time-varying noise, it is necessary to adopt an adaptive control system that can adjust the parameters automatically. 'At present, the adaptive active noise control system mainly applied to the earphone belt is a feedback type, and the method is to configure one inside the earphone. The micro-microphone is used to obtain the noise in the ear, and the anti-noise is estimated through the adaptive control system to output through the speaker in the earphone, thereby achieving the purpose of eliminating the sound. However, the main limitation of the feedback active noise control system is that after the speaker outputs the anti-warmer tone, the microphone inside the earphone is combined with the anti-sounding sound, and the microphone inside the earphone obtains residual noise, and adapts! The noise reference signal required by the raw control system is not directly available and can only be generated using synthetic techniques. When the synthesized noise reference signal is inaccurate, the estimated anti-noise will be inaccurate, thus greatly reducing the performance of noise cancellation. In view of the problems in the design of the above-mentioned conventional noise suppression technology, how to develop a better sound-eliminating effect, more ideal and practical, a soil and a juice, it is waiting for the relevant industry to think about breakthroughs. Goal and
◎ 方向者Q —有鑑於此，發明人本於多年從事相關產品之製造開發 十、要驗，4十對上述之目標’詳加設計與審慎評估後， 終私·一確具實用性之本發明。 【發明内容】 立，卜發明之主要目的’係在提供-種語音通訊裝置之噪 構及方法’其所欲解決之問題點，係針對習知主 動^ °呆音控制技術仍舊存在反噪音估測不準確而致噪音消 5 201042985 ·::除性低之問題點加以思索突破；所述語音通訊裝置传包括 =置本體、揚聲器、耳内麥克風、講話麥克風 控制系統所構成； 應性 本發明 〇 部增設一外 向性之麥克 性控制系統 明對照先前 得受話端在 音干擾的近 干擾的近端 估測反噪音 性。 解決問 部噪音 風，藉 的噪音 技術而 接聽遠 端語音 語音傳 正確性 題之技 麥克風 以取得 參考訊 言，俾 端語音 中分離 送到遠 並提升 術特點 ，該外 耳鼓夕卜 號，藉 可正確 時可降 噪音與 端；故 主動式 ，主要 部°喿音 部的0柴 此創新 地估測 低噪音 近端語 本發明 噪音消 是在該 麥先風 音，以 獨特設 出耳内 干擾； 音後， 將可確 除性能 裝置本體外 為具單一指 作為該適應 計，使本發 反噪音，使 並可從受噪 再將無噪音 實達到增加 之實用進步 〇 3、4圖 方法之較 申請上並 語音通訊 、行動電 的哚音消 克風、耳 第2、3 於—外掛 行動電話 統架構圖 所示’係 佳實施例 不受此結 裝置（如 話、固定 除結構， 塞式耳機 圖與第4 式耳機麥 i〇c之具 ，該噪音 【實施方式】 請參閱第2 、 之噪音消除結構及 說明之用，在專利 本發明係針對 耳塞式耳機麥克風 授式噪音控制技術 置於外掛式耳機麥 電話話機内部。如 噪音消除結構配置 機麥克風10B以及 的雙向音頻通訊系 本發明語音通訊襞置 准此等實施例僅供 構之限制。 外掛式耳機麥克風、 式弘§舌等）提出具前 該噪音消除結構可設 麥克風或直接配置於 圖所示，分別為所述 克風10A 、耳塞式耳 前授式噪音控制技術 ’肖除結構包括： 6 201042985 一裝置本體’泛指可用於進行雒。 •如行動電話、固定式電話等）之主音通訊的裝置（ .之外掛式耳機麥克風m 、第3圖所且如第2圖所示 風10B、以及第3圖所示之行動電γ、耳塞式耳機麥克 一揚聲器20，係可配置於該外掛1〇C等。 耳機内部（如第2圖所示）；式s 式耳機麥克風10A的 ο 克風10B的耳機内部（如第3圖:=置於該耳塞式耳機麥 動電話10C内部（如第4圖所示）下），或是配置於該行 輸出反噪音及遠端語音，輸出反噪五:述揚聲益20係負責 鼓附近的噪音，本發明中稱之A S 、目的疋為了消除耳 丨ΤΓ〜两斗内鸣立 土 則是進行雙向語音通訊時對方所傳來的二音退端語音指的 —耳内麥克風30 ’係可配置於:◎ Directions Q - In view of this, the inventor has been engaged in the manufacture and development of related products for many years. It is necessary to test, and after 40 detailed design and careful evaluation of the above objectives, the final private and practical invention. SUMMARY OF THE INVENTION The main purpose of the invention is to provide the noise and method of providing a voice communication device. The problem is to solve the