M268002 八、新型說明： 【新型所屬之技術領域】 本創作係關於-種If聲感測裝4，特別是指一種以非察覺性 方式監測並紀錄睡眠打軒之情形，擷取後的㈣訊號經由放大、 /慮波、A/D轉換等處理，並以程式判讀、 八到貝翰出，精以提供整體睡 眠品質評估之鼾聲感測裝置。 【先前技術】 根據調查’國人至少有五分之—有打㈣情形，而其中三分 7打軒程度病變較嚴重’已造成健康等問題七了軒只是輕哼 幾下，還無傷大雅，但嚴重相可能时隨呼料致呼吸暫停或 中止的危險情況發生，甚至導致腦部缺氧、睡眠中斷;在經年累 月後，記憶力、注意力、咮王5 & 、 力月匕、工作效率和工作品質都會逐 .漸變差，甚至會有痴呆等現象 矛兄冢冋柃，由於睡眠中斷會造成睡眠 ，不足，白天會有嗜睡之傾向’發生車禍或工作意外的比例也遠超 出—般正常人，因此打鼾的問題不能不重視。 睡眠相關呼吸疾病（sleep related breathing dis〇rd_一般利 用多項生理監測儀（P〇lysomnograph)在醫院中進行檢測，較為不 便且無法作長期監測。針對夜間多項生理監測儀的不便，本創作 人希望能發展-種較為簡易的量測儀器’在不失準確率的情況 了 ’以非察覺性的方式對受測者的打㈣情形做長期監控，告知 以者打#f的程度，做為是否就醫的參考，整體的設計需求如圖 所示’分成感測方式 & ^ 月b、輸出貢訊及商品化四個方向進行 M268002 考i ’使有打鼾問題的人可以先在家中做初步的檢測，若有需要 再至醫I1元做詳細檢查，初步檢測的資料也可做為醫生的參考。 睡眠時的打軒行為外界所能觀察到的變化有聲音以及鼻部 的氣體流動，在打鼾監測裝置的訊號擷取方式上，也是區分為聲 音感測器與氣體監測兩方面。如圖二所示，於西元】994年乙邛以 等學者使用鼻罩將鼻孔出氣導入空氣壓力感測器，量測鼻部氣體 /瓜動末皿/則打奸狀況，如此做法的優點是除了可以清楚紀錄打軒 情形之外，甚至連呼氣中氮氣、氧氣等氣體所佔比例也能測量， 缺點是量測時試驗者需載上鼻罩，既不方便又影響睡眠。 於西兀1995年Alfaro等學者以及於西元1999年幻爪等學 者使用夕克風收曰的方式針對軒聲所造成的聲音進行量測，M268002 8. Description of the new model: [Technical field to which the new model belongs] This creation is about a kind of If acoustic sensing device 4, especially a kind of monitoring and recording the situation of sleeping and playing in a non-perceptive manner. After processing such as amplification, wave filtering, A / D conversion, etc., and program interpretation, eight to Behan output, it is refined to provide a snoring sensing device for overall sleep quality assessment. [Previous technology] According to the survey, "at least one fifth of the Chinese people have hiccups, and three-sevenths and seven dozens of Xuanxuan have more serious lesions." This has caused health problems and other problems. When possible, it occurs with the dangerous situation of apnea or cessation caused by exhalation, and even leads to hypoxia and interrupted sleep in the brain; after years and months, memory, attention, King of Kings 5 &, force and efficiency, and work quality Will gradually worsen, even dementia and other phenomena, spear brother Tsukasa, because sleep interruption will cause sleep, lack of sleep, there will be a tendency to drowsiness during the day. 'The proportion of car accidents or work accidents is far more than normal people- The issue of snoring cannot be ignored. Sleep related breathing disease (sleep related breathing dis〇rd_) Generally, multiple physiological monitors (Polisomnograph) are used for testing in hospitals, which is inconvenient and cannot be monitored for a long time. In view of the inconvenience of multiple physiological monitors at night, the author hopes Able to develop a relatively simple measuring instrument 'without losing accuracy' to monitor the snoring situation of the subject in a non-perceptive manner for a long-term monitoring, and inform the person who hit #f to determine if For medical reference, the overall design requirements are shown in the figure 'Divided into sensing methods & ^ month b, output tribute and commercialization for M268002 test i' so that people with snoring problems can do preliminary at home first For testing, if you need to go to the doctor I1 for detailed inspection, the preliminary test data can also be used as a reference for the doctor. The changes in the behavior of Daxuan during sleep can be observed by sound and gas flow in the nose. The signal acquisition method of the monitoring device is also divided into two aspects: the sound sensor and the gas monitoring. As shown in Figure 2, in AD [Yi Yiyi et al. 994] Use a nasal mask to introduce the air from the nostril into an air pressure sensor to measure the nasal gas / melon / trapping situation. The advantage of this method is that in addition to clearly recording the situation of beating Xuan, even the nitrogen in the breath, The proportion of gases such as oxygen can also be measured. The disadvantage is that the tester needs to wear a nasal mask during the measurement, which is inconvenient and affects sleep. Yu Xiwu and other scholars in 1995 and Alfaro and other scholars in 1999 used Yu Ke The wind harvest method measures the sound caused by Xuansheng.
