TW407402B - Method for sub frequency band encoding/decoding and apparatus thereof - Google Patents

Method for sub frequency band encoding/decoding and apparatus thereof Download PDF

Info

Publication number
TW407402B
TW407402B TW88102911A TW88102911A TW407402B TW 407402 B TW407402 B TW 407402B TW 88102911 A TW88102911 A TW 88102911A TW 88102911 A TW88102911 A TW 88102911A TW 407402 B TW407402 B TW 407402B
Authority
TW
Taiwan
Prior art keywords
sub
band
encoding
decoding
bit allocation
Prior art date
Application number
TW88102911A
Other languages
Chinese (zh)
Inventor
Wen-Yuan Chen
Original Assignee
Winbond Electronics Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Winbond Electronics Corp filed Critical Winbond Electronics Corp
Priority to TW88102911A priority Critical patent/TW407402B/en
Application granted granted Critical
Publication of TW407402B publication Critical patent/TW407402B/en

Links

Landscapes

  • Compression, Expansion, Code Conversion, And Decoders (AREA)

Abstract

The invention provides a method for sub frequency band encoding/decoding and apparatus thereof By the property that the variation of the difference of the adjacent ratio factors is smaller than that of the value of the respectively discrete ratio factors, a specific encoding/decoding method is used to encode the ratio factor of sub frequency band to substantially reduce the bit rate of the digitalized real timbre.

Description

4224twf.doc/006 407402 A7 B7 經濟部中央橾準局貝工消费合作社印製 五、發明説明(/ ) 本發明是有關於一種語音壓縮方法及裝置,且特別 是有關於一種次頻帶編解碼之方法及裝置。 基於人類聽覺系統之次頻帶編解碼器(sub-band coder),通常被用來壓縮一寬頻多樣的語音輸入信號,如 吵鬧的交談、眾多的說話者及音樂。次頻帶編解碼的基本 原則爲’將輸入語音頻譜劃分成許多頻帶,然後這些頻率 頻帶會被分別編碼(encode)。通常是利用一濾波器組(filter bank)來分離§吾音輸入柄號。在語音頻譜被分割成爲n個 頻帶之後,每一頻帶被依據Nyquist原理(N個中取1個)再 取樣、正規化(normalized)、量化(quantized)、編碼、多工 化(multiplexed)及傳送。再者,以聽覺模型(pSych〇acoustic model)建立一組數據,用以控制量化裝置(quantizer)與編 碼。 在習知技藝中,正規化程序係以找出一比例因數 (scalefactor)來調整這些次頻帶樣本的振幅。此比例因素的 編解碼係藉由一預定資料表(pre-determined table)查表而得 一指數(index),其中此指數的對應値最接近此比例因數。 下表 1 顯示 MPEG1/語音層 I 與 II(MPEGl/Audio Layer 1,11) 所使用的資料表。此指數以6位元來表示,並且唯有位元 中非零的數字被分配給次頻帶此指數才會被傳送。 請 先 閱- 讀 背 i- 填 訂 線 本紙張尺度適用中國國家操準(CNS ) A4蛛格(2丨〇><297公釐) 4224twf.doc/006 五、發明説明(2) 407402 A7 B7 經濟部中央標準局員工消費合作社印製 (表1) Ind ex Scalefacto r ind ex Scalefacto r ind ex scalefactor ind ex Scalefacto r 0 2.0000000 0000000 1 1.5874010 5196820 2 1.2599210 4989488 3 1.0000000 0000001 4 0.7937005 2598411 5 0.6299605 2494744 6 0.5000000 0000001 7 0.3968502 6299206 8 0.3149802 6247372 9 0.2500000 0000001 10 0.1984251 3149603 11 0.1574901 3123686 12 0.1250000 0000000 13 0.0992125 6574802 14 0.0787450 6561843 15 0.0625000 0000000 16 0.0496062 8287401 17 0.0393725 3280922 18 0.0312500 0000000 19 0.0248031 4143700 20 0.0196862 6640461 21 0.0156250 0000000 22 0.0124015 7071850 23 0.0098431 3320230 24 0.0078125 0000000 25 0.0062007 8535925 26 0.0049215 6660115 27 0.0039062 5000000 28 0.0031003 9267963 29 0.0024607 8330058 30 0.0019531 2500000 31 0.0015501 9633981 32 0.0012303 9165029 33 0.0009765 6250000 34 0.0007750 9816991 35 0.0006151 9582541 36 0.0004882 8125000 37 0.0003875 4908495 38 0.0003075 9791257 39 0.0002441 4062500 40 0.0001937 7454248 41 0.0001537 9895629 42 0.0001220 7031250 43 0.0000968 8727124 48 0.0000305 1757813 49 0.0000242 2181781 50 0.0000192 2486954 51 0.0000152 5878906 52 0.0000121 1090890 53 0.0000096 1243477 54 0.0000076 2939453 55 0.0000060 5545445 56 0.0000048 0621738 57 0.0000038 1469727 58 0.0000030 2772723 59 0.0000024 0310869 60 0.0000019 0734863 61 0.0000015 1386361 62 0.0000012 0155435 63 4 -----;----襄-- (請先Ba·讀背面之注意事产.