.3 8.3 8
五、發明説明（/ ) (本發明的技術領域） 本發明係關於音樂表演的計分方法和系統。 (發明背景） 卡拉(《統是大家熟知的…個歧多的演唱者唱 歌可經由像CD片的預錄音樂的音源_同伴唱。原唱者的聲 音被消除而演唱使用者的聲音痤由麥克風的拾音使與原來 的背景音樂爲合輪出至味〗队。 一首音樂的構成包含了所有基本種種成分，像音節，音 長和速度等β爲了娛樂的目的，已有二些卡拉仳系統在表 演結束後提供計分。已發現早先的卡拉〇{[機器的計分並未 確實地依卡拉0Κ的演唱者和原先原唱者的聲音做完全的比 較。 (發明概述） 依本發明的具體實例，一種計分系統和方法提供了歌曲 結束時的計分，以其反應至輿參考信息接近音樂的程度。 此方法包含了偵測原唱預綠音樂和系統使用者兩者的音調 比較而做計分。 (附圖的敘述） 圖1爲卡拉0Κ系統的方塊圖； 圖2爲本發明之具體實例的功能方塊圖； 圖3爲圖2具體實例的更詳細的方塊圖； 圖4爲音調偵測電路的操作圖； 圖5Α和5Β所示爲音節的最終評價；且 圖6所示爲縮小的視窗寬度表。 3 - 卜紙張尺度適用中國國家標牟（CNS ) Μ规格（210x297公釐） ----------^1,------,ιτ------% - (請先閎讀背面之注意事項再填寫本頁) 經濟部中央榡準局貝工消費合作社印$. 經濟部中央標準局員工消費合作社印製 316961 A7 B7 五、發明説明（2 ) (本發明的詳細説明） 蹰1爲一方塊圖乃依先前所示卡拉0K機器10的結構，其 包括了雷射影碟音樂伴唱設備11 »這個雷射影碟音樂伴唱 設備11的组成由一雷射影碟自動換片機用以更換多量的影 碟11a，作爲音樂伴唱時的資訊記憶媒體〇機器1〇包括了 一個控制器12用以控制影碟自動換片機11選擇想要的影碟 。雷射影碟.自動換片機需由使用者操作輸入終端機輪入其 需求。這终端機在一些卡拉OK系統是一個可遙控的單元。 機器10更包括了一個信號處理器13,其包含混音器13a和 放大器13b，左和右邊的喇久14用於輸出重新產生的聲音 信號的輸出，一個影像顯示單元15用於顯示重新產生的影 像信號，其來自影碟11a的影像，且一個麥克風16 徒 用者的聲音使協調於背景音樂輸入放大器二：二 自雷射影碟自動換片機11混合了背景聲音信號，其爲伴奏5. Description of the Invention (/) (Technical Field of the Invention) The present invention relates to a scoring method and system for music performances. (Background of the invention) Kara ("All are well-known ... many singers can sing through the source of pre-recorded music like CDs_sing-song. The original singer ’s voice is eliminated and the singing user ’s voice is caused by The pickup of the microphone makes the team work in harmony with the original background music. The composition of a music contains all the basic components, such as syllables, length and speed. For entertainment purposes, there are already two Kara The system provides scoring after the performance. It has been found that the previous karaoke {{the machine ’s scoring does not exactly compare the sound of the karaoke and the original singer. (Overview of the invention) A specific example of the invention, a scoring system and method provides scoring at the end of a song, and responds to the extent that the public reference information is close to the music. This method includes the detection of both the pre-green music and the system user. Tone comparison and scoring. (Description of the drawings) Figure 1 is a block diagram of the Kara OK system; Figure 2 is a functional block diagram of a specific example of the present invention; Figure 3 is a more detailed block diagram of the specific example of Figure 2 Figure 4 is the operation diagram of the tone detection circuit; Figures 5A and 5B show the final evaluation of the syllable; and Figure 6 shows the reduced window width table. 3-The paper size is applicable to the Chinese National Standard (CNS) Μ specifications (210x297mm) ---------- ^ 1, ------, ιτ ------%-(please read the precautions on the back before filling this page) Ministry of Economic Affairs Printed by the Central Bureau of Industry and Fisheries Consumer Cooperatives. Printed by the Employees and Consumers Cooperative of the Central Bureau of Standards of the Ministry of Economic Affairs 316961 A7 B7 V. Description of the invention (2) (Detailed description of the invention) Step 1 is a block diagram according to the previously shown Kara 0K The structure of the machine 10 includes a laser disc music accompaniment device 11 »This laser disc music accompaniment device 11 is composed of a laser disc autochanger for replacing a large number of discs 11a as information memory during music accompaniment The media machine 10 includes a controller 12 to control the automatic disc changer 11 to select the desired disc. The laser disc. The automatic changer needs to be operated by the user to input the terminal into its needs. This terminal In some karaoke systems is a remote control unit. Machine 10 is more A signal processor 13 is included, which includes a mixer 13a and an amplifier 13b, the left and right Raju 14 is used to output the output of the regenerated sound signal, and an image display unit 15 is used to display the regenerated image signal. It comes from the video of the video disc 11a, and the sound of a microphone 16 is coordinated to the background music input amplifier two: two from the laser disc automatic changer 11 which mixes the background sound signal, which is the accompaniment.