problem. Inaccurate measurement and noise reduction 5 201042985 ·:: The problem of low de-sexuality is considered to be a breakthrough; the voice communication device includes: a body, a speaker, an in-ear microphone, a speech microphone control system; An extroverted mic control system is added to the crotch to estimate the anti-noise of the near-end of the near-interference of the previously accepted end. Solve the noise of the part, borrow the noise technology and answer the remote voice voice to the correctness of the technical microphone to obtain the reference message, the voice is separated and sent to the far and enhance the characteristics of the voice, the foreign ear drums It can reduce the noise and the end when it is correct; therefore, the active part, the main part of the 喿 部 的 此 此 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新 创新Interference; After the sound, it will be able to confirm the performance of the device as a single finger as the adaptor, so that the present anti-noise, and from the noise and then no noise to achieve practical progress 〇 3, 4 map method Compared with the application of voice communication, mobile audio, the sound of the wind, the ear 2, 3 in the external mobile phone system architecture diagram shown in the 'good example is not subject to this device (such as words, fixed removal structure, plug The type of earphones and the type 4 headphones are used for the noise. [Embodiment] Please refer to the second, noise cancellation structure and description. In the patent, the invention relates to earphones. The microphone-based noise control technology is placed inside the external headset microphone telephone. For example, the noise cancellation structure configures the microphone 10B and the two-way audio communication system. The voice communication device of the present invention is only applicable to these embodiments. Microphone, type hong tongue, etc.) proposed that the noise cancellation structure can be set up with a microphone or directly arranged in the figure, respectively, the gram wind 10A, earplug type ear pre-arrival noise control technology 'Shaw removal structure includes: 6 201042985 A device body 'generic' can be used to carry out 雒. • A device for the main tone communication such as a mobile phone, a stationary telephone, etc. (external headset microphone m, Figure 3, and wind 10B as shown in Figure 2, and the mobile power γ, earplug shown in Figure 3) Headphones microphone one speaker 20, can be configured in the plug-in 1〇C, etc. The inside of the earphone (as shown in Figure 2); the s-type headphone microphone 10A ο 克 10B inside the earphone (as shown in Figure 3: Placed inside the earphone headset 10C (as shown in Figure 4), or configured to output anti-noise and far-end voice in the line, output anti-noise five: Shu Yang Sheng 20 is responsible for the drum The noise, the AS in the present invention, the purpose of the purpose of the purpose of eliminating the deafness ~ two buckets inside the sound of the soil is the two-way voice communication when the two-way voice communication refers to the two-in-one voice-in-ear microphone 30 ' Can be configured at:
的:機内部並位於該揚聲器2。旁：第；匕 20;軸耳塞式耳機麥克風10B的耳機内部於該揚聲J 〇 位於揚t第3圖所不）’或是配置於該行動電話10C内； Μ聲器2。旁（如第4圖所示卜所述耳内麥克風3。： 個，型的麥克風用以取得耳機内部的聲音，包； =端語音以及次要近端語音；所述殘餘嚼聲； 近;：：反噪音與耳㈣音互相抵消後剩下的噪音，；I 道^_；則是使用者講話時的語音經由口腔内部回傳到耳 掛式耳機麥克 ί:::::，…機“於::: 麥香Λ J话10C背面（如第4圖所示）。哕外邻％立 参克風40是一個且„ 一 币4圃 孩外部噪音 部的啤音，太早一扣向性的麥克風，用以取得耳鼓外 “本發明中稱之為耳外噪音。該單一指向性的噪 201042985 :音麥克風配置的方向必需能夠避免 產生的主要近端語音。 田使用者4活所 :講話麥克風5。，係可配置於該外掛式 0A之靠近使用者口部 夕克風1 冲位置處（如第2圖所示）.赤3 於該耳塞式耳機麥克㈤議， 或疋配置 、寸恢夕見風1〇β (如第3圖所 於該行動電話10C下大^ # )，或疋配置 卜方（如苐4圖所示）。嗜謹样办+ 50是一個全向性麥> δ麥克風 主要近端語音以及；者講話時所產生的 接收到嶋，在本；該講話麥克風50所 ❹ 十知明〒稱之為次要耳外峰立 一適應性控制系@ Rn # θ .+1Q. ln 統6〇，係可使用數位訊號處理器（Dis ai s，ai pr〇_r)做為該系統的運算核心。可獨立配置二 爾式耳機麥克…卜部（如第2圖所 式耳機麥克風⑽外部（如第3圖所示）；或是配：； 行動電話10C内部（4m _ -置於〇亥 (如第4圖所示）。本發明中所描 適應性控制系統60主要肖括 出的 語音估測濾波器以及-欠晷折被吃立…# 王要近蜢 人要近知浯音估測濾波器的設計，且 ? 6亥外部噪音麥克風所 ❾ 音估測濾波器的參考浐， V音做為該反噪 哭 輪入汛唬以及主要近端語音估測濾波 Π: 訊號’並令主要近端語音估測渡波器的輸出 ‘、、、-人要近端語音估測濾波器的參考輸入訊號者。 又其中，該第2、3圖所示之方塊7〇，係代表電話話 '、々行動A A或固^式電話），經由該適應性控制系統 6〇進行嗓音消除後之主要近端語音將透過該電話話機？〇傳 送到遠端。 本 t 明所提出以LMS(least-mean-square)以及 FXLMS (filtered Heast-mean-square)演算法則為基礎的適應性控制系統完整 8 ’圖中 波器或 代表噪 一路控 經過的 通濾波 需經過 濾波器 轉移函 測濾波 近端語 ：W3(Z) 次要近 器 ％(z) 輸入訊號，並輸出反噪音_乃⑷，其 ~ D(z) = Xl(z)Wl(2) 反噪音-3⑷結合遠端語音G(z) 函數災Θ ，再與耳内噪音以及 ，因此耳内麥克風取得的聲音[/⑺ 使用Ζ 系統裝 音麥克 的等效 各項電 器、類 的各項 ) >如 數，用 器的轉 音估測 是次要 端語音 是以主 轉換表 置，其 風到耳 轉移函 子裝置 比數位 電子裝 第6圖 於近似 移函數 濾波器 近端語 201042985 :方塊圖如第5圖所示 ./輪出訊號、估測濾 徑的等效轉移函數， .傳輸路徑；％)為第 克風戴取聲音時所需 前置放大器、前置低 七聲器輪出聲音時所 比轉換器、後置低通 二路徑估測濾波器的 〇數，％〇)是反σ喿音估 反噪音；％⑷是主要 於估測主要近端語音 轉移函數，用於估測 反噪音估測濾波 〇 不法來表示輪入 中ρ(ζ)為主要路 内麥克風的聲波 數，主要包含麥 (包括麥克風、 轉換器），以及 置（包括數位類 所示；心）為第 第二路徑轉移函 ，用於估測耳内 的轉移函數，用 音估測濾波器的 要耳外噪音冬(ζ)為參考 關係可表示如下： 後，經過第二路徑轉移 次要近端語音ρ2⑺結合 可表示如下 · U(z) = D(z) + Q2 (ζ) + (G(z) - D(z))S(z) =G(z)S(z) + Q2 (z) + (D(z) _ D{z)S{z)) = G{z)S{z) + Q2(z) + r(^z) 其中為殘餘噪音，如果耳内噪音可以被反噪音抵 消，即= 0 ’貝ij 9: The inside of the machine is located in the speaker 2. Next: the first; 匕 20; the inside of the earphone of the shaft earphone microphone 10B is located in the speaker J 〇 is located in the mobile phone 10C or the mobile phone 10C; Next (as shown in Figure 4, the in-ear microphone 3::, the type of microphone is used to obtain the sound inside the earphone, the package; = the end voice and the secondary near-end voice; the residual chewing sound; ::The noise remaining after the anti-noise and the ear (four) sounds cancel each other out; I channel ^_; is the voice of the user speaking back to the ear-hook earphone via the inside of the mouth. Mike:::::,... "于:::麦香Λ J words 10C back (as shown in Figure 4). 