填寫本頁)4224twf.doc / 006 407402 A7 B7 Printed by the Central Laboratories Bureau of the Ministry of Economic Affairs, Shellfish Consumer Cooperative, V. Description of the invention (/) The present invention relates to a method and device for speech compression, and more particularly to a sub-band encoding and decoding method. Method and device. A sub-band coder based on the human auditory system is usually used to compress a wide variety of voice input signals, such as loud conversations, many speakers, and music. The basic principle of subband coding and decoding is to 'divide the input speech spectrum into a number of frequency bands, and these frequency bands are then individually encoded. A filter bank is usually used to separate the vowel input handles. After the speech spectrum is divided into n frequency bands, each frequency band is resampled, normalized, quantized, encoded, multiplexed, and transmitted according to the Nyquist principle (1 of N). . Furthermore, a set of data is established by using an acoustic model to control the quantizer and encoding. In the conventional art, the normalization procedure is to find a scale factor to adjust the amplitude of these sub-band samples. The encoding and decoding of the scale factor is obtained by looking up a table in a pre-determined table, and an index is obtained, wherein the corresponding index of the index is closest to the scale factor. Table 1 below shows the data sheets used by MPEG1 / Voice Layer I and II (MPEGl / Audio Layer 1, 11). The index is expressed in 6 bits, and the index is transmitted only if a non-zero number in the bit is assigned to the sub-band. Please read first-Read back i- Filling line The paper size is applicable to China National Standards (CNS) A4 spider (2 丨 〇 > < 297 mm) 4224twf.doc / 006 V. Description of the invention (2) 407402 A7 B7 Printed by the Consumer Cooperative of the Central Bureau of Standards of the Ministry of Economic Affairs (Table 1) Ind ex Scalefacto r ind ex Scalefacto r ind ex scalefactor ind ex Scalefacto r 0 2.0000000 0000000 1 1.5874010 5196820 2 1.2599210 4989488 3 1.0000000 0000001 4 0.7937005 2598411 5 0.6299605 2494744 6 0.5000000 0000001 7 0.3968502 6299206 8 0.3149802 6247372 9 0.2500000 0000001 10 0.1984251 3149603 11 0.1574901 3123686 12 0.1250000 0000000 13 0.0992125 6574802 14 0.0787450 6561843 15 0.0625000 0000000 16 0.0496062 8287401 17 0.0393725 3280922 18 0.0312500 0000000 19 0.0248031 4143700 20 23 0.0098431 3320230 24 0.0078125 0000000 25 0.0062007 8535925 26 0.0049215 6660115 27 0.0039062 5000000 28 0.0031003 9267963 29 0.0024607 8330058 30 0.0019531 2500000 31 0.0015501 9633981 32 0.0012303 9165029 33 0.0009765 6250000 34 0.0007750 9816991 35 0.0006151 9582541 36 0.0004882 8125000 37 0.0003875 4908495 38 0.0003075 9791257 39 0.0002441 4062500 40 0.0001937 7454248 41 0.0001537 9895629 42 0.0001220 7031250 43 0.0000968 80.0090 52 0.0090 2901878906 50 0.0000305 1757813 49 0.0000242 2181878906 50 0.00 0.0000096 1243477 54 0.0000076 2939453 55 0.0000060 5545445 56 0.0000048 0621738 57 0.0000038 1469727 58 0.0000030 2772723 59 0.0000024 0310869 60 0.0000019 0734863 61 0.0000015 1386361 62 0.0000012 0155435 63 4 -----; -------- Please read Ba (Notes on the back. Fill out this page)