Alfar〇等學者將麥克風放在離試驗者口部30cm處，Kim等學者 則以兩個麥克風’―個放在喉頭上，—個放在頭上同時進行監 測。麥克風收音的缺點為雜訊多，易受環境背景聲音影響，優點 為麥克風體積小可㈣隱藏、設備便宜、靈敏度高，不^試驗 者接觸口乎非察覺性監測的原則，因此本打軒監測裝置也將以 聲音感測方式進行設計研究。 以聲音感测方式作軒聲監測，除了來自外界的雜音會影響訊 號的可U外，4驗者躺在床上的位置、面部所朝方向等因素， 可能也都會所_於西元㈣年―做了有㈣聲評估 的實驗’將感測麥克風擺設於試驗者頭上恢^，實驗發現試 驗細與仰臥造成的誤差小於2分貝，用揚聲器模擬打軒音源 M268002 放在床的中央與兩側，誤差也在2分貝内。根據這樣的實驗結 果他〜為軒每音里判斷的公差應為±2分貝，且將麥克風擺設 在試驗者頭部上方是—個適當的位置。 搁取軒聲訊號，於西元1995年Smithson則使用一種「音量 累。十分配圖（cumulative distributi〇n 〇f 加 〇v⑽响 wund ki)」，來量化打軒的等、級。如圖三戶斤*，紀錄整晚房間内的音 置大小，並依音量大小與所佔時間比例繪圖，如此一來只須觀察 曲線與x_y軸所圍面積的大小，即可判斷打鮮情形的嚴重程度 這個方法可將整晚打鮮的情形簡化成_個數字表示，大幅節省資 料儲空間，亦不失參考價值與可信度。 、 以麥克風作為軒聲訊號的擷取容易受到雜訊以及背景雜音 的影響’在消除雜訊的手段，大致可分成從硬體的增加狀電路 低通濾、波器、軟體數位低㈣波器、以及利用特殊演算法...等方 理仍有不足 式。硬體低通濾、波器能去除高頻、低頻雜訊，只留下鮮聲頻帶内 的訊號’ Smith議於西元1995年之研究中表讀聲的最低頻率 略J於1 00Hz，可以作為聲音濃波的頻帶標準。雖然軒聲多半分 佈於100Hz以下的頻域’但一般日常的談話聲、笑聲。咳嗽聲 等也多包含有這個頻帶内的聲音，因此僅以低通濾波器方式做處 /於西元2_年Jane等學者使用⑵個聲音樣本訓練類神經 系統，藉由22項頻域與時域的聲音特徵輸入值，系統可以在離 線狀態下分析經由錄音設備所儲存的聲音資料，辨識㈣並摒除 M268002 其他雜音與雜訊。如圖四所示，為—充滿雜訊的訊號，經過處理 後仍能清楚分辨鼾聲與非鼾聲；如圖五所示，為顯示一突波聲音 出現，系統則將其視為如咳嗽等雜訊而不列入考慮。 根據文獻Smithson於西亓i qqs含4匕,· 6 士 70 1995年指出鼾聲主要集中在 麵2的低頻地帶’我們收集了日常生活常見的34種聲音，藉 由Matlab的Simulink模組將樣本聲音分別通過〇〜^、 ㈣销、*、心職的模擬帶通低通較器，以驗證文獻 的結果。34個樣本聲音分為軒聲、人類聲音、以及曰常動㈣ 生活器具三大類’觀察其在頻域上的分佈情形，紀錄得到如表丄 的聲音頻域分佈表，㈣“〇”代表 立 耳曰私含该頻帶的聲 _ 可發現3個樣本㈣大都包含在〇〜2_z 頻帶内，其他聲音分布頻域則較廣。因此本創作人首 收聲音的頻率著手’採用濾波曲線較為平滑的丨接 波器（如圖六所示），對外界接收聲 自動式低通滤 4 曰先行濾波，去除高頻訊號。Scholars such as Alfar0 placed the microphone 30cm away from the mouth of the test subject, while Kim and other scholars used two microphones—one on the throat and one on the head for monitoring. The disadvantage of the microphone radio is that it has a lot of noise and is easily affected by environmental background sounds. The advantages are that the microphone is small and can be hidden, the equipment is cheap, and the sensitivity is high. The tester's contact is based on the principle of non-perceptual monitoring. The device will also be designed for sound sensing. Xuan sound monitoring with sound sensing method, in addition to the noise from the outside will affect the signal U, 4 factors such as the position of the bed in the bed, the direction of the face, etc., may also be _ Yu Yuannian _ do In the experiment with snoring evaluation, the sensor microphone was placed on the subject's head. The experiment found that the error caused by the test and the supine was less than 2 decibels. The loudspeaker M268002 was simulated with a speaker and placed on the center and sides of the bed. The error Also within 2 decibels. Based on the results of such experiments, the tolerance judged in each tone of Xuan Xuan should be ± 2 dB, and the microphone should be placed at an appropriate position above the subject's head. Taking the Xuansheng signal, in 1995 AD Smithson used a "cumulative distributi. Ten distribution map (cumulative distributi0n 〇f plus 〇v⑽ wund ki)" to quantify the level and level of playing Xuan. As shown in the figure of three households *, record the sound setting in the room all night, and plot it according to the volume and the proportion of time occupied. In this way, you only need to observe the size of the area surrounded by the curve and the x_y axis to determine the freshness. This method can simplify the situation of refreshing the whole night to _ a number, which greatly saves data storage space without losing reference value and credibility. The use of microphones as Xuan sound signals is easy to be affected by noise and background noise. In the method of eliminating noise, it can be roughly divided into hardware low-pass filter, wave filter, and software digital low wave filter. , And the use of special algorithms ... and so on. The hardware low-pass filter and wave filter can remove high-frequency and low-frequency noise, leaving only the signal in the fresh sound frequency band. Smith suggested that the lowest frequency of table reading in the 1995 study was slightly lower than 100Hz, which can be used as Band standard for