哕外邻% Lishen Kefeng 40 is one and „ one coin 4 圃 children's external noise department of the beer sound, too early buckle The directional microphone is used to obtain the outside of the eardrum. "The extra-earth noise is referred to in the present invention. The single-directional noise 201042985: the direction of the tone microphone configuration must be able to avoid the generation of the main near-end speech. : Speaking microphone 5. It can be placed in the position of the external 0A near the user's mouth, as shown in Figure 2. Red 3 is in the earphone microphone (5), or 疋 configuration寸 夕 夕 see the wind 1 〇 β (as shown in Figure 3 on the mobile phone 10C under the big ^ #), or 疋 configuration Bu Fang (苐4 picture)) 嗜 样 + + + 50 is an omnidirectional wheat > δ microphone main near-end speech and; when the person speaks, the received 嶋, in the present; the speech microphone 50 ❹ 十知明〒 〒 次 次 次 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应 适应The core of the operation. It can be independently configured with the Erphone Mic... Microphone (such as the earphone microphone (10) shown in Figure 2 (as shown in Figure 3); or with:; Mobile phone 10C internal (4m _ - placed in 〇 Hai (as shown in Fig. 4). The adaptive control system 60 described in the present invention mainly includes a voice estimation filter and a 晷 晷 被 is eaten... #王要近蜢人要知知浯音Estimate the design of the filter, and the reference of the sound estimation filter of the 6-MHz external noise microphone, the V-sound as the anti-noise crying wheel and the main near-end speech estimation filter: signal 'and Let the main near-end speech estimate the output of the waver ',,, - the reference input signal of the near-end speech estimation filter In addition, the block 7〇 shown in the second and third figures represents a telephone call, a mobile action AA or a fixed telephone, and the main control is performed after the voice cancellation is performed via the adaptive control system 6〇. The voice will be transmitted to the remote end through the telephone. The proposed adaptive control system based on LMS (least-mean-square) and FXLMS (filtered Heast-mean-square) algorithm is complete 8 ' In the figure, the filter or the pass-through filter representing the noise path is filtered by the filter transfer filter. The near-language: W3 (Z) Minor near the %(z) input signal, and the output anti-noise _ is (4), its ~ D(z) = Xl(z)Wl(2) Anti-noise-3(4) combined with far-end speech G(z) function disaster, and then with the in-ear noise as well as the sound obtained by the microphone in the ear[/(7) using Ζ system The equivalent of various electrical appliances and classes of sound microphones) > As the number, the sound estimation of the device is that the secondary voice is set by the main conversion table, and the wind-to-ear transfer functor device is more than the digital electronic device. Figure 6 shows the approximate shift function filter near-end 201042985: block diagram as shown in Figure 5. ./round signal, estimated equivalent transfer function of filter path, .transmission path; %) is the ratio of the preamplifier and the front low seven sounder when the sound is taken by the first wind. The number of turns of the post-low-pass two-path estimation filter, %〇) is the inverse σ 估 估 estimated anti-noise; %(4) is mainly used to estimate