、1T |線- 本紙張尺度適用中國國家樣準(CNS ) Α4祝格(210X297公釐) 4224twf.doc/006 407402 經濟部中央標準局員工消費合作杜印製 發明説明(多) 然而,如此之編解碼模式並沒有利用到相異樣本之 間的相關性。另一方面,爲大家所熟知的,自適性差分脈 衝碼調變(adapnve differential PulSe CQde m()dulat_,aDPCM) f法Ulgonthm)可以達成每樣本4位元(4 blts/samples)之位 元率的高音質(quality)數位化。由於比例因數的特性,相 鄰比例因數的差之變動量小於個別比例因數的數値。因 此,一種編解碼模式,其基於籩化差之變動量的槪念,可 以將一數位化實際音質之位元率大幅減低。 有鑑於此,本發明提供〜種次頻帶編解碼之方法及 裝置,利用相鄰比例因數的差之變動量小於個別比例因數 的數値之特性,以一特定編解碼方法將次頻帶的比例因數 編碼。 本發明提供一種次頻帶編解碼之方法及裝置,利用 ADPCM方法編碼次頻帶的比例因數,大幅減低數位化實 際音質之位元率。 本發明提出一種次頻帶編解碼方法。將一數位語音 信號進行次頻帶分析’產生複數個次頻帶。計算次頻帶相 對應之複數個比例因數。編解碼比例因數,依據一特定編 碼方法。將數位語音信號進行聽覺模型分析。將數位語音 信號進行位元分配,依據聽覺模型分析之資訊及一預定的 位元率,分配複數個次頻帶樣本所需的位元數目。依據位 元分配之資訊,量化及編解碼次頻帶。編解碼位元分配之 資訊。格式化被量化及編解碼之次頻帶、被編解碼後的比 例因數及位元分配之資訊。因爲每一次頻帶的比例因數經 本紙張尺度適用中國國家標準(CNS ) A4祝格(210X297公釐) 請 先 閲· 讀 r 填 聚裝 頁 訂 線 4224twf,d〇c/006 407402 A7 B7 經濟部中央標準局員工消費合作社印製 五、發明説明(¥ ) 由ADPCM方法來編解碼,所以可以大幅減低數位化實際 音質之位元率。 本發明提出一種次頻帶編解碼之裝置,包括一次頻 帶分析裝置、一計算裝置、一編解碼裝置、一聽覺模型分 析裝置、一位元分配裝置、一量化裝置、一位元分配編解 碼裝置及一格式化裝置。次頻帶分析裝置,用以將一數位 語音信號進行次頻帶分析,產生複數個次頻帶。計算裝置, 用以計算次頻帶相對應之複數個比例因數。編解碼裝置, 用以編解碼比例因數’依據自適性差分脈衝碼調變方法。 聽覺模型分析裝置’用以將數位語音信號進行聽覺模型分 析。位元分配裝置,用以將數位語音信號進行位元分配, 依據聽覺模型分析之資訊及一預定的位元率,分配複數個 次頻帶樣本所需的位元數目。量化裝置,用以依據位元分 配之資訊’量化及編解碼次頻帶。位元分配編解碼裝置, 用以編解碼位元分配之資訊。格式化裝置,用以格式化被 量化及編解碼之次頻帶、被編解碼後的比例因數及位元分 配之資訊。因爲編解碼裝置的作用,次頻帶的比例因數可 依據ADPCM方法來編解碼,所以可以大幅減低數位化實 際音質之位元率。 爲讓本發明之上述和其他目的、特徵、和優點能更 明顯易懂,下文特舉一較佳實施例,並配合所附圖式,作 詳細說明如下: 圖式之簡單說明: 第1圖是一方法流程圖,敘述依照本發明一較佳實施 6 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公釐) ---1 Ί-----^-- (請先M.讀背面之注意事Js-ι,填寫本頁) 訂 407402 4224twf.doc/〇〇6、 1T | Line-This paper size is applicable to China National Standard (CNS) Α4 Zhuge (210X297mm) 4224twf.doc / 006 407402 Employees' cooperation cooperation with the Central Standards Bureau of the Ministry of Economic Affairs Du printed invention description (multiple) However, this is the case The codec mode does not take advantage of the correlation between disparate samples. On the other hand, as everyone knows, adaptive differential pulse code modulation (adaptive differential PulSe CQde m () dulat_, aDPCM) f method Ulgonthm) can achieve a bit rate of 4 bits per sample (4 blts / samples) High-quality digitalization. Due to the nature of the proportionality factor, the amount of change in the difference between adjacent proportionality factors is less than the number of individual proportionality factors. Therefore, a codec mode, which is based on the idea of the amount of variation of the conversion difference, can greatly reduce the bit rate of a digitized actual sound quality. In view of this, the present invention provides a method and device for encoding and decoding sub-bands, utilizing the characteristic that the variation of the difference between adjacent scaling factors is less than the number of individual scaling factors, and using a specific encoding and decoding method to scale the sub-bands. coding. The invention provides a method and a device for encoding and decoding a sub-band. The ADPCM method is used to encode a sub-band proportionality factor, which greatly reduces the bit rate of digitalized actual sound quality. The invention proposes a sub-band encoding and decoding method. Sub-band analysis of a digital speech signal 'produces a plurality of sub-bands. Calculate the multiple scale factors corresponding to the sub-bands. The codec scale factor is based on a specific coding method. Auditory model analysis of digital speech signals. The digital speech signal is bit-allocated, and the number of bits required for the plurality of sub-band samples is allocated based on the information analyzed by the auditory model and a predetermined bit rate. The subbands are quantized and coded based on the bit allocation information. Information on codec bit allocation. Format the quantized and coded sub-band, the coded scale factor and bit allocation information. Because the scale factor of each frequency band is subject to the Chinese National Standard (CNS) A4 (210X297 mm) according to this paper scale. Please read · Read r Filler binding line 4224twf, doc / 006 407402 A7 B7 Central Ministry of Economic Affairs Printed by the Consumer Bureau of the Standards Bureau 5. The invention description (¥) is encoded and decoded by the ADPCM method, so the bit rate of the actual digital sound quality can be greatly reduced. The invention provides a sub-band encoding and decoding device, which includes a primary band analysis device, a computing device, a codec device, an auditory model analysis device, a bit allocation device, a quantization device, a bit allocation codec device, and A formatting device. The sub-band analysis device is used for performing sub-band analysis on a digital speech signal to generate a plurality of sub-bands. A computing device is used to calculate a plurality of scaling factors corresponding to the sub-bands. A codec device for encoding and decoding a scaling factor 'according to an adaptive differential pulse code modulation method. The auditory model analysis device 'is used for analyzing auditory models of digital speech signals. A bit allocation device is used for bit allocation of digital speech signals, and allocates the number of bits required for a plurality of sub-band samples based on the information analyzed by the auditory model and a predetermined bit rate. A quantization device is used to quantize and encode and decode the sub-band based on the bit-allocated information '. A bit allocation codec device is used to encode and decode bit allocation information. A formatting device is used to format the quantized and coded sub-band, the coded scale factor, and the bit allocation information. Because of the function of the encoding and decoding device, the scale factor of the sub-band can be encoded and decoded according to the ADPCM method, so the bit rate of the actual digital sound quality can be greatly reduced. In order to make the above and other objects, features, and advantages of the present invention more comprehensible, a preferred embodiment is given below in conjunction with the accompanying drawings for detailed description as follows: Brief description of the drawings: FIG. It is a method flow chart describing a preferred implementation according to the present invention. 6 This paper size applies the Chinese National Standard (CNS) A4 specification (210X297 mm) --- 1 -1 ----- ^-(please first M. (Read the note on the back Js-ι, fill out this page) Order 407402 4224twf.doc / 〇〇6