音樂的信號，輿自麥克風16的歌聲的信號混合輸出至喇A 14 〇 依另一卡拉0K機器放音器11爲一CD自動換 音機用以容納多個_或卡”作音樂伴奏資訊記憶媒禮 和再生。 控制器12控制CD自動換片機或卡帶，以允許根要的 CD或卡帶，且’片機或卡帶乃經由使用者的: 擇。使用者也可經由-遙控單缝選擇 喇叭14輸出和再生聲音的音頻信號。於某些具 -圖形解碼器15a (虚線處）轉換圖形資料，由⑶實中^中生 本紙張尺度適用中國國家標準（CNS ) Λ4規格 (請先閲讀背面之注意事項再填寫本頁) —裝· 訂The signal of music, the signal of the singing voice from the microphone 16 is mixed and output to the La A 14. According to another karaoke device 0K, the player 11 is a CD automatic changer for accommodating multiple _ or cards "as music accompaniment information memory Matchmaking and regeneration. The controller 12 controls the CD automatic changer or cassette to allow the essential CD or cassette, and the 'disc or cassette is selected by the user. The user can also choose via -remote Speaker 14 outputs and reproduces the audio signal of the sound. The graphics data is converted in some graphics decoder 15a (the dotted line), and the paper standard is applied to the Chinese national standard (CNS) Λ4 specifications (please first Read the precautions on the back and then fill out this page)-Pack · Order
*316961 A7 ____ B7 ______ 五、發明説明（3 ) 的副碼資料成爲一影像資料，顯示於影像顯示器15上。麥 克風16的輸出被混合在處理器13。一個更詳細的卡拉0K機 器的敛述可以被發現在不同的專利内，如Oakamura et al, 的美國專利號5,194,682，在此可一同參考。 麥考圖2目的信號和參考信號兩者爲卡拉〇Κ***的歌唱 聲音。對分離原唱的歌聲系統，"目的信號"被定義爲卡拉 0K使用者所唱的歌聲，而"參考信號"被定義爲原唱者的歌 聲。因此，對於此系統的技術爲顯示目的歌聲對於麥考歌 聲的計分和水平。對於其他的系統，"目的信號••被定義爲 卡拉0K的輸出，即，卡拉弧使用者的歌聲加上背景音樂輸 出至味I队，而"參考信號"被定義爲卡拉〇Κ的歌的唱聲加音 樂。這來自卡拉0KCD。因此，這技術被用於決定唱的歌聲 和唱聲加音樂的分數和水平。另外，本技術也可使用於決 定音樂的分數和水平。在任何情形，分數和水平被決定於 目的信號對於參考信號。 經濟部中央樣準局貝工消費合作社印裝 ---------jr 1 裝— ί請先聞鲭背面之注意事項再填寫本耳) -訂_ 目的信號與現今發明的具體實施例一致，爲一麥克風16 輸出’在任何卡拉〇Κ系統。另一具體例爲其爲輸出至喇队 包括背景音樂的信號。於大部分的情形，此爲類比信號。 參考信號可能是類比或數位的，全依系統決定。一般地， 這信號來自雷射唱盤，CD或錄音/綠影帶，全依先前討論 的卡拉〇κ系統的型態而定。在此發生在圖1的處理器13的 計分過程’參考圖2，A/D轉換器和選擇器或21 (ADCM)在 圖1中的處理器13中，轉換目的信號和參考信號成爲數位 信號；〇若任何信號是數位的，此信號被選定爲數位化，且 經濟部中央橾準局員工消費合作社印製 A7 _______B7_ 五、發明説明（4 ) 不需A/D轉換器的作業。前置信號處理誤差校正器（pREM) 22處理自ADCM接收的數位信號。此過程使在評價音節時減 少錯誤。音節評償器（PPEM) 23 (—處理器）評償吸收自 PREM 22的數位信號的音節。此評償依固定間隔進行◊對 歌聲此間隔爲16秒，僅量測聲音的架構。後信息處理誤差 校正器（P0EM)24校正其誤差，由於ρρεμ引起的二倍或三倍 音調。一般，地，大部分的音節許償器演算引起二倍或三倍 的誤差，於某些評償音節。資料存儲器（DSTM)25可將資料 儲存的記憶裝置。被校正的音節，參考信號和目的信號兩 者被儲存在這裡。對應的錯誤資料也被儲存在相同的位置 ，被重覆寫上對應的正確資料。這是由PPEM 23所做的。 計分和水準評償器（26) (SLEM)爲一邏輯的組合，決定了目 的信號與參考信號的計分和水準。爲了達成此計分，它自 元件DSTM 25中存取記憶資料。當SLEM 26是有效時，ADCM 21和POEM 24是無效的，且保持一計分和水準的記錄。計 分和水準顯示器（SLDM) 15爲一影像顯示單元（VDU)，使用 於任何的卡拉0K系統》VDU 15也可應用於商業上的電視機 上。這也能用LCD或LED顯示元件，用於有卡拉0K的影音系 統，即家庭卡拉0K。也可爲一點矩陣顯示單元。 此技術執行於卡拉系統’如圖1的處理器13。參考信 號和目的信號被輸入。在執行時，目的信號的計分和水準 被產生。此技術連績地工作。其平行架構是可能的。 首先，它把參考信號槔爲輸入。 若輸入不是數位信號，那麼*SDCM轉成數位化’否則數 -6 - 本紙張尺度適用中國國家標準（CNS ) A4規格（210X297公釐） ---------f ά-- (請先閲讀背面之注$項再填寫本頁) 訂 於6 J刁 飯A邻1^^* 316961 A7 ____ B7 ______ 5. The subcode data of the invention description (3) becomes an image data, which is displayed on the image display 15. The output of the microphone 16 is mixed in the processor 13. A more detailed description of the Kara OK machine can be found in different patents, such as U.S. Patent No. 5,194,682 by Oakaka et al, which can be referred to here. Both the target signal and the reference signal of McCaw Figure 2 are the singing voices of the Karaoke system. For the singing voice system that separates the original singing, " purpose signal " is defined as the singing voice of the Kara 0K user, and " reference signal " is defined as the original singing voice. Therefore, the technology for this system is to show the score and level of the target singing voice to the McCaw singing voice. For other systems, " purpose signal •• is defined as the output of Kara 0K, that is, the output of Kara Arc user ’s singing voice plus background music is output to Ai I team, while " reference signal " is defined as Kara Kara The singing of the song plus music. This comes from Kara 0KCD. Therefore, this technique is used to determine the score and level of singing voice and singing voice plus music. In addition, this technique can also be used to determine music scores and levels. In any case, the score and level are determined by the target signal versus the reference signal. Printed by the Beigong Consumer Cooperative of the Central Bureau of Standards of the Ministry of Economic Affairs --------- jr 1 Pack — Please listen to the precautions on the back of the mackerel before filling in the ears) -Subscribe_ the purpose signal and the specific implementation of the present invention The example