the main near-end speech transfer function, used to estimate the anti-noise estimation filter 〇Important to indicate that ρ(ζ) is the number of sound waves in the main road microphone, mainly including wheat (including microphone, converter), and setting (including the digital class; heart) as the second path transfer function. It is used to estimate the transfer function in the ear. The noise of the extra-ear noise (冬) of the sound estimation filter can be expressed as follows: After the second path is transferred, the secondary near-end speech ρ2(7) can be expressed as follows. U(z) = D(z) + Q2 (ζ) + (G(z) - D(z))S(z) =G(z)S(z) + Q2 (z) + (D(z) _ D{z)S{z)) = G{z)S{z) + Q2(z) + r(^z) where is residual noise, if the noise in the ear can be offset by anti-noise, ie = 0 ' Ij 9
201042985 D(z) = D(z)S(z) 將D⑷=义,(Z)P(Z)以及—乃⑻= Zi(z)%⑷代入上式，可得到 %⑺的最佳解如下： S(z) 換言之，如果反噪音估測濾波器γ(ζ)可以同時估測 主要路徑的轉移函數，以及第二路徑的反轉移函數，則 即時估測出有效的耳内反噪音，達到噪音消除的目的。 f豢 而’反°桑音估測渡波器％⑺所使用的FXLMS演算法則要正 地收斂到最佳解，必需有正確的殘餘噪音做為調整濾波 參數的依據。但是，該耳内麥克風所取得的聲音除了殘 术曰之外還包括退纟而语音成份G(z)«S(z) _以及次要近端語 ，因此為了此夠取得正確的殘餘n喿音，必需移除耳 麥克風所取得的遠端語音成份與次要近端語音。 為此’本發明中係將遠端語音G(z)通過該第二路徑 測濾波器％)來估測近似耳内麥克風所包含的遠端語音 ❹伤G(艰(2)，再將耳内麥克風取得的聲音⑺減去G(z)左⑺ 剩下次要近端語音及殘餘噪音的估測值，表示如下： U,(z)^U(z)-G(Z)S(z) -G(z)(S(z) - S(z)) + Q2 (z) + = Q2(z) + R(z) 其中S(z)-兔Z)4。為了再移除其中的次要近端語音估 值么⑷，必需再藉助本發明中所提出的主要近端語音估 濾波器％⑺’以及次要近端語音估測濾波器％㈤。 主要近端§吾音估測濾波器％⑺的參考輸入訊號為主 出 可 缺 確 器 餘 音 内 估 成 測 測 要 ❹ 〇 201042985 ·:耳外嗓音不⑺，目標輸入訊號為講話麥克風所取得的 近端語音以及次要耳外噪音，似啦4⑺。假設主要近 音q⑻與次要耳外噪音A⑻之間具有統計意義上互不 •的關係，以及主要耳外噪音岑⑺與次要耳外噪音a⑻ 具有高度相關性’則當主要近端語音估測濾波器_ 數收斂到最佳解時，主要近端語音估測濾波器％㈡的 訊號為目標輸入訊號中與參考輸入訊號相關的成份， 之，主要近端語音估測濾波器％⑺的輸出訊號為次要 噪音的估測值尤⑷，誤差訊號（_+从卜毛㈣即為 近端語音的估測值ή⑺。主要近端語音的估測值⑺ 噪音消除後的主要近端語音，可傳送到遠端。 、次要近端語音估測濾波器％⑺的參考輪入訊號為 近端語音的估測值么⑺，目標輸入訊號為",⑴（次要 :音及殘餘噪音的估測值，么⑻以⑻）。假設次要近 音估測值么⑷與殘餘噪音估測值及⑷之間具有統計意 互不相關的關係，以及主要近端語音的估測值&⑻與 近端語音估測值4⑺之間具有高度相關性，則當次要 θ估測濾波器％⑺的參數收斂至最佳解時，次要近 音估測濾波器％⑺的輸出訊號為目標輪入訊號中與參 入汛娩相關的成份’換言之，該次要近端語音估測濾 %⑷的輪出訊號為次要近端語音的估測值么⑷，誤差 即為殘餘聲音的估測值如）。因此，所得到的殘：噪 測值Λ(ζ)可提供給反噪音估測濾波器做為調整參數的 …本發明中估測第二路徑轉移函數的適應性控制 弟6圖所不，所述估測第二路徑的方法，是採用系 主要 端語 相關 之間 的參 輸出 換言 耳外 主要 即是 主要 近端 端語 義上 次要 近端 端語 考輸 波器 訊號 音估 依據 統如 辨識 11 201042985 的原理 用以產 ti|乡東IfL 5(z) > 後置低 置放大 兩者的 徑估測 圖中第 號，目 果&) 解時， 該第二 S(z)。 Ο ’在適應性控制系統中設置-組白隨機訊號產生器 生富+所有步員率成㈣白&機訊號做為系統辨識的 號，該白隨機訊號同時輸入到第二路經估測渡波器 以及實際的第二路徑(包括數位類比轉換器、 通濾波器、揚聲器、耳機内一維聲場、麥克風、前 器、前置低通遽波器以及類比數位轉換器），如果 輸出差異很小時’即厂⑺处時，則第二路 慮波器如）可用於近似實際的第二路徑吣)' 。^ 6 二路徑估測遽波器仏）的參考輪入訊號為白隨機訊 標輸入訊號為白隨機訊號經過實際第二路徑後的結 ，當第二路徑估測濾波器女z)的參數值收:到最: 誤差訊號阶）達到最小值時，gp科2) = y⑷一仏卜〇時， 路徑估測淚波器々⑷將可用近似於實際的第二路徑 本發明之優點： 1. 主要藉由該裝置本體外部增設一外部噪音麥克風之創 新設計’相較於習知回授式適應性控制系統使用合成 方式產生噪音參考輪入訊號之作法而t，本發明採用 獨立外料音麥克風取得的主要耳外嚼音做為反噪音 估測滤波器的參考輸入訊號（可稱為前授式適應性控 制系統），因為沒有噪音合成時的誤差，因此可以择 加估測反噪音的正確性，並提升主動式噪音消除的 能。 2. 可消除主要近端語音所受到的噪音干擾：本發明 *近端語音估測濾波器的設計，以獨立噪音麥克風所 12 201042985 - 取得的主要耳外噪音做為參考輸入訊號，可以分離講 話麥克風中的次要耳外噪音與主要近端語音，因此可 以將沒有噪音干擾的主要近端語音傳送到遠端。 3. 了避免夂要近知s吾音對於反°喿音的估測造成干擾：反 噪音估測濾波器是根據殘餘噪音做為調整濾波器參數 的依據因此近端語音進入耳内麥克風後，會直接影 妻反本音估測濾波器的參數調整。為此，本發明特別 提出次要近端語音估測濾波器的設計，以主要近端語 音的估測值做為參考輸入訊號，可以分離出耳内麥克 〇 風的次要近端語音成份，因此可避免次要近端語音對 於反噪音的估測造成干擾。