五、發明説明(爻) 經濟部中央標隼局員工消費合作社印製 例之次頻帶編解碼裝置; 第2圖是一電路方塊圖,說明依照本發明一較佳實施 例之一種次頻帶編解碼之裝置。 圖式之標記說明: _ 28 :數位語音信號 30 :次頻帶分析裝置 32 :計算裝置 34:編碼裝置 36 :聽覺模型分析裝置 38 :位元分配裝置 40 :量化裝置 42 :位元分配編碼裝置 44 :格式化裝置 較佳實施例 本發明提出一種特定編碼方法,特別是以ADPCM方 法來進行比例因數的編碼(coding)。依照第1圖所示,敘述 本發明之次頻帶編解碼器的方法流程圖。輸入脈衝碼調變 (pulse code modulation, PCM)樣本的取樣率例如爲 8 KHz ’ 而且每一輸入樣本係以16位元表示。 次頻帶解析 解析次頻帶之濾波器組被使用來將8 KHz取樣頻率 之輸入信號分離,而成爲8個頻寬0.5 KHz等距之次頻帶。 解析次頻帶濾波程序係以濾波器(quadrature mirror filters, QMFs)來實現。利用濾波器,最初將原始的輸入信號濾波 請 先 閱 讀 背V. Description of the invention (ii) Sub-band coding and decoding device printed by the employee's cooperative of the Central Bureau of Standards of the Ministry of Economic Affairs; Figure 2 is a circuit block diagram illustrating a sub-band coding and decoding according to a preferred embodiment of the present invention; Of the device. Explanation of the drawing symbols: _ 28: digital voice signal 30: sub-band analysis device 32: calculation device 34: encoding device 36: auditory model analysis device 38: bit allocation device 40: quantization device 42: bit allocation encoding device 44 : A preferred embodiment of a formatting device The present invention proposes a specific encoding method, in particular, the ADPCM method is used for encoding the scaling factor. A flowchart of a method for a sub-band codec according to the present invention will be described with reference to FIG. The sampling rate of the input pulse code modulation (PCM) samples is, for example, 8 KHz ′, and each input sample is represented by 16 bits. Sub-band analysis The filter bank that analyzes the sub-band is used to separate the input signal of the 8 KHz sampling frequency into 8 sub-bands of 0.5 KHz equidistance. The analysis sub-band filtering program is implemented by quadrature mirror filters (QMFs). Use filters to filter the original input signal initially.