sound rich waves. Although Xuansheng is mostly distributed in the frequency domain below 100Hz ’, daily conversation and laughter are common. Cough sounds also include sounds in this frequency band, so they only use low-pass filters. In 2nd year AD, scholars such as Jane used a sound sample to train the neural system. With 22 frequency domains and time Domain voice characteristic input value, the system can analyze the sound data stored through the recording device in offline state, identify and eliminate other noise and noise of M268002. As shown in Figure 4, it is a noise-filled signal that can clearly distinguish between snoring and non-snoring after processing; as shown in Figure 5, in order to show the emergence of a sudden sound, the system considers it as a cough and other noise. Information without consideration. According to the literature Smithson Yu Xiqin qqs containing 4 daggers, 6 shi 70 In 1995, it was pointed out that the snoring was mainly concentrated in the low frequency zone of face 2. 'We collected 34 kinds of sounds common in daily life, and sampled the sounds through Matlab's Simulink module Pass the analog band-pass low-pass comparators of 0 ~ ^, ㈣ pin, *, and heart to verify the results of the literature. The 34 sample sounds are divided into three categories: Xuan sound, human sound, and daily life. Observe the distribution in the frequency domain. The record can be obtained from the audio-frequency distribution table shown in Table ㈣, where “〇” stands for It is said that most of the three samples are contained in the 0 ~ 2_z frequency band, while other sounds have a wider frequency domain. Therefore, the author started with the frequency of the first received sound 'using a smoother filter (as shown in Figure 6) to automatically receive low-pass filtering of externally received sound. This means that high-frequency signals are removed.
M268002M268002
-------—- , I _ ―」--±_ - 現有之軒聲監測方式雖可紀錄並分析鼾聲之狀況，然其於傾 用時，對於使用者仍不方便且又影響睡眠，且必須在醫院中進t 一整晚之監測’未能作長期、整體的睡眠品質監測，甚為可惜 由此可見，上述習用方式仍有諸多不便，乃亟思加以曰‘ 新，亚經多年研究後終於成功研發完又J 【新型内容】 牛鼾名感測裝置。 本創作之目的即在於提供—種體積小〜 及靈敏度高，不需與試驗者接觸即可γ別:、隱藏、設備便宜 察覺性」監測原則之鼾聲感測裝置。 D乎「非侵入性與非 M268002 本創作之次一目的係在於提供一種更精巧、更輕量化及操作 谷易之許聲感測裝置。 可達成上述新型目的之鼾聲感測裝置，其主要包含有： -電源供料、統，該電源供應㈣係供給裝置中各子系統所 需之工作電源； 聲曰感測系統，該聲音感測系統係可接收各種不同之聲音 桌唬，亚配合音訊放大器轉換成電壓訊號； 濾波系統，该濾波系統係透過一組低通低通濾波器進行聲 'V的;慮波’並接收來自聲音感測系、統之聲音訊號，並將其區 分為鼾聲與非鼾聲； 一 PIC單晶片，該PIC單晶片係具備驅動pic單晶片内各系 統之運作，其内之系統包含有： -A/D轉換系統，該A/D轉換系統係接收來自低通濾波器 之鼾聲頻帶，並將其進行類比數位轉換； 方〒聲辨硪軟體系統’該鼾聲辨識軟體系統係接收A/D轉換 糸統轉換後之數位訊號，以辨別該數位訊號為鼾聲、非鼾聲或苴 他環境異音； * ~ -顯示介面，該顯示介面係將鼾聲辨識軟體系統之辨識結 果如.&測時間、打軒時間、打#次數及有睡眠呼吸中止症 (〇sASm徵之異諸料f訊透過咖直接㈤1提供使用 者整體性睡眠品質評估。 猎由上述各系統相互搭配使用，即以非察覺性之方式監測並 M268002 紀錄睡眠打鼾之情形。 【實施方式】 ^ 斤示，為本創作所提供之鼾聲感測裝置之架構示 思圖’其主要包括有： 一電源供應系統1，該電源供應系統1係包含12V變壓器， 並透過穩壓1C轉換，葬以报认# @^ v 猎以供給裝置中各子系統所需之工作電源； 一聲音感測系統2，該聲音感測系統2係可為電容式麥克 風藉以接收各種不同之聲音訊號，並配合音訊放大器放大其輸 出電壓訊號； -濾-波系統3,係依據—截止頻率為2_2的低通遽波器作 為軒聲辨識之依據，因此係可透過—組低㈣波器進行聲音頻帶 的濾波’並接收來自聲音感測系統2之聲音訊號，並將其區分為 軒聲與非軒聲：該濾波系統3係可為—階主動式低通濾波器，因 其具有低成本及低耗電量的優點，同時由於其濾波性能曲線較為 平級，也可避免少部份頻率較高的鼾聲被濾掉的情形，· PIC單晶片4 ’該PIC單晶片4係具備驅動其内各系統之 運作，其内包含有·· 一 A/D轉換系統41，該A/D轉換系統41係接收來自濾波系 統3之鼾聲頻帶，並將其進行類比數位轉換； 一鼾聲辨識軟體系統42,該鼾聲辨識軟體系統42係接收a/d 轉換系統轉換後之數位訊號，以辨別該數位訊號為鼾聲、非鼾聲 或其他環境異音； M268002 一顯示介面43，該_ 一人 之辨… ^不"面43係將鼾聲辨識軟體系統42 之辨I、，果，如··監測時間 mh間、打次數、異常鼾聲等 貝矾透過LCD直接顯千 # 估…“ ‘…错以提供使用者整體性睡眠品質評 估，该顯示介面43係可為數 、 _ 、 要文位輸出介面或其他顯示介面； 當聲音感測系統2接收蓉立^^咕、，一 、皮李蛴3 曰5 5儿亚經由放大電路後，透過濾 及糸、，先3區分軒聲盘非富鼓 魅^ 聲’再將其透過A/D轉換^ 41進行 類比數位轉換成數位訊號， 細 且丄由鼾聲辨識軟體系統U之判 斷，即可將擷取到的聲立八Μ* ώ ★ 曰刀辨為鼾聲或非鼾聲或其他環境显音， 並將擷取到的訊號計算為相” 貝戒使仵貧訊可透過顯示顯 〜 直接顯示，藉以提供整體性睡眠品質評估。 ，口月參閱圖八所不’其為梅錄自—整晚钟聲的五分鐘聲音波 形，其中每個波形起伏 …… 疋末自於鮮聲，即可發現鼾聲是非常有 規：的重覆進行。再請參閱圖九所示，其係以更仔細的角度觀察 四二連續叫其每次維持的時間和間隔的崎 此，、要知道正峰的鼾聲維持時 Τ 了 U和間隔打間特性，就可以利用 此一特性來區分所收錄的簦 ^ ㈣卓曰5fl唬是屬於鼾聲或非鼾聲。若有一 聲音訊號維持的時間符合鼾聲 貯车的特丨生，且在間隔一段時間後又再 有一次符合鼾聲長度特性的聲音 吁1的耳曰Λ旒產生，且間隔的時間長度也 必需要符合If聲特性時，就可蔣 5纟可將遠聲音判別為軒聲而非一般雜 音，同時計數器也會開始計數。利 1双刊用相冋的想法，受測者是否串、 有睡眠呼吸中止症（〇SAS)產生 " 、 ；王 < 異㊉鼾聲也有可能被量測到。 不過在曰常生活中其他整音叆早 他是曰還疋有機會呈現上述間隔模式，因此 12 M268002 可再增加另一項規則，就是此間隔模式必須連續發生兩次，若口 有發生-次則不列入計算’在連續第二次發生時才開始計算，並 將第一次未計算的鼾聲補正回來（如圖+ 汴不），如此一來誤判 的機會就會大幅減少。 