is the same, as a microphone 16 outputs' in any Karaoke system. Another specific example is the signal output to the cheerleading team including background music. In most cases, this is an analog signal. The reference signal may be analog or digital, depending on the system. Generally, this signal comes from a laser turntable, CD, or recording / green tape, depending on the type of karaoke system previously discussed. Here the scoring process of the processor 13 of FIG. 1 'refer to FIG. 2, the A / D converter and selector or 21 (ADCM) in the processor 13 of FIG. 1, the conversion destination signal and the reference signal become digital Signal; 〇If any signal is digital, this signal is selected to be digitized, and the A7 _______B7_ is printed by the Employee Consumer Cooperative of the Central Bureau of Economics of the Ministry of Economic Affairs. 5. Description of the invention (4) Operation without A / D converter. A pre-signal processing error corrector (pREM) 22 processes the digital signal received from ADCM. This process reduces errors when evaluating syllables. A syllable evaluation device (PPEM) 23 (—processor) evaluates the syllables of digital signals absorbed from PREM 22. This evaluation is performed at a fixed interval. ◊ Pair of singing voices. This interval is 16 seconds, and only measures the structure of the sound. The post-information processing error corrector (POEM) 24 corrects its error due to the double or triple tone due to ρρεμ. In general, most syllable compensator calculations cause double or triple errors, which can be used to evaluate syllables. Data storage (DSTM) 25 is a memory device that can store data. The corrected syllable, both the reference signal and the destination signal are stored here. Corresponding incorrect data is also stored in the same location, and the corresponding correct data is overwritten. This is done by PPEM 23. The scoring and level compensator (SLEM) (SLEM) is a logical combination that determines the scoring and level of the target signal and the reference signal. To achieve this score, it accesses the memory data from the device DSTM 25. When SLEM 26 is valid, ADCM 21 and POEM 24 are invalid, and a record of scores and standards is maintained. The Scoring and Level Display (SLDM) 15 is an image display unit (VDU). It can be used in any Kara OK system. VDU 15 can also be applied to commercial TV sets. This can also use LCD or LED display elements for audio and video systems with karaoke, ie home karaoke. It can also be a dot matrix display unit. This technique is implemented on the processor 13 of the karaoke system as shown in FIG. The reference signal and the destination signal are input. During execution, the score and level of the destination signal are generated. This technology works continuously. Its parallel architecture is possible. First, it takes the reference signal as input. If the input is not a digital signal, then * SDCM is converted to digitized, otherwise the number is -6 Please read the note $ item on the back before filling out this page) Booked at 6 J Diaofan A Lin 1 ^^
IT*發明說明（) 位信號資料可由ADCM 2〗直接選取。數位資料然後被輸入 PREM由框偏壓成一框，每一框爲2〇 ms。 由PPEM 23評價每一音節，檢查是否此框是有聲音的， 若有，此評償音節被存放於DSTM 25。此過程持續，每16 秒的聲音資料（每框2〇ms)ppEil 23在16秒後停止且啓動 SLEM 26。 SLEM 26俵ADCM 21無效且啓動poem 24。POEM 24更正存 放於DSTM 25的錯誤音節。在更正時，儲存在耵诎25的音 節’是由SLEM 26所輸入。SLEM 26然後評償其分數和水平 ，其計分式爲 100(1-/(PD) * (PD))% 或爲 100 (1- I PD | ) % (此處PD爲目的信號基本頻率對參考信號基本頻率的偏差 百分比） PD==(參考信號基本頻率一目的信號基本頻率）/ (參考 信號基本頻率） 或，如PD依音節表示， 經 濟 部 中 央 標 準 % 貝 工 消 费 合 作 社 (請先閱讀背面之注^^^項再填寫本頁> 訂 PD==(目的信號音節一參考信號音節）/(目的信號音節） 以上乃是對目的信號的計分定義，基於目的信號對參考 信號的基本頻率的偏差百分比。 爲了獲得評償計分的好結果，建議取基本頻率（或音 節）的平均，在一定間隔内。對唱的歌聲，取樣在矿kHz 每一框内20毫秒，平均取音框持續16秒是最好的選擇。此 本紙法尺度遥用中國國家梂準（CNS ) Α4规格（210X297公釐） Α7 Β7 316561 五、#明叙明（6) " 16秒不應包括任何無聲之框。對其它取樣速度，框的總數 應相稱地增加或減少〇水準的定義乃基於PD値。若ρρ等於 (請先閲讀背面之注意事項再填寫本頁) 零，目的信號水平相對於參考信號是相同的。稱爲正常水 平（NORMAL LEVEL)。 如果PD大於零’目的信號水平被定義爲正常水平之上。 如果PD小於零’此水平被定義爲正常之下。 目的信號的水準可被定義爲依多個1/4的半音變化。此 定義可使某些客户进喪一特别是不善於卡拉〇1[的人，或 勒至卡拉0K的人。因爲這理由，三種水準因而被定義。 在每一卡拉0K表演結束，表演的計分和水準二者將被顯 示。此水準將依下面滿意的信息。如下例： NORMAL LEVEL正常水平（良好的表演） BELOW NORMAL LEVEL水平之下（請試著再唱高調可表現 更好） ABOVE NORMAL LEVEL (水平之上）（請試著再唱低調些可 表現更好） 經濟部中央標準局員工消费合作社印製 在已發表的文獻中有一種計算數量的方法，它可對目的 fs说和參考彳§说兩者做坪價其基本頻率或音節。Gold-. Rabiner音節評價器被應用且之後將討論。Slem 26記綠數 値直到它自ADCM 21接收一個"選定"的信號。