I I裝 .訂 線 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公嫠) 經濟部中央標準局員工消費合作社印製 407402 4224twf.doc/006 A7 ______B7 五、發明説明(ό ) 成兩個頻帶,一低頻帶(low band)與一高頻帶(high band)。 高頻帶的取樣頻率下降2倍,其包含一半的頻譜,代表爐 波器組的最高頻帶(/ 2€ω<)。同樣地,低頻帶的取樣頻 率一樣下降2倍,並且高頻帶與低頻帶兩者皆被送入一第 二濾波階段,每一輸入信號再被九十度相位差的鏡射濾波 器分割成兩個相等的頻帶。此濾波程序以三階段重複來得 到8頻帶輸出。在每一頻帶中,16個次頻帶樣本被組合來 處理。下表2列出本發明較佳實施例所使用之濾波器的16 個係數。 (表 2)_ H[0]=H[15]=0.10501670e-2 H[l]=H[14]=-0.50545260e-2 H[2]=H[13]=-0.25897560-2 H[3]=H[12]=0.27641400e-l H[4]=H[ll]=-0.96663760e-2 H[5]=H[10]=-0.9Q392230e-l H[6]=H[9]=0.97798170e-l H[7]=H[8]=0.48102840e0___ 計算比例因數 每一頻帶的比例因素計算係以每16個次頻帶進行 之。並且這些16個樣本的絕對値之最大値被決定,例如 最大値爲256。 8 — — 11 — 裝—— 1 -iJT線 (請先閲讀背面之注意事笋_填寫本頁) 本紙張尺度適用中國國家標準(CNS ) A4規格(210 X297公釐) 407402 4224twf.doc/006 A7 B7 經濟部中央標準局貝工消費合作社印製 五、發明説明(q) 比例因數的編解碼 以每樣本4位元(4 bits/sample)之ADPCM方法將比例 因素編碼。唯有位元中一非零數値被分配至次頻帶,比例 因素才被傳送。因爲每—次頻帶的比例因數經由ADPCM 方法來編解碼,所以可以大幅減低數位化實際音質之位元 率。例如相較於習知每秒64千位元(64 kbps)的位元率,此 方法可以將位元率降低至每秒只需16千位元。 聽覺模型分析 聽覺模型計算濾波器組中任一頻帶的人耳能感覺的 聲音位準(noise level)。此聲音位準被用於位元分配(bit allocation) ’決定實際的量化裝置(quantizer)與量化裝置位 準。本發明一實施例運用一種MPEG/Audio的聽覺模型。 此模型的最終輸出爲每一頻帶的信號遮罩比(signal t0 mask ratio, SMR)。 位元分配 分配裝置(allocator)依據濾波器組的輸出樣本與聽覺 模型的SMR値’調整位元分配,以符合位元率之要求。 在調整至固定位元率之前,次頻帶樣本之編解碼所用之位 元數目必須先決定。上述的位元數目”adb”係由可用的位元 總數目”cb”減去位元分配所需的位元數目” bbal”及比例因 數所需的位元數目”bbcf”(adb=cb-bbal-bbcf)。此位元結果數 目”adb”被用來編碼次頻帶樣本。此位元分配的原則是,在 可用位元數目下,要使得整個雜訊的遮罩比最小。分配給 一個樣本的可能位元數目可由下表3査得;其數値爲〇, 3,4, 9 本紙张尺度適用中國國家標準(CNS ) A4規格(210X297公釐) Λ 身 (請先閱讀背面之注意事^/¼寫本頁) -a 407402 4224twf.doc/006 A7 B7 五、發明説明(公 5 ° (表3) 碼(CODE) 位元(BITS) “00” 0 “01” 3 “10” 4 “1Γ 5 經濟部中央標準局員工消費合作社印製 暈化及編解碼次頻帶樣本 每一次頻帶樣本被正規化,以符合比例因數方式來 分割次頻帶樣本的數値,獲得一新樣本;並且,以N個位 元來量化次頻帶樣本,其中N代表編碼次頻帶樣本的所需 位元數目。 位元分配的編解碼 位元分配的編解碼所依據的位元分配之二位元碼(2-bit code),例如上表3所示。 格式化及傳送 位元串(bitstream)的格式化規格包括有位元分配資 訊、比例因數及次頻帶樣本。將資訊格式化之後,再將格 式化後的資訊傳送。 接下來,請參照第2圖,其說明依照本發明一較佳實 施例之一種次頻帶編解碼之裝置。 依照本發明較佳實施例之次頻帶編解碼之裝置包括 一次頻帶分析裝置30,用以將一數位語音信號28進行次 10 ----:-------^--------II;------^ (請先閲讀背面之注意事項广^寫本頁) 本紙張尺度適用中國國家標準(CNS ) A4规格(210X297公釐) 407402 4224twf.doc/006 A7 __B7 五、發明説明(巧) 頻帶分析’產生複數個次頻帶;一計算裝置32,用以計算 上述次頻帶相對應之複數個比例因數;一編解碼裝置34, 用以編解碼上述比例因數,依據自適性差分脈衝碼調變方 法:一聽覺模型分析裝置36 ’用以將上述數位語音信號28 進行聽覺模型分析;一位元分配裝置38,用以將上述數位 語音信號進行位元分配’依據上述聽覺模型分析之資訊及 一預定的位元率,分配複數個次頻帶樣本所需的位元數 目;一量化裝置40 ’用以量化及編解碼上述次頻帶,依據 上述位元分配之資訊;一位元分配編解碼裝置42,用以編 解碼上述位元分配之資訊;以及,一格式化裝置44,用以 格式化被量化及編解碼之上述次頻帶、被編解碼後的上述 比例因數及上述位元分配之資訊。藉由上述本發明之次頻 帶編解碼之裝置,可以大幅減低數位化實際音質之位元 率。 雖然本發明已以較佳實施例揭露如上,然其並非用 以限定本發明,任何熟習此技藝者,在不脫離本發明之精 神和範圍內,當可作各種之更動與潤飾,因此本發明之保 護範圍當視後附之申請專利範圍所界定者爲準。 --:--Z-----¾II (請先閱讀背面之注意事f-k寫本頁)II. Binding line. The paper size is applicable to Chinese National Standard (CNS) A4 (210X297). Printed by the Staff Consumer Cooperative of the Central Standards Bureau of the Ministry of Economic Affairs. 407402 4224twf.doc / 006 A7 ______B7 5. Description of the Invention Frequency band, a low band and a high band. The sampling frequency of the high frequency band is reduced by a factor of two, which contains half of the frequency spectrum, representing the highest frequency band of the furnace group (/ 2 € ω <). Similarly, the sampling frequency in the low frequency band is also reduced by 2 times, and both the high frequency band and the low frequency band are sent to a second filtering stage. Each input signal is divided into two by a 90-degree phase difference mirror filter. Equal bands. This filtering procedure is repeated in three stages to obtain an 8-band output. In each band, 16 sub-band samples are combined for processing. Table 2 below lists the 16 coefficients of the filter used in the preferred embodiment of the present invention. (Table 2) _ H [0] = H [15] = 0.10501670e-2 H [l] = H [14] =-0.50545260e-2 H [2] = H [13] =-0.25897560-2 H [ 3] = H [12] = 0.27641400el H [4] = H [ll] =-0.96663760e-2 H [5] = H [10] =-0.9Q392230e-l H [6] = H [9] = 0.97798170el H [7] = H [8] = 0.48102840e0 ___ Calculating the scaling factor The scaling factor calculation for each frequency band is performed every 16 sub-bands. And the absolute maximum of these 16 samples is determined, for example, the maximum is 256. 8 — — 11 — Install — 1-iJT line (please read the notes on the back _ fill in this page) This paper size is applicable to China National Standard (CNS) A4 specification (210 X297 mm) 407402 4224twf.doc / 006 A7 B7 Printed by the Shellfish Consumer Cooperative of the Central Standards Bureau of the Ministry of Economic Affairs. 