請參閱如圖十一所示，其為綜合上述軒磬 4舯耸感測裝置之設計概 念流程圖’其係先利用聲音感測器將音波訊號轉換為電壓訊號， 並將訊號經由音訊放大器放大訊號，以進行後續處理，接著進入 截止頻率為2’低通低通濾波器，再經A/D轉換系統…進行 類比數位轉換’待轉換完成後，該鼾聲辨識軟體系統Ο即會進 行軒聲間隔模式之辨識，辨識後的㈣次數·示介面43進行 輸出；其中該鼾聲訊號經過放大以及濾波後，輸入至pic單晶 片4以進行A/D轉換系統41並儲存於電腦中，即可蟓製出如圖 十二之波形。其中，該聲波訊號的特性是會以依—個電壓準位做 上下振盪，但這樣的訊號形態不易處理，因此，第—步係先將電 壓準位以程式的方式歸零，所採用的方法為將1〇筆之a/d值計 算〜人平均’即可得到上下振盪的中心線，如圖十三所示，再將 新進的值與中心線相減，即可得到如圖十四準位歸零後之波形， 此時的梅取頻率為2_Hz。由於該訊號係呈上下振盪，然而向 下的資料點對鼾聲判斷並沒有幫助，因此採用尋找區域最大值的 方式來去掉不必要的訊號，也就是每丨〇個資料點中比較出一個 最大值來代表這10個點，如此就可將向下的點去除，該圖十四 的波形即簡化成如圖十五所示，此時的運算頻率為200Hz，播放 13 M268002 的鼾聲音量約為75分貝，圖中鼾聲所造成的最大電壓為 300mV，而無聲的電壓為75mV，兩者僅相差225mV，若鼾聲的 音置再更小，可能就會造成判斷上的困難，因此使用丨〇筆做— 次總合也就是類似積分的方式，來強化訊號的能量，拉大有聲與 無聲時的差距’以利於判斷（如圖十六所示）。然而其波形還是 有過多的起伏，因此再使用5點做平均的方式進行平滑化之曲線 (如圖十七所不)。當訊號成為如圖十七的狀態後，鼾聲由複雜 的上下《成為平滑的連續遞增與遞減訊號，此時程式每秒產生 2〇個資料點’我敎義若訊號連續5個資料點（約G2秒）均 為遞增的時間點為tl，表示㈣開始；連續連續5”料點（約 0·2秒）均為遞減的時間點為t2，表示軒聲結束；^幻2的差 為軒聲持續時間；她下一次連續5個資侧 的時間點為t3，t2盥t3的# Τ9 ρ & 曰 — -◊差丁2即為兩次鼾聲的間隔時間（如 圖十八所示）。如此即可 於入,辨硪軟體系統42所需要的兩個 矸聲持續時間丁1及鼾聲間隔時間T2。 、;〔鼾聲持頦％間與間隔時間的計算方法確定後，我們 觀察四位受測者連續Μ、 我們可 小可能，以定羞a …走错以传到兩者的的最大與最 疋義—次完整鼾聲的時間特性。另 十所示，四位受、、則去 * 0十九與二 又而者的鼾聲持續時間各 〇·5〜2.0秒之間， ϋ，但絕大部分落在 1起出2.0秒的情形只佔所有 1.7%視為量剛時 ）·7/❶，我們將這 的块差不列入計算，因此， 義即為0.5〜2 〇禾丨、夕„ έ v耳持、喟時間的定 間。而㈣間隔時間則絕大部分落在1.6〜3·6 M268002 秒之間’在這個間隔之外的情形只佔i 3%，也視為誤差不列入 片开藉由上述即可得到一次完整的軒聲定義為秒的有 聲“伴Ik著1.6〜3.6秒的無聲情形’並且必需要有規律的持續 進行。 另’該實際測試分係分成兩階段進行，第—階段係先在實驗 室中使用錄製的㈣測試’其主要目的係在一控制的環境下測試 本軒聲監測裝置對不同受測者㈣辨識率；第二階段再進行實地 測:式，主要目的在測試在實際睡眠環境下，受到包含環境雜音、 又列者翻纟#動等因素影響下，本軒聲監測裳置的軒聲辨識 率-中帛又鼻f聲辨識測試的實驗設備佈置如圖二十一所 示丄擴音器5與感測器6間間隔5Gem，並放置—個分貝計7。 放旦刀貝7的原因為一般軒聲的音量範圍在60〜90分貝之間， 因此播放時也必需在這個 礼固m，右最大鼾聲超過9()分貝， :又㈣所旎造成的音量，則該次實驗不算，將播放音 調小後再重新進行實驗。 其實驗結果如表2所示，共有四組料編號由A〜D，每_ 聲都進行三次測試，並且將_讀除以實際次數的值定義為: 二率，測試結果得到平均成功率最低為74%，最高到達心 日守母組的3次測試所 门 ㈣接近，相差最多為⑻ … ^的毅性及及可重複m透過μ 辨識軟體系統〜卩可有效且準確的判斷是否_ = 得之結果更為可靠且更有依據。 使其所 15 M268002 表2----------, I _ ― ”-± _-Although the existing Xuansheng monitoring method can record and analyze the status of snoring, it is still inconvenient and affects the user when it is used. Sleep, and must be monitored in the hospital overnight. 'It is not possible to monitor long-term and overall sleep quality. It is a pity that from the above, there are still many inconveniences in the above-mentioned customary methods. After many years of research, it has finally been successfully developed and developed. [New content] Burdock name sensing device. The purpose of this creation is to provide a kind of snore sensing device that is small in size and high in sensitivity, and does not require contact with the tester. D "Non-intrusive and non-M268002 The second purpose of this creation is to provide a more sophisticated, lighter, and easier to operate Gu Yi Xu sound sensing device. The snoring sound sensing device that can achieve the above-mentioned new purpose, which mainly includes There are:-Power supply and system, the power supply is the working power required to supply the various subsystems in the device; Acoustic sensing system, which can receive a variety of different sound tables, sub-audio The amplifier is converted into a voltage signal; a filtering system that performs sound 'V; considers waves' through a set of low-pass and low-pass filters, and receives sound signals from the sound sensing system and system, and distinguishes them into chirps And non-sound; a PIC single chip, which has the function to drive the various systems in the pic single chip, the systems within it include:-A / D conversion system, the A / D conversion system receives the low-pass The vocal frequency band of the filter is converted by analog to digital; the vocal identification software system 'the vocal identification software system receives the digital signal after the A / D conversion system to identify the digital signal The number is snoring, non-snoring, or other environmental noise; * ~-Display interface, which displays the recognition results of the snoring software system such as. &Amp; measurement time, Daxuan time, ## times, and sleep apnea The different materials of the disease (〇sASm sign) provide users with overall sleep quality assessment through coffee directly. 