此信號告知 是否卡拉0K的表演的歌聲是完成與否。在接收選擇信號時 ，SLEM 26顯示其記綠的分數和水平於SLDM 15上。 VDU 15,用於商業上矽TV，一般使用於專業的卡ί〇Κ系 統’對此系統，SLDM 15爲VDU或TV。於專業的卡拉0Κ系統 經濟部中央標準局員工消費合作社印製 A7 --------—B7_______ 五、發明説明（7 ) ，圖形顯示可被用於顯示分數和水平。例如一個動態彩 色的筆畫過VDU或TV螢幕，寫下分數和水平。 您的分數：71.4% 您的歌聲：在正常水平之下（BELOW NORMAL· (請再試唱高一路的Key，以改善水平，享受卡拉管弦 樂隊） 對家庭卡，拉0K系統，SLDM 15爲LCD，LED或點矩陣顯示 單元。數位顯示器解決這些顯示單元。 圖3爲圖2的更詳細欽述，即，圖2中每一方塊更詳細 地敘述。每一方塊的功能依下面箭頭流程描述。 ADCM 21包括了一個類比轉數位的轉換器2ia，和一個信 號選擇遜輯21b。它以目的信號和參考信號做輸入。目的 信號在卡拉0K系統中爲麥克風16的輸出，且一般爲類比信 號。參考信號包括了兩種元件。一爲"參考歌聲"，另一爲 "參考背景音樂”。參考信號一般是數位信號。 於某些有DVS (數位顯示系統）的機器記綠卡拉0K的參 考歌聲在一分離的頻道（未混入背景音樂）。於曰本的碟 影（LD)中，歌聲在碟片中爲一分離的頻道。對LD而言，歌 聲的輸出爲類比，因此類比到數位的轉換是必須的。而 DVS卡拉OK CD爲數位的。選擇遜輯21b，選擇這些機器， 自麥克風16來的目的信號的歌聲，由系統信號處理器13 LD或DVS卡拉OK CD 11a的參考信號歌聲。若任何被選到的 信號爲類比的，A/D轉換、器23a被啓動，否則不會啓動❶ 一般的影像卡拉0K機器，背景音樂和原唱歌聲爲混合的 本紙張尺度適用中國國家標準（CNS ) A4規格（210X297公釐） ---------f 1 裝-----—訂-----"瘃 (請先閲讀背面之注意事項再填寫本頁) 經濟部中央標準局員工消費合作社印裝 A7 B7 五、發明説明（8 ) ，參考信號爲經由CD组合原唱歌聲加上背景音樂，和目的 信號爲類比的卡拉0K歌聲加上背景音樂輸出至喇久。這是 使用A/D轉換器21a轉換爲數位的。 選擇邏輯21b首先選擇目的信號。如果需要則轉換爲數 位的。此信號之後被應用至PREM 22。PREM 22是由偏移暫 存器22a和減法器22b組成。它接收的數位資料爲一 16位元 的正整數。，每一資科在偏移暫存器22a中往左偏移1個位 元。然後1減去偏移暫存器中的資料，藉由減法器。然後 偏移暫存器的資料往右偏移1個位元。再後資料再往左偏 移一個位元。處理過的資料再應用至PPEM 23。 PPEM 23包括音節評償器23a和比較測定儀22b與控制暫 存器23c的結合。此音節評價器其描述和Gold和Rabiner於 義國聽覺公會期刊1969年Vo丨.46, NO.2 (第二部），頁 數442-448，抬頭爲"Parallel Processing Techniques for Estimating Pitch Period of speech in Time Domain"。此系統包括低通過濾器51以去除第一共振區域 。此低通過濾波形藉由波峰和波谷許價器53處理。波峰和 波谷的六組測量被引出。有六個”單一”相同的評價器55， 每一個由偵測器53的六組中的一個工作。每一個評價器爲 一波峰偵測下降電路。看圖4，循著每一個偵測脈波，在 空的間隔中有一單純的指數衰減。無論何時一個脈衝超過 此下降電路的水平（在衰減中），則被偵測且此電路被重 置。此下降時間常數和每一偵測器的空白時間爲平垣地評 償偵測器的音節的函數。最终的音節計算數値乃基於來自 本紙張尺度適用中國國家標準（CNS )八4規格（210Χ297公釐） I I I I ϋϋ———^ 1 裝^ I I I 訂I I I .γ# (請先閲讀背面之注意事項再填寫本頁) 316361 A7 B7 五、發明説明（9 ) 每一·•簡單"音節評償器和多數規律選取被完成以決定依六 者結論來決定音調的檢查。最终計算數値被執行在決定的 製作者57 ’其可被視爲一有記憶的電腦，一算術邏輯演算 和控制硬體，操作進來的信號。在任一時間t 〇 —個音節 的許償被製成由： 1. 形成一個6X6矩陣的音節評償。看圈5B。行爲個别的 偵測器和列,爲音節的評償。首三列爲三個最近的音節評價 。第4列爲第1和第2列的總和；第5列爲第2和第3列 的總和；第6列爲首3列的總和。形成這矩陣的技術如圖 5A所示。此矩陣最末三列的理由爲有時個别的偵測器將指 定第二或第三調和波比基本波且將輸入最後三列被校正比 較於三個最近的音節評償。 2. 比較在矩陣的首列的每一個輸入値到矩陣與數符合的 數目的另35個數値。特别地Pn (i = l, 2, 3, 4, 5，6)爲 最普遍（符合的最大數目）被用做最終評價的音節。 經濟部中央標準局員工消費合作社印製 ---------^ 1 裝-- (請先閲讀背面之注項再填寫本頁) .叫線 爲了決定是否兩音節評價爲"符合"可觀察其比率優於觀 察其差異。然而，比率測量可能非常接近避免分劄數値的 需求。因爲在許多演説的片段中，有相當大的變量在連續 的音節測量，這是有用地包括幾個出發値，以定義"符合” ’且然後試著選擇》對每一全部音節計算，此出發點獲得 最一致的回答。由這説明，我們現在定義圖3中方塊57的 計算數値。 圖6所示爲16個符合艰窗寬度的表列。如圖5所指，僅 有自给予的偵測器中最接近的評償音節爲一 ••被選者"做爲 "11 - 本紙張尺度適用中國國家標準（CNS ) A4規格（210X297公釐） A7 B7 經濟部中央標準局員工消費合作杜印製 五、發明説明（10 ) 最終選擇。此被選者爲六個可能選擇中的一個，爲"正確 的"音節。爲了決定此"勝利者"，每一候選者被做數値的 比較與所有剩餘的35個音節者〇此比較被重覆四次，相對 於圈6表格的每一行。自每一行，適當的視窗寬度被選擇 ，視爲關係候選者評償的功能。 符合的數字被製成表格後，此數量減去1的偏移量。第 二行然後重，覆此測量；這次視窗更寬了，增加了符合的可 能性，但，在捕償中，自編輯的數量減去2的偏移量。用 此方法對全部的四行重覆此計算之後，最大的偏移量被用 作符合的數量，代表特别的音節評償。現在剩餘的5個候 選者重覆所有的過程，獲勝者的選擇爲有最大符合偏移量 的數目。 每20毫秒做評價一次且結論的平均爲每一 2〇毫秒計算完 成一次，該，10秒即50X 10或500做爲平均。這決定聲音 的音調》此評償音調於比較器23b中被比較，輿檢查是否 輸入框對此音調爲聲音。如此框爲聲音，控制暫存器23c 記綠此框的長度或時間，以秒記綠。聲音資料然後被寫入 DSTM 23中，且任何非聲音的資料被控制暫存器23c捨棄。 DSTM 23也可爲信息處理器13的記憶。 比較器23b和控制暫存器23c方塊維持累積時間，依聲音 框的每一框，且維持記綠音節於DSTM 23直到全部時間爲 16秒。如果這已完成時，控制暫存器23c傳送一信號至 SLEM 26中的啓用/取消居生器26a。然後此啓動/取消產 生器26a傳送一信號到ADCM 21。然後選擇邏輯21b選擇參 本紙i尺度適用f國國家標準（C&S ) A4規格_( 210x297公釐） (請先閲讀背面之注$項再填寫本頁) 丨裝_ *•11IT * invention description () Bit signal data can be directly selected by ADCM 2〗. The digital data is then input into PREM and biased by the frame into a frame, each frame is 20 ms. Each syllable is evaluated by PPEM 23, and it is checked whether there is sound in this box. If so, the evaluation syllable is stored in DSTM 25. This process continues, every 16 seconds of sound data (20ms per frame) ppEil 23 stops after 16 seconds and starts SLEM 26. SLEM 26. ADCM 21 is invalid and poem 24 is activated. POEM 24 corrects incorrect syllables stored in DSTM 25. At the time of correction, the syllables stored in 诵 诎 25 'are input by SLEM 26. SLEM 26 then evaluates its score and level, its scoring formula is 100 (1-/ (PD) * (PD))% or 100 (1- I PD |)% (here PD is the basic frequency of the target signal Percentage deviation of reference signal basic frequency) PD == (reference signal basic frequency-destination signal basic frequency) / (reference signal basic frequency) Or, if PD is expressed in terms of syllables, the Ministry of Economic Affairs Central Standard% Beigong Consumer Cooperative (please read first Note ^^^ on the