5. Description of the Invention (q) The encoding and decoding of the scale factor uses the ADPCM method of 4 bits / sample to encode the scale factor. The proportionality factor is transmitted only if a non-zero number of bits in the bit is allocated to the sub-band. Because the scale factor of each sub-band is encoded and decoded by the ADPCM method, the bit rate of the digitized actual sound quality can be greatly reduced. For example, compared with the conventional bit rate of 64 kilobits per second (64 kbps), this method can reduce the bit rate to only 16 kilobits per second. Auditory Model Analysis The auditory model calculates the noise level that the human ear can perceive at any frequency band in the filter bank. This sound level is used for bit allocation 'to determine the actual quantizer and quantizer level. An embodiment of the present invention uses an MPEG / Audio hearing model. The final output of this model is the signal t0 mask ratio (SMR) for each frequency band. The bit allocation device (allocator) adjusts the bit allocation according to the output samples of the filter bank and the SMR 値 'of the auditory model to meet the bit rate requirements. Before adjusting to a fixed bit rate, the number of bits used for encoding and decoding the subband samples must be determined. The above-mentioned number of bits "adb" is the total number of available bits "cb" minus the number of bits "bbal" required for bit allocation and the number of bits "bbcf" (adb = cb- bbal-bbcf). This bit result number "adb" is used to encode sub-band samples. The principle of this bit allocation is to minimize the mask ratio of the whole noise under the number of available bits. The number of possible bits allocated to a sample can be found in Table 3 below; the number 〇 is 0, 3, 4, 9 This paper size applies to the Chinese National Standard (CNS) A4 specification (210X297 mm) Λ body (please read first Note on the back ^ / ¼Write this page) -a 407402 4224twf.doc / 006 A7 B7 V. Description of the invention (5 ° (Table 3) Code (BIT) Bits (00) 0 “01” 3 “10” 4 “1Γ 5 Printed by the Consumer Cooperatives of the Central Standards Bureau of the Ministry of Economic Affairs and coding and decoding sub-band samples. Each time the band samples are normalized, and the number of sub-band samples is divided in accordance with the scale factor method to obtain a new one. And quantizing the subband samples with N bits, where N represents the number of bits required to encode the subband samples. Bit allocation codec Bit allocation The two bits of the bit allocation on which the codec is based 2-bit code, as shown in Table 3. Format specifications for formatting and transmitting bitstreams include bit allocation information, scale factors, and subband samples. After formatting the information, Send the formatted information again. Please refer to FIG. 2, which illustrates a subband encoding and decoding device according to a preferred embodiment of the present invention. A subband encoding and decoding device according to a preferred embodiment of the present invention includes a primary band analysis device 30 for A digital voice signal 28 times 10 ----: ------- ^ -------- II; ------ ^ (Please read the precautions on the back first. Page) This paper size is in accordance with Chinese National Standard (CNS) A4 specification (210X297 mm) 407402 4224twf.doc / 006 A7 __B7 V. Description of the invention (smart) Band analysis' produces multiple sub-bands; a computing device 32 is used to Calculate a plurality of scale factors corresponding to the above-mentioned sub-bands; a codec device 34 for encoding and decoding the above-mentioned scale factors, according to an adaptive differential pulse code modulation method: an auditory model analysis device 36 'for converting the above-mentioned digital voice signal 28 for auditory model analysis; a bit allocation device 38 for bit allocation of the above-mentioned digital speech signal 'according to the information of the above-mentioned auditory model analysis and a predetermined bit rate, allocating bits required for a plurality of sub-band samples Yuan number; one A digitizing device 40 ′ is used to quantize and encode the above-mentioned sub-bands according to the above-mentioned bit allocation information; a one-bit allocation codec device 42 is used to encode and decode the above-mentioned bit allocation information; and a formatting device 44, It is used to format the quantized and coded sub-band, the scale factor after coding and decoding, and the bit allocation information. With the above-mentioned sub-band coding and decoding device of the present invention, the actual digital audio quality can be greatly reduced. Bit rate. Although the present invention has been disclosed as above with a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make various modifications and retouches without departing from the spirit and scope of the present invention. Therefore, the present invention The scope of protection shall be determined by the scope of the attached patent application. -: --Z ----- ¾II (Please read the note on the back f-k first to write this page)