201042985 D(z) = D(z)S(z) Substituting D(4)=, (Z)P(Z) and — (8)= Zi(z)%(4) into the above equation, the best solution for %(7) is as follows S(z) In other words, if the anti-noise estimation filter γ(ζ) can simultaneously estimate the transfer function of the main path and the inverse transfer function of the second path, the effective in-ear anti-noise is estimated immediately. The purpose of noise cancellation. f豢 And the FXLMS algorithm used by the 'anti-Sangyin estimation waver%(7) should converge to the optimal solution correctly, and the correct residual noise must be used as the basis for adjusting the filtering parameters. However, the sound obtained by the microphone in the ear includes the retreat and the speech component G(z)«S(z) _ and the minor near-end language, so that the correct residual is obtained for this purpose. Tone, it is necessary to remove the far-end speech component and the secondary near-end speech obtained by the ear microphone. To this end, in the present invention, the far-end speech G(z) is passed through the second path filter (%) to estimate the far-end speech bruise G contained in the approximate in-ear microphone (hard (2), then the ear The sound obtained by the inner microphone (7) minus G(z) left (7) The estimated value of the secondary near-end speech and residual noise is expressed as follows: U, (z)^U(z)-G(Z)S(z -G(z)(S(z) - S(z)) + Q2 (z) + = Q2(z) + R(z) where S(z) - rabbit Z)4. In order to remove the secondary near-end speech estimate (4), it is necessary to rely on the main near-end speech estimation filter %(7)' and the secondary near-end speech estimation filter %(5) proposed in the present invention. The main near-end § my sound estimation filter% (7) of the reference input signal is the main out of the sufficiency of the residual sound estimate within the measurement ❹ 42 201042985 ·: the ear outside the voice is not (7), the target input signal is obtained by the speech microphone Near-end speech and secondary extra-earth noise, like 4 (7). Suppose there is a statistically unrelated relationship between the main near-sound q(8) and the secondary extra-earthy noise A(8), and the main extra-earth noise 岑(7) is highly correlated with the secondary extra-earal noise a(8)' then when the main near-end speech is estimated When the measured filter _ number converges to the optimal solution, the signal of the main near-end speech estimation filter %(2) is the component related to the reference input signal in the target input signal, and the main near-end speech estimation filter %(7) The output signal is the estimated value of the secondary noise (4), the error signal (_+ from Bu Wo (4) is the estimated value of the near-end speech ή (7). The estimated value of the main near-end speech (7) The main near-end speech after noise cancellation Can be transmitted to the far end. The reference round-robin signal of the secondary near-end speech estimation filter %(7) is the estimated value of the near-end speech (7), the target input signal