back then fill in this page> Order PD == (Destination signal syllable-reference signal syllable) / (Destination signal syllable) The above is the scoring definition of the destination signal, based on the destination signal to the reference signal The deviation of the basic frequency. In order to obtain a good result of the evaluation score, it is recommended to take the average of the basic frequency (or syllable) within a certain interval. The singing of the duet is sampled within 20 milliseconds in each frame of the mine kHz, and the average sound is taken. The frame lasting 16 seconds is the best choice. This paper method scale is used in China National Standard (CNS) Α4 specification (210X297 mm) Α7 Β7 316561 V. # 明 述 明 (6) " 16 seconds should not include any Silent frame. Take other Speed, the total number of frames should be increased or decreased proportionally. The definition of the level is based on the PD value. If ρρ is equal to (please read the precautions on the back before filling this page) zero, the target signal level is the same as the reference signal. Normal level (NORMAL LEVEL). If PD is greater than zero, the target signal level is defined as above the normal level. If PD is less than zero, the level is defined as below normal. The level of the target signal can be defined as multiple 1 The semitone change of / 4. This definition can make some customers enter into one, especially those who are not good at Kara 〇1 [, or those who reach Kara 0K. For this reason, three levels are thus defined. In each Kara 0K At the end of the performance, both the performance score and the level will be displayed. This level will be based on the following satisfactory information. The following example: NORMAL LEVEL normal level (good performance) BELOW NORMAL LEVEL level below (please try to sing high-profile again) Better performance) ABOVE NORMAL LEVEL (above the level) (please try to sing a lower profile for better performance) Printed in the published article by the Staff Consumer Cooperative of the Central Standards Bureau of the Ministry of Economic Affairs There is a method for calculating the quantity, which can be used to evaluate the basic frequency or syllable of the target fs and reference §. Both Gold-. Rabiner syllable evaluator is applied and will be discussed later. Until it receives a " selected " signal from ADCM 21. This signal tells whether the singing of Kara 0K ’s performance is complete or not. When receiving the select signal, SLEM 26 displays its green score and level on SLDM 15. . VDU 15, used in commercial silicon TVs, is generally used in professional card systems. For this system, SLDM 15 is VDU or TV. Printed on the professional Kara 0K system. Employee's consumer cooperative of the Central Bureau of Standards of the Ministry of Economic Affairs. A7 --------— B7_______ V. Description of the invention (7). The graphic display can be used to display scores and levels. For example, a dynamic color stroke has been drawn on a VDU or TV screen, and the score and level are written. Your score: 71.4% Your singing voice: below the normal level (BELOW NORMAL · (Please try singing the key of the higher way to improve the level and enjoy the karaoke orchestra) For family cards, pull 0K system, SLDM 15 for LCD, LED or dot matrix display unit. The digital display solves these display units. Figure 3 is a more detailed description of Figure 2, that is, each block in Figure 2 is described in more detail. The function of each block is described according to the arrow flow below. ADCM 21 includes an analog-to-digital converter 2ia, and a signal selection series 21b. It takes the destination signal and the reference signal as input. The destination signal is the output of the microphone 16 in the Kara OK system, and is generally an analog signal. Reference The signal includes two kinds of components. One is "quote the reference song", the other is "quote the background music". The reference signal is generally a digital signal. For some machines with DVS (digital display system), the green karaoke 0K is recorded. Refer to the singing voice on a separate channel (no background music). In Japanese DVD (LD), the singing voice is a separate channel on the disc. For LD, the input of singing