1. The above systems are used in conjunction with each other, that is, in a non-perceptive manner, M268002 is used to record sleep snoring. 【 Implementation method] ^ Shown, the schematic diagram of the structure of the snoring sound sensing device provided by the author, which mainly includes: A power supply system 1, which includes a 12V transformer, and is converted by a regulated 1C [@ ^ V Hunting to provide the working power required by the various subsystems in the device; a sound sensing system 2, which can be a condenser microphone to receive a variety of different sound signals, And cooperate with the audio amplifier to amplify its output voltage signal;-filter-wave system 3, based on-the low-pass chirp wave with a cutoff frequency of 2_2 is used as the basis for Xuan sound identification, so it can pass through- The wave filter performs sound frequency band filtering and receives sound signals from the sound sensing system 2 and distinguishes them into Xuan sound and non-Xuan sound: The filtering system 3 can be a first-order active low-pass filter, because It has the advantages of low cost and low power consumption. At the same time, because its filtering performance curve is relatively flat, it can also avoid the situation where a few high-frequency snoring sounds are filtered out. · PIC single chip 4 'This PIC single chip 4 It is equipped to drive the operation of the various systems in it, including an A / D conversion system 41. The A / D conversion system 41 receives the snoring frequency band from the filtering system 3 and performs analog digital conversion; Snoring recognition software system 42, which is a snoring recognition software system 42 that receives digital signals converted by the a / d conversion system to identify the digital signals as snoring, non-snoring, or other environmental noise; M268002 display interface 43, the _ one person Discrimination ... ^ No " face 43 is the discrimination of the snoring identification software system 42. If the monitoring time is mh, the number of hits, the abnormal snoring, etc., the alum is directly displayed on the LCD through the LCD # Estimate ... "'... Wrong to provide use For overall sleep quality evaluation, the display interface 43 can be a number, _, or text output interface or other display interfaces; when the sound sensing system 2 receives Rong Li ^ ^ Gu, I, Pi Li 3 3 5 5 After passing through the amplifying circuit, it will pass through the filter and 糸, 3 distinguish the Xuan Sheng Pan non-rich drum charm ^ sound ', and then through A / D conversion ^ 41 for analog digital conversion to digital signal, thin and 丄 sound recognition software Based on the judgment of the system U, the captured sound can be recognized as a snoring or non-snoring sound or other environmental sounds, and the captured signal is calculated as a phase. It can be displayed through display ~ direct display to provide overall sleep quality assessment. See the figure in Figure 8 and it's a five-minute sound waveform recorded from Mei—the bells all night, each of which fluctuates ... At the end of the sound from the fresh sound, you can find that the sound of the snoring is very regular: heavy Repeated. Please refer to FIG. 9 again, which is a more careful observation of the time and interval of the continuous maintenance of four or two times. To know the characteristics of U and interval between the snoring of the positive peak, You can use this feature to distinguish the recorded 簦 ^ ㈣ Zhuo Yue 5fl bluffs belong to snoring or non-snoring. If a sound signal is maintained for a period of time corresponding to the characteristics of the snoring car, and after a period of time, a sound that meets the characteristics of the snoring length is generated, and the length of the interval must also be consistent. If the sound characteristics, Jiang 5 纟 can distinguish far sounds