is ", (1) (minor: tone and residual Estimated value of noise, (8) to (8)). Assume that there is a statistically unrelated relationship between the secondary near-sound estimate (4) and the residual noise estimate and (4), and the estimated value of the main near-end speech. & (8) and near-end speech estimate 4 (7) With high correlation, when the parameter of the secondary θ estimation filter %(7) converges to the optimal solution, the output signal of the secondary near-imaging filter %(7) is related to the target delivery in the target round-in signal. In other words, in other words, the secondary near-end speech estimation filter (4) is the estimated value of the secondary near-end speech (4), and the error is the estimated value of the residual sound, for example). Therefore, the obtained residual: noise measurement value ζ(ζ) can be provided to the anti-noise estimation filter as an adjustment parameter... In the present invention, the adaptive control of the second path transfer function is estimated. The method of estimating the second path is to adopt the parameter output between the main end-of-speech correlations, in other words, the main end-end semantics, the last time, the near-end language test, the signal, the sound estimate, the basis, the identification 11 The principle of 201042985 is used to produce ti|乡东IfL 5(z) > post-low-amplification both of the diameter estimation maps in the first, the result &) solution, the second S(z). Ο 'Set in the adaptive control system - group white random signal generator Sheng Fu + all the step rate into (four) white & machine signal as the system identification number, the white random signal is simultaneously input to the second path estimated The waver and the actual second path (including digital analog converters, pass filters, speakers, 1D sound field in the earphone, microphone, front, front low pass chopper, and analog digital converter), if the output is different When it is very small, that is, at the factory (7), the second path filter can be used to approximate the actual second path 吣)'. ^ 6 The two-path estimation chopper 仏) reference wheel input signal is the white random signal input signal is the white random signal after the actual second path, when the second path estimation filter female z) parameter value Receive: to the most: error signal level) When the minimum value is reached, gp is 2) = y(4), when the path is estimated, the path estimation tear wave 々(4) will be used to approximate the actual second path. Advantages of the invention: 1. The innovative design of the external noise microphone is mainly provided by the external body of the device. Compared with the conventional feedback adaptive control system, the noise is used to generate the noise reference wheel signal. The present invention uses an independent external sound microphone. The main external ear chew sound is used as the reference input signal of the anti-noise estimation filter (which can be called the pre-admission adaptive control system). Because there is no error in noise synthesis, the correct anti-noise can be estimated. Sexuality and enhance the ability of active noise cancellation. 