voice For analogy, the conversion of analog to digital is necessary. The DVS karaoke CD is digital. Choose the 21b, select these machines, the singing of the destination signal from the microphone 16, the system signal processor 13 LD or DVS Karaoke CD 11a reference signal song. If any selected signal is analog, the A / D converter and the device 23a are activated, otherwise it will not be activated ❶ General video karaoke 0K machine, background music and original singing sound are mixed The size of the paper is applicable to the Chinese National Standard (CNS) A4 specification (210X297 mm) --------- f 1 pack -----— order ----- " 瘃 (please read the back first Please pay attention to this page and fill in this page) A7 B7 printed by the Employees ’Consumer Cooperative of the Central Bureau of Standards of the Ministry of Economic Affairs 5. Description of the invention (8), the reference signal is the original singing sound combined with the background music via CD, and the target signal is the analog Kara 0K The singing voice plus background music is output to Raju. This is converted to digital using the A / D converter 21a. The selection logic 21b first selects the destination signal. If necessary, it is converted to digital. This signal is then applied to PREM 22. PREM 22 is temporarily offset by The memory 22a is composed of a subtractor 22b. The digital data it receives is a 16-bit positive integer. Each resource is offset by 1 bit to the left in the offset register 22a. Then 1 is subtracted from the offset The data in the register is shifted by a subtractor. Then the data in the offset register is shifted to the right by 1 bit. The data is then shifted to the left by one bit. The processed data is then applied to PPEM 23. The PPEM 23 includes a combination of a syllable assessor 23a and a comparator 22b and a control register 23c. The description of this syllable evaluator and Gold and Rabiner in the Journal of the U.S. Hearing Association 1969 Vo 丨 .46, NO.2 (Part 2), pages 442-448, titled " Parallel Processing Techniques for Estimating Pitch Period of speech in Time Domain ". This system includes a low-pass filter 51 to remove the first resonance region. This low-pass filter shape is processed by the peak and valley quotient 53. Six sets of measurements of peaks and troughs are derived. There are six "single" identical evaluators 55, each operated by one of the six groups of detectors 53. Each evaluator is a peak detection drop circuit. Looking at Figure 4, following each detected pulse, there is a simple exponential decay in the empty interval. Whenever a pulse exceeds the level of the falling circuit (in attenuation), it is detected and the circuit is reset. The fall time constant and the blank time of each detector are a function of evaluating the detector's syllables. The final syllable calculation value is based on the Chinese standard (CNS) 84 specifications (210Χ297 mm) from this paper scale. IIII ϋϋ ———— ^ 1 装 ^ III 定 III .γ # (Please read the notes on the back first (Fill in this page again) 316361 A7 B7 Fifth, the description of the invention (9) Each • Simple " syllable evaluator and majority rule selection are completed to determine the tone check based on the conclusion of the six. The final calculation of the numerical value is carried out in the decision of the creator 57. It can be regarded as a computer with memory, an arithmetic logic calculation and control hardware, operating the incoming signal. At any time t 〇-a syllable's compensation is made by: 1. Form a 6X6 matrix of syllable evaluation. Look at circle 5B. The individual detectors and rows act as syllable evaluations. The first three columns are the three most recent syllable evaluations. Column 4 is the sum of columns 1 and 2; column 5 is the sum of columns 2 and 3; column 6 is the sum of the first 3 columns. The technique for forming this matrix is shown in Figure 5A. The reason for the last three columns of this matrix is that sometimes individual detectors will specify the second or third harmonic wave fundamental wave and will input the last three columns corrected for comparison with the three most recent syllables. 