as Xuan sound instead of ordinary noise, and the counter will start counting. Lee 1 uses the same idea, whether the subject is strung, and there is sleep apnea (〇SAS) produced ", " Wang " strange sounds may also be measured. However, in other daily life, he has not had the opportunity to present the above interval pattern. Therefore, another rule can be added to 12 M268002, that is, this interval pattern must occur twice in a row, if the mouth occurs-times The calculation will not be included until the second consecutive occurrence, and the first uncalculated snoring will be corrected (see Figure + 如图 No), so the chance of misjudgment will be greatly reduced. Please refer to Figure 11, which is a flowchart of the design concept of the above-mentioned Xuanyuan 4-sensor sensing device. It first uses a sound sensor to convert a sound wave signal into a voltage signal, and amplifies the signal through an audio amplifier. Signal for subsequent processing, and then enter the cut-off frequency of 2 'low-pass low-pass filter, and then through the A / D conversion system ... analog digital conversion' After the conversion is completed, the snoring identification software system 0 will perform Xuansheng Interval mode identification, the number of times of identification and display interface 43 are output; the 鼾 sound signal is amplified and filtered and input to pic single chip 4 for A / D conversion system 41 and stored in the computer. Make the waveform shown in Figure 12. Among them, the characteristic of the acoustic wave signal is that it will oscillate up and down according to a voltage level, but this signal form is not easy to handle. Therefore, the first step is to zero the voltage level in a programmatic manner. The method used is In order to calculate the a / d value of 10 strokes ~ average person, you can get the centerline of up and down oscillations, as shown in Figure 13, and then subtract the newly entered value from the centerline to get the accuracy shown in Figure 14. The waveform after the bit returns to zero, the culling frequency at this time is 2_Hz. Because the signal is oscillating up and down, but the downward data points are not helpful for the snoring judgment, so the method of finding the regional maximum value is used to remove unnecessary signals, that is, a maximum value is compared for each data point To represent these 10 points, the downward points can be removed. The waveform in Figure 14 is simplified as shown in Figure 15. At this time, the operating frequency is 200Hz, and the volume of chirping sound that plays 13 M268002 is about 75. In decibels, the maximum voltage caused by the snoring sound is 300mV, while the silent voltage is 75mV. The difference between the two is only 225mV. If the pitch of the snoring sound is even smaller, it may cause difficulty in judgment, so use 丨 〇 to do — The sub-summary is a method similar to the integral to strengthen the energy of the signal and widen the gap between the sound and the sound 'to facilitate judgment (as shown in Figure 16). However, its waveform still has excessive fluctuations, so it is smoothed with a 5-point average (see Figure 17). When the signal becomes as shown in Figure 17, the beeping sound will change from a complex up and down line to a smooth continuous increasing and decreasing signal. At this time, the program generates 20 data points per second. G2 seconds) are both increasing time points tl, indicating the beginning of ㈣; continuous 5 "material points (about 0.2 seconds) are decreasing time points are t2, indicating the end of Xuansheng; the difference between ^ Magic 2 is Xuan The duration of the sound; her next consecutive 5 asset-side time points are t3, t2 and t3 # Τ9 ρ & —— -◊ 差 丁 2 is the interval between the two sounds (as shown in Figure 18) . In this way, you can identify the two snoring durations D1 and snoring interval T2 required by the software system 42. After the calculation method of the snoring percentage and interval time is determined, we observe four The testees have continuous M, we can have a small chance, to determine shame a… go wrong to pass the time characteristics of the largest and most ambiguous—the second complete sound of the two. The other ten shows that four subjects, and then go * The duration of snoring sounds between 0 and 19 and 0.5 and 2.0 seconds each, ϋ, but never Most of the cases that fall from 1 to 2.0 seconds account for all 1.7% of the total amount.) 