JT 線 經濟部中央標準局貝工消費合作社印製 本紙張尺度適用中國國家棣準(CNS ) A4規格(21〇x297公釐)Printed on JT Line Printed by Shellfish Consumer Cooperative, Central Bureau of Standards, Ministry of Economic Affairs This paper is sized for China National Standards (CNS) A4 (21 × 297 mm)

Claims (1)

經濟部中央標準局員工消費合作社印裝 407402 六、申請專利範圍 1. 一種次頻帶編解碼方法,包括: 將一數位語音信號進行次頻帶分析,產生複數個次 頻帶; 計算該些次頻帶相對應之複數個比例因數; 編解碼該些比例因數,依據一特定編碼方法; 將該數位語音信號進行聽覺模型分析; 將該數位語音信號進行位元分配,依據該聽覺模型 分析之資訊及一預定的位元率,分配該些次頻帶樣本所需 的位元數目; 量化及編解碼該些次頻帶,依據該位元分配之資訊; 編解碼該位元分配之資訊;以及 格式化被量化及編解碼之該些次頻帶、被編解碼後 的該些比例因數及該位元分配之資訊。 2. 如申請專利範圍第1項所述之次頻帶編解碼方 法,其中,該數位語音信號包括一脈衝碼調變信號。 3. 如申請專利範圍第1項所述之次頻帶編解碼方 法,其中,該特定編碼方法包括一自適性差分脈衝碼調變 方法。 4. 如申請專利範圍第1項所述之次頻帶編解碼方 法,其中,該次頻帶分析包括以一濾波器組將該數位語音 信號分割成該些次頻帶。 5. 如申請專利範圍第4項所述之次頻帶編解碼方 法,其中,該聽覺模型分析計算濾波器組中任一頻帶的一 合理顯著的聲音位準,輸出爲每一頻帶的信號遮罩比。 本紙張尺度適用中國國家標準(CNS ) A4規格(210X297公釐) ------裝---r--:丨訂--^----線 (請先閲讀背面之注意事填寫本頁) ABCD 401402 4224twf.doc/006 六、申請專利範圍 6. 如申請專利範圍第5項所述之次頻帶編解碼方 法,其中,該位元分配程序係依據該濾波器組的輸出樣本 與該聽覺模型分析的信號遮罩比,調整位元分配,以符合 位元率之要求。 7. 如申請專利範圍第5項所述之次頻帶編解碼方 法,更包括傳送格式化後的資訊。 8. —種次頻帶編解碼之裝置,包括: 一次頻帶分析裝置,用以將一數位語音信號進行次 頻帶分析,產生複數個次頻帶; 一計算裝置,用以計算該些次頻帶相對應之複數個 比例因數; 一編解碼裝置,用以編解碼該些比例因數,依據自 適性差分脈衝碼調變方法; 一聽覺模型分析裝置,用以將該數位語音信號進行 聽覺模型分析; 一位元分配裝置,用以將該數位語音信號進行位元 分配,依據該聽覺模型分析之資訊及一預定的位元率,分 配該些次頻帶樣本所需的位元數目; 一量化裝置,用以量化及編解碼該些次頻帶,依據 該位元分配之資訊; 一位元分配編解碼裝置,用以編解碼該位元分配之 資訊;以及 一格式化裝置,用以格式化被量化及編解碼之該些 次頻帶、被編解碼後的該些比例因數及該位元分配之資 訊。 :---τ-----裝-----^丨订-------線 (請先閲讀背面之注意事^Λ填寫本頁) 經濟部中央標準局員工消費合作社印製 本紙張尺度適用中國國家標準(CNS ) Α4規格(210X297公嫠)Printed by the Consumers' Cooperative of the Central Standards Bureau of the Ministry of Economic Affairs 407402 VI. Patent Application Scope 1. A sub-band encoding and decoding method includes: performing a sub-band analysis on a digital voice signal to generate a plurality of sub-bands; calculating the corresponding sub-bands A plurality of scaling factors; encoding and decoding the scaling factors, according to a specific encoding method; analyzing the digital speech signal by an auditory model; performing bit allocation of the digital speech signal, according to the information analyzed by the auditory model, and a predetermined Bit rate, the number of bits needed to allocate the sub-band samples; quantize and encode the sub-bands, based on the bit allocation information; encode and decode the bit allocation information; and format the quantized and encoded Information about the decoded sub-bands, the coded scale factors, and the bit allocation. 2. The sub-band encoding and decoding method according to item 1 of the scope of patent application, wherein the digital voice signal includes a pulse code modulation signal. 3. The sub-band encoding and decoding method as described in item 1 of the patent application scope, wherein the specific encoding method includes an adaptive differential pulse code modulation method. 4. The sub-band encoding and decoding method according to item 1 of the scope of patent application, wherein the sub-band analysis includes dividing the digital voice signal into the sub-bands by a filter bank. 5. The sub-band encoding and decoding method described in item 4 of the scope of patent application, wherein the auditory model analyzes and calculates a reasonably significant sound level in any frequency band in the filter bank, and outputs a signal mask for each frequency band ratio. This paper size is applicable to China National Standard (CNS) A4 specification (210X297 mm) ------ install --- r--: 丨 order-^ ---- line (please read the notes on the back first and fill in (This page) ABCD 401402 4224twf.doc / 006 6. Patent application scope 6. The sub-band encoding and decoding method described in item 5 of the patent application scope, wherein the bit allocation procedure is based on the output samples of the filter bank and The signal mask ratio analyzed by the auditory model adjusts the bit allocation to meet the bit rate requirements. 7. The sub-band encoding and decoding method described in item 5 of the scope of patent application, further includes transmitting formatted information. 8. —A kind of sub-band encoding and decoding device, including: a primary band analysis device for performing a sub-band analysis on a digital voice signal to generate a plurality of sub-bands; and a computing device for calculating the corresponding sub-bands. A plurality of scale factors; a coding / decoding device for coding and decoding the scale factors according to an adaptive differential pulse code modulation method; an auditory model analysis device for performing an auditory model analysis of the digital speech signal; one bit An allocating device for allocating bits of the digital speech signal, and allocating the number of bits required for the sub-band samples based on the information analyzed by the auditory model and a predetermined bit rate; a quantizing device for quantizing And encode and decode the sub-bands according to the bit allocation information; a bit allocation codec device for encoding and decoding the bit allocation information; and a formatting device for formatting the quantization and encoding and decoding The sub-bands, the scale factors after encoding and decoding, and the bit allocation information. : --- τ ----- install ----- ^ 丨 order ------- line (please read the notes on the back first ^ Fill this page) The paper size of the paper is applicable to China National Standard (CNS) Α4 (210X297 cm)
TW88102911A 1999-02-26 1999-02-26 Method for sub frequency band encoding/decoding and apparatus thereof TW407402B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
TW88102911A TW407402B (en) 1999-02-26 1999-02-26 Method for sub frequency band encoding/decoding and apparatus thereof