2. Compare each input value in the first column of the matrix to another 35 number values that match the number of the matrix. In particular, Pn (i = l, 2, 3, 4, 5, 6) is the most commonly used (maximum number of matches) syllable used as the final evaluation. Printed by the Employee Consumer Cooperative of the Central Bureau of Standards of the Ministry of Economic Affairs --------- ^ 1 set-(please read the notes on the back and then fill out this page). Call line in order to determine whether the two syllables are evaluated as " compliant " It can be observed that the ratio is better than the difference. However, the ratio measurement may be very close to the need to avoid dividing values. Because in many segments of the speech, there are quite large variables measured in consecutive syllables, it is useful to include several starting values to define " in accordance with '' and then try to select "calculated for each syllable, this The starting point is the most consistent answer. From this description, we now define the calculated value of box 57 in Figure 3. Figure 6 shows 16 tabular columns that fit the width of the hard window. As indicated in Figure 5, only self-giving The closest evaluation syllable in the detector is one •• The selected person " as " 11-This paper scale is applicable to the Chinese National Standard (CNS) A4 specification (210X297 mm) A7 B7 Employee consumption of the Central Standards Bureau of the Ministry of Economic Affairs Cooperative Duprinting 5. Description of invention (10) Final choice. This candidate is one of the six possible choices, which is the "correct" syllable. To determine this "winner", each candidate is selected Do a numerical comparison with all the remaining 35 syllables. This comparison is repeated four times, relative to each row of the circle 6 table. From each row, the appropriate window width is selected and considered as a candidate for relationship evaluation Features. After the combined numbers are tabulated, this number is subtracted by an offset of 1. The second line is then repeated, and this measurement is repeated; this time the window is wider, increasing the likelihood of coincidence, but, in the case of compensation, since The number of edits minus the offset of 2. After repeating this calculation for all four lines in this way, the maximum offset is used as the number of matches, which represents the special syllable evaluation. Now there are 5 remaining candidates The winner repeats all the processes, and the winner chooses the number with the largest matching offset. The evaluation is made every 20 milliseconds and the average of the conclusions is calculated once every 20 milliseconds, which is 50X 10 or 500 in 10 seconds. As an average. This determines the pitch of the sound. This evaluation pitch is compared in the comparator 23b, and check whether the input box is a sound for the tone. If the box is a sound, control the register 23c to record the length of the green box or The time is recorded in seconds in green. The sound data is then written into the DSTM 23, and any non-sound data is discarded by the control register 23c. The DSTM 23 can also be the memory of the information processor 13. The comparator 23b and the control register 23c block maintains cumulative time , According to each box of the sound box, and keep the green syllable in DSTM 23 until the total time is 16 seconds. If this has been completed, the control register 23c sends a signal to the enable / disable living device 26a in the SLEM 26 . Then the activation / deactivation generator 26a sends a signal to ADCM 21. Then select logic 21b to select the reference paper i scale applicable to the national standard (C & S) A4 specification _ (210x297mm) (please read the note on the back first $ Item then fill out this page) 丨 装 _ * • 11