7 / ❶, we do not include this block difference in the calculation, so the meaning is 0.5 ~ 20. Evening „v v ear hold, time interval. However, the interval time between the most of them is between 1.6 ~ 3.6 M268002 seconds. Outside this interval, only 3% of the i is considered as an error. Xuansheng is defined as a "soundless situation with Ik with 1.6 ~ 3.6 seconds" and must be continued on a regular basis. In addition, the actual test line is divided into two stages. The first stage is first in the laboratory. The main purpose of using the recorded ㈣ test is to test the recognition rate of different test subjects by the Xuansheng monitoring device under a controlled environment; the field test is performed in the second stage, the main purpose is to test in the actual sleeping environment. Affected by factors including environmental noise, and other people who turn around, and other factors, Benxuan Sound monitors the Xuansheng recognition rate of the set-the test equipment layout of the Zhongyi and Nasal sound recognition test is shown in Figure 21. The microphone 5 and the sensor 6 are spaced 5Gem apart and placed with a decibel meter 7. The reason for putting the scallop 7 is that the volume range of the general Xuan sound is between 60 and 90 decibels, so it must also be at this time during playback. Ligu m, the right maximum snoring sound is more than 9 () dB : The volume caused by the sound is not counted, the experiment is not counted, and the experiment is performed again after the tone is small. The experimental results are shown in Table 2. There are four groups of materials numbered from A to D. Three tests were performed, and the value of _read divided by the actual number of times was defined as: the second rate. The average success rate of the test results was as low as 74%.毅… ^ 's perseverance and reproducibility m through μ identification software system ~ 卩 can effectively and accurately determine whether _ = The results obtained are more reliable and more basis. M268002 Table 2
本創作所提供之鼾聲感測裝置，與其 具有下列之優點： 他習用方式比較時，更 1·本創作係針對睡眠環境中 …U 中砧取到的所有聲音訊號，採用_ 聲與非軒聲交互比對的方式，並以 、 鼾 U , ¥ β线孝人體糸統以及根攄時 域作判斷’使其明確計算出使用者打_次數。 、 2·本創作係能準確判斷鼾聲盘 /、非7^聲，經以睡眠環境中可处 產生之各種樣本聲音作測試 h中了月匕 估。 使其犯更準確進行睡眠品質之評 3·本創作係在提供一種 度高，不需與試驗者接觸即1^易隱藏、設備便宜及靈敏 测。 可行實施例之具體說明， 之專利範圍，凡未脫離本創作技藝 監測原則，能在使用者家，作:，合乎「非侵入性與非察覺性」 τ π長期監 上列詳細說明係針對本釗作之—σ ^ ^ ^ A,/ ^ m 一 3〜貝心丹歷說明，惟 4 η施例並非用以限制本創作 乃m 16 M268002 精神所為之等效實施或變更，均應包含於本案之專利範圍中。 綜上所述，本案不但在空間型態上確屬創新，並能較習用物 品增進上述多項功效，應已充分符合新穎性及進步性之法定新型 專利要件，爰依法提出申請，懇請貴局核准本件新型專利申請 案’以勵創作，至感德便。 圖式簡單說明】 施例之附圖為： 請參閱以下有關本創作一較佳實施例之詳細說明及其附 圖，將可進-步瞭解本創作之技術内容及其目的功效；有關該實 圖一為本創作鼾聲感測裝置之整體設計需求圖； 圖二為量測鼻部氣體流動的實驗設備示意圖； 圖三為以一房間為例之音量累計分配示意圖； 圖四為雜訊訊號經過類神經演算法之示咅圖· 圖五為一突波經過類神經演算法輸出示意圖； 圖六為本創作軒聲感測裝置之低通濾波器頻譜圖； 圖七為本創作鼾聲感測裝置之架構示意圖； 圖八為本創作軒聲感測裝之連續㈣:錄圖； 圖九為本創作鼾聲感測裝置之連續鼾聲波形圖； 圖十為本創料聲感測裝置 貝7^耳間隔模式示意圖； 圖十一本轉聲感測裳置之設計概念流程圖； 圖 圖十二為本創特聲感測裝置之-I晶片紀㈣聲波形 17 M268002 冽裝置之訊號處理前之電壓訊號與 圖十三為本創作鼾聲感 中心線波形圖； 形圖； 圖十四為本創作㈣❹找置之訊號處理前之準位歸零 波 圖十五為本創作_測震置之以每1〇個資料點中一最大 值來代表這1 0個點之波形圖； 圖十六為本創作㈣感測裝置之以每1G點做—次總合之波 形圖； 圖十七為本創作軒聲感測裝置之以5點做平均之波形圖； 圖十八為本創作軒聲感測裝置之鼾聲判斷波形圖； 圖十九為本創作軒聲感測裝置之鼾聲持續時間統計圖； 圖二十為本創作軒聲感測裝置之鼾聲間隔時間統計；以及 圖一十一為本創作鼾聲感測裝置之實施例視圖。 