2. It can eliminate the noise interference of the main near-end speech: the design of the near-end speech estimation filter of the present invention uses the main external noise obtained by the independent noise microphone 12 201042985 as a reference input signal, and can separate the speech. The secondary out-of-earth noise in the microphone is combined with the main near-end speech, so the main near-end speech without noise interference can be transmitted to the far end. 3. To avoid the interference caused by the estimation of the anti-sounding sound: the anti-noise estimation filter is based on the residual noise as the basis for adjusting the filter parameters. Therefore, after the near-end speech enters the in-ear microphone, It will directly adjust the parameters of the wife's anti-negative sound estimation filter. To this end, the present invention particularly proposes the design of a secondary near-end speech estimation filter, which uses the estimated value of the main near-end speech as a reference input signal to separate the secondary near-end speech components of the in-ear microphone. Therefore, secondary near-end speech can be prevented from causing interference to anti-noise estimation.
上述實施例所揭示者係藉以具體說明本發明，且文中 雖透過特定的術語進行說明’當不能以此限定本發明之專 利範圍；熟悉此項技術領域之人士當可在瞭解本發明之精 神與原則後對其進行變更與修改而達到等效之目^，而^ 等變更與修改’皆應涵蓋於如后所述之申請專利範圍所I ❹ 13 201042985 -【圖式簡單說明】 第1圖：係為°柴音 夂殘餘噪音波形 異對照圖。 第2圖··本發明應用於外掛式耳機來 見風之簡示圖。 第3圖：本發明應用於耳塞式耳機來 _ 見風之簡示圖。 笫4圖：本發明應用於行動電話之簡示圖 第5圖··本發明之適應性控制系统方塊圖 第6圖：本發明之估測第二路徑轉移 此一 由婁欠的適應性控制备 ^ 統方塊圖。 t剌糸 【主要元件符號說明】 外掛式耳機麥克風 1 〇 a 耳塞式耳機麥克風 ιοΒThe above embodiments are intended to be illustrative of the present invention, and are intended to be illustrative of the invention. After the principle is changed and modified to achieve the equivalent goal, and the changes and modifications of ^ should be covered by the scope of the patent application as described later. ❹ 13 201042985 - [Simple description of the diagram] Figure 1 : It is the contrast diagram of the residual noise waveform of Chai Yinyu. Fig. 2 · The present invention is applied to an external earphone to see a brief view of the wind. Figure 3: The present invention is applied to an earphone type _ See a brief view of the wind.笫4 diagram: a schematic diagram of the present invention applied to a mobile phone. FIG. 5 is a block diagram of the adaptive control system of the present invention. FIG. 6 is an estimation of the second path of the present invention. Prepare the system block diagram. T剌糸 [Main component symbol description] External headphones microphone 1 〇 a Earbud microphone ιοΒ
行動電話 1 0 C 揚聲器 20 耳内麥克風 3〇 〇噪音麥克風 4〇 講話麥克風 50 適應性控制系統 60 電话話機 7 Π 14Mobile Phone 1 0 C Speaker 20 In-Ear Microphone 3〇 Noise Microphone 4〇 Speech Microphone 50 Adaptive Control System 60 Telephone 7 Π 14