T 經濟部中央標準局員工消費合作社印製 A7 -----^____ 五、發明説明（11 ) 考的歌聲。此過程是持續地參考歌聲在相同的樣子當目的 歌聲持續16秒。此過程之後，在DSTM 25的兩方塊音節資 料是可用的。一個方塊爲目的信號且另一方塊爲參考信號 。然後控制暫存器23c傳送一信號至SLEM 26中的啓動/取 消產生器26a。然後啓動/取消產生器26a如前述傳送信號 到ADCM 21。這時間選擇遲輯21於ADCM 21中選擇背景音樂 且持續檢查，以知道何時背景音樂被完成，即卡拉〇ί[歌聲的 背景音樂被完成。在背景音樂結束時，選擇遲輯21傳送一 信號到SLEM 26中的啓動/取消產生器26a。 藉由SLEM中的啓動/取消產生器26a，自ADCM 21接收最 終信號；SLEM 26中的啓動/取消產生器26a取消ADCM 21 和啓動POEM 24。 此POEM 24校正儲存在DSTM 24中的兩個方塊資料。偏移 暫存器24a首先自DSTM 25讀取第一個方塊資料。然後經由 比較器24h的協助決定方塊資料的最小値。依第一方塊中 決定的實料最小値，比較器24b保持此資料的記綠。然後偏 移暫存器24a自開始處讀取全部的方塊。然後比較器24b比 較由偏移暫存器24a讀取的任一資料，以檢查是否此資料 大於最小記綠資料的兩倍。如果是，此資料於偏移暫存器 24a中右移一個位元。然後新資料被重寫入dstjj 25中舊資 料相同的位置。此過程重覆於第二方塊的資料。然後p〇EM 24傳回一信號至SLEM 26中的啓動/取消產生器26a。 SLEM 26由啓動/取消產生器26a，計分評價器26b，水 平評價器26c組合而成。依接收P0EM 24中的偏移暫存器 -13 - 本紙張U迺用中國國家標準（CNS)从胁（21(}><297公瘦） (請先閲讀背面之注意事項再填寫本頁) -裝· 訂 年 316961 A7 B7 五、發明説明（12 ) 24a的信號，SLEM 26中的啓動/取消產生器263啓動計分 評償器26b。然後計分評價器26b讀取儲存在DSTM 25中的 目的信號和參考信號的音節。然後計算目的信號對於參考 信號的百分比變化量。此評償的百分比變化量错存在Dstjj 25。然後傳送一信號到啓動/取消產生器26a。啓動^取 消產生器26a啓動水平評償器26c。水平評償器26c讀取错 存在DSTM 25中的百分比變化量，同時評價其水平和顯示 於 SLDM 15 上。 DSTM可爲RAM或碟片記憶。也可爲信息處理器13上的晶 元或外在晶元記憶。 ------I--f ·装------Ί ^ I,------(.^ (請先S讀背面之注意事項再填寫本頁) 經濟部中央橾準局員工消費合作社印製 本紙張尺度適用中國國家標隼（CNS ) A4規格（210X297公釐）T Printed by the Employee Consumer Cooperative of the Central Bureau of Standards of the Ministry of Economy A7 ----- ^ ____ 5. The description of the invention (11) The singing voice of the exam. This process is to continuously refer to the singing voice in the same way as the purpose. The singing voice lasts 16 seconds. After this process, two square syllable data in DSTM 25 is available. One block is the destination signal and the other block is the reference signal. The control register 23c then sends a signal to the activation / deactivation generator 26a in the SLEM 26. The activation / deactivation generator 26a then transmits a signal to the ADCM 21 as described above. At this time, select the late album 21 to select the background music in the ADCM 21 and continue to check to know when the background music is completed, that is, the background music of the karaoke song. At the end of the background music, the selection delay 21 sends a signal to the activation / deactivation generator 26a in the SLEM 26. The final signal is received from the ADCM 21 by the activation / deactivation generator 26a in SLEM; the activation / deactivation generator 26a in SLEM 26 cancels the ADCM 21 and activates the POEM 24. This POEM 24 corrects the two square data stored in DSTM 24. The offset register 24a first reads the first block data from the DSTM 25. Then, with the help of the comparator 24h, the minimum value of the block data is determined. According to the minimum value of the actual material determined in the first box, the comparator 24b keeps the green mark of this data. Then the offset register 24a reads all the blocks from the beginning. The comparator 24b then compares any data read from the offset register 24a to check whether the data is greater than twice the minimum green data. If so, the data is shifted to the right by one bit in the offset register 24a. Then the new data is rewritten to the same location as the old data in dstjj 25. This process repeats the data in the second box. The POEM 24 then returns a signal to the activation / deactivation generator 26a in the SLEM 26. The SLEM 26 is composed of a start / cancel generator 26a, a score evaluator 26b, and a horizontal evaluator 26c. According to the offset register -13 in the received P0EM 24-This paper uses the Chinese National Standard (CNS) from the threat (21 (} > < 297g)) (Please read the precautions on the back before filling in this Page)-Installation · Yearbook 316961 A7 B7 V. Description of the invention (12) The signal of 24a, the start / cancel generator 263 in SLEM 26 activates the scoring evaluator 26b. Then the scoring evaluator 26b reads and stores in DSTM The syllables of the destination signal and the reference signal in 25. Then calculate the percentage change of the destination signal with respect to the reference signal. The percentage change of this compensation error exists in Dstjj 25. Then send a signal to the start / cancel generator 26a. Start ^ Cancel The generator 26a activates the level evaluator 26c. The level evaluator 26c reads the percentage change in the DSTM 25, and evaluates its level and displays it on the SLDM 15. The DSTM can be RAM or disc memory. It can also be Memory of the wafer or external wafer on the information processor 13. ------ I--f · install ------ Ί ^ I, ------ (. ^ (Please first Read the precautions on the back and then fill out this page) The paper printed by the Consumer Cooperative of the Central Provincial Bureau of the Ministry of Economic Affairs The size of the paper is applicable to China Home Standard Falcon (CNS) A4 size (210X297 mm)