【主要元件符號說明】 1電源供應系統 籲 2聲音感測系統 3濾波系統 4 PIC單晶片 41 A/D轉換系統 42鼾聲辨識軟體系統 43顯示介面 5擴音器 18 M268002 6感測器 7分貝計The snoring sensing device provided by this creation has the following advantages: Compared with his customary methods, 1. This creation is aimed at all sound signals obtained from the anvil in the sleeping environment ... U, using _ 声 和 非 轩轩 声The method of cross-comparison and judging by ', U, ¥ β lines to the human body system and the root time domain' makes it clear to calculate the number of user hits. 2. This creative department can accurately judge the snoring disks, non-7 ^ sounds, and test it with various sample sounds that can be generated in a sleeping environment. Make it more accurate to evaluate the quality of sleep 3. This creation is to provide a high degree, without the need to contact the tester, namely easy to hide, cheap equipment and sensitive detection. The specific description of the feasible embodiments, the scope of the patent, without departing from the principle of monitoring the creative skills, can be done at the user's home: "non-intrusive and non-perceived" τ π The long-term detailed description above is for this article Zhao Zuozhi—σ ^ ^ ^ A, / ^ m 1-3 ~ Bei Xin Dan Li explained, but the 4 η embodiment is not intended to limit the creation or equivalent implementation or change of the spirit of m 16 M268002, should be included in Within the scope of the patent in this case. In summary, this case is not only innovative in terms of space type, but also enhances the above-mentioned multiple effects over conventional items. It should have fully complied with the novel and progressive statutory new patent requirements, apply in accordance with the law, and ask your office for approval. This new type of patent application is' created with inspiration, to the utmost. Brief description of the drawings] The drawings of this embodiment are: Please refer to the following detailed description of a preferred embodiment of the present invention and the accompanying drawings, which will further understand the technical content of this creation and its purpose and effectiveness; Figure 1 shows the overall design requirements of the snoring sensor device. Figure 2 is a schematic diagram of the experimental equipment for measuring the gas flow in the nose. Figure 3 is a schematic diagram of the cumulative volume distribution of a room as an example. Figure 4 is the noise signal process Schematic diagram of a neural-like algorithm. Figure 5 is a schematic diagram of the output of a surge through a neural-like algorithm. Figure 6 is the low-pass filter spectrum of the creative Xuan sound sensing device. Figure 7 is the creative sound sensor device. Schematic diagram of the structure; Figure 8 shows the continuous sound of the creative sound sensor: recording; Figure 9 shows the continuous sound waveform of the creative sound sensor; Figure 10 shows the sound sensor of the original sound sensor. Schematic diagram of interval mode; Figure 11: Design concept flow chart of the rotary sound sensing device; Figure 12: The -I chip of the original sound sensor device, the chirp sound waveform 17 M268002 The voltage before the signal processing of the device Signal and Figure 13 is the waveform diagram of the center line of the sound of the creative work; Figure 14; Figure 14 is the zero-level wave before the signal processing for the creative work. Figure 15 is the creative work. A maximum value of each data point represents the waveform of these 10 points; Figure 16 is a waveform diagram of every 1G point of the creative device for the sensor; Figure 17 is the sound of Xuan Sheng The waveform diagram of the sensing device averaged at 5 points; Figure 18 is the snoring judgment waveform diagram of the creative Xuan sound sensing device; Figure 19 is the statistical diagram of the duration of the snoring sound of the Creative Xuan sound sensing device; Figure 2 Ten is the statistics of the snoring interval time of the creative Xuan sound sensing device; and FIG. 11 is a view of the embodiment of the creative snoring sound device. [Description of Symbols of Main Components] 1 Power supply system 2 Sound sensing system 3 Filter system 4 PIC single chip 41 A / D conversion system 42 Sound recognition software system 43 Display interface 5 Loudspeaker 18 M268002 6 Sensor 7 Decibel meter