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
TW88102911A TW407402B (en) 1999-02-26 1999-02-26 Method for sub frequency band encoding/decoding and apparatus thereof

Publications (1)

Publication Number Publication Date
TW407402B true TW407402B (en) 2000-10-01

Family

ID=21639778

Family Applications (1)

Application Number Title Priority Date Filing Date
TW88102911A TW407402B (en) 1999-02-26 1999-02-26 Method for sub frequency band encoding/decoding and apparatus thereof

Country Status (1)

Country Link
TW (1) TW407402B (en)

Similar Documents

Publication Publication Date Title
EP1914724B1 (en) Dual-transform coding of audio signals
KR101192241B1 (en) Mixing of input data streams and generation of an output data stream therefrom
JP4506039B2 (en) Encoding apparatus and method, decoding apparatus and method, and encoding program and decoding program
JP5175028B2 (en) Digital signal encoding method and apparatus, and decoding method and apparatus
JP4168976B2 (en) Audio signal encoding apparatus and method
TWI585749B (en) Lossless-encoding method
EP1914725B1 (en) Fast lattice vector quantization
JP6535466B2 (en) Speech sound coding apparatus, speech sound decoding apparatus, speech sound coding method and speech sound decoding method
JP3277692B2 (en) Information encoding method, information decoding method, and information recording medium
KR100419546B1 (en) Signal encoding method and apparatus, Signal decoding method and apparatus, and signal transmission method
EP0884850A2 (en) Scalable audio coding/decoding method and apparatus
JPH0851366A (en) Adaptive encoding system
CN111179946B (en) Lossless encoding method and lossless decoding method
WO1994028633A1 (en) Apparatus and method for coding or decoding signals, and recording medium
US20080140393A1 (en) Speech coding apparatus and method
JP3297240B2 (en) Adaptive coding system
JPH05206866A (en) High efficiency encoder
TW200406096A (en) Improved low bit-rate audio coding systems and methods that use expanding quantizers with arithmetic coding
JP6388624B2 (en) Method, encoder, decoder, and mobile device
JPH0846518A (en) Information coding and decoding method, information coder and decoder and information recording medium
WO1995012920A1 (en) Signal encoder, signal decoder, recording medium and signal encoding method
JP2005037949A (en) Compressing device and restoring device of wide band audio signal, and compressing method and restoring method
JPH0653911A (en) Method and device for encoding voice data
TW407402B (en) Method for sub frequency band encoding/decoding and apparatus thereof
JP2000151413A (en) Method for allocating adaptive dynamic variable bit in audio encoding

Legal Events

Date Code Title Description
GD4A Issue of patent certificate for granted invention patent