JP3139602B2 - Acoustic signal encoding method and decoding method - Google Patents
Acoustic signal encoding method and decoding methodInfo
- Publication number
- JP3139602B2 JP3139602B2 JP07065622A JP6562295A JP3139602B2 JP 3139602 B2 JP3139602 B2 JP 3139602B2 JP 07065622 A JP07065622 A JP 07065622A JP 6562295 A JP6562295 A JP 6562295A JP 3139602 B2 JP3139602 B2 JP 3139602B2
- Authority
- JP
- Japan
- Prior art keywords
- signal
- code
- encoding
- frequency
- band
- Prior art date
- Legal status (The legal status 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 status listed.)
- Expired - Lifetime
Links
Abstract
Description
【０００１】[0001]
【産業上の利用分野】この発明は、楽音や音声などの音
響信号を周波数領域で帯域分割して階層符号化する符号
化方法及びその復号化方法に関する。BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an encoding method for hierarchically encoding an audio signal such as a musical sound or a voice by dividing it into bands in a frequency domain and a decoding method thereof.
【０００２】[0002]
【従来の技術】音響信号を周波数領域で帯域分割して符
号化する方法として、サブバンド符号化方法がある。サ
ブバンド符号化方法はＱＭＦ（Ｑｕａｄｒａｔｕｒｅ
Ｍｉｒｒｏｒ Ｆｉｌｔｅｒ）を用いて入力信号を複数
の周波数帯域に分割し、その各帯域に適切なビット割り
当てを行いつつ各帯域を独立に符号化する。2. Description of the Related Art There is a sub-band encoding method as a method of encoding an audio signal by dividing the frequency band into bands. The subband encoding method is QMF (Quadrature).
The input signal is divided into a plurality of frequency bands using a Mirror Filter, and each band is independently encoded while allocating appropriate bits to each band.
【０００３】現在、楽音及び音声などの音響信号の符号
化方法は使用目的、復号品質、符号化速度などに応じて
多種多様な方法が有るが、１つの音響信号に対して複数
の符号化方法を得ることなく１つの符号化方法でのみ符
号化するのが普通である。しかし、例えば図１Ａに示す
ように音響信号１１を周波数軸上で低域側から３つのサ
ブバンドＳＢ1 ，ＳＢ2 ，ＳＢ3 に分割して階層化し、
図２に示すようにその下位層（階層１）であるサブバン
ドＳＢ 1 は符号化品質は低い、すなわち復号再生音の周
波数帯域が狭く、量子化誤差も大きい符号化方法、例え
ば符号駆動線形の予測符号化法：ＣＥＬＰにより高圧縮
率で符号化し、逆に上位層（階層３）であるサブバンド
ＳＢ3 の符号化は符号化品質が高く、すなわち復号再生
音の周波数帯域が広く、量子化誤差が小さい符号化方法
（例えば離散コサイン変換符号化方法などの変換符号化
法で低圧縮率で符号化し、中位層（階層２）であるサブ
バンドＳＢ2 に対しては下位層の符号化方法と、上位層
の符号化方法との中間の符号化方法とし、利用者の要求
に応じて階層１のみを符号化送出し、あるいは階層１と
２を符号化送出し、又は全ての階層を符号化送出すると
いう符号化方法も、考えられる。At present, there are various methods for encoding audio signals such as musical tones and voices, depending on the purpose of use, decoding quality, encoding speed, and the like. It is common to encode with only one encoding method without obtaining. However, for example, as shown in FIG. 1A, the acoustic signal 11 is divided into three sub-bands SB 1 , SB 2 , and SB 3 from the low-frequency side on the frequency axis to form a hierarchy.
As shown in FIG. 2, the sub-band SB 1 which is the lower layer (layer 1) Is a coding method with low coding quality, that is, a narrow frequency band of decoded reproduced sound and a large quantization error, for example, coding with a code-driven linear predictive coding method: CELP at a high compression rate, and conversely, an upper layer ( coding the subband SB 3 is a hierarchy 3) has a high encoding quality, i.e. wide frequency band of the decoded sound data, transform coding such as quantization error is small coding method (e.g. discrete cosine transform coding method encoded at a low compression ratio in the law, and the encoding method of a lower layer, an intermediate encoding method and encoding method of the upper layer with respect to the sub-band SB 2 is moderate layer (hierarchy 2), use A coding method in which only layer 1 is coded and transmitted, or layers 1 and 2 are coded and transmitted, or all layers are coded and transmitted in response to a user's request, is also conceivable.
【０００４】あるいは前述のように３つに階層符号化さ
れた各種の楽音又は音声信号を例えばデータベースとし
て設け、利用者からのそのデータベースをアクセスし、
所望の楽音信号を受け取り、その利用者の復号器に応じ
て、階層１の符号のみを復号して狭帯域かつ量子化誤差
の大きい低品質の再生音を得、あるいは階層１及び２の
符号を復号、又は階層１，２，３の全ての符号を復号し
て広帯域かつ量子化誤差の小さい高品質な再生音を得る
ことが考えられる。[0004] Alternatively, various musical sounds or audio signals hierarchically encoded into three as described above are provided as, for example, a database, and the database is accessed by a user.
A desired tone signal is received, and according to the decoder of the user, only the code of the layer 1 is decoded to obtain a low-quality reproduced sound with a narrow band and a large quantization error, or the code of the layer 1 and 2 is decoded. It is conceivable to obtain high-quality reproduced sound with a wide band and a small quantization error by decoding or decoding all codes in layers 1, 2, and 3.
【０００５】又は、例えば、音声が支配的な広帯域の音
響信号を２階層に分けて符号化し、その下位層符号のみ
を復号すれば主に音声的な性質を有する音響信号をきれ
いに復号し、下位層と上位層との両符号を復号すれば、
更に、非音声的な性質を有する音響信号も含めた信号の
復号ができる、ということが考えられる。またこれらの
場合において、下位層符号のみを受け取り、その際の伝
送路の利用時間を短かくしまたは、伝送容量の小さな伝
送路を使用し、かつ実時間で復号したり、長い時間かけ
て上位層符号をも受けとり、一度蓄積した後、改めて再
生復号することにより高品質の復号信号を得ることもで
きる。[0005] Alternatively, for example, a wideband acoustic signal in which speech is dominant is divided into two layers and encoded, and if only the lower layer code is decoded, the acoustic signal mainly having speech characteristics is decoded clearly, and the lower layer is decoded. By decoding both the upper and lower layer codes,
Further, it is conceivable that a signal including an acoustic signal having a non-speech property can be decoded. Also, in these cases, only the lower layer code is received, and the use time of the transmission path at that time is shortened, or a transmission path with a small transmission capacity is used, and decoding is performed in real time, It is also possible to obtain a high-quality decoded signal by receiving the code, storing it once, and then reproducing and decoding it again.
【０００６】あるいは、これらの場合において、下位・
上位層の全ての符号を一度蓄積した後、下位層符号のみ
を、小型かつ経済的な遅延時間の小さい復号器により実
時間で復号したり、高品質な音を再生したい時には、上
位層符号をも含めて、大型かつ遅延時間の大きな復号器
により、時間をかけて復号し、その後で一度に再生する
こともできる。[0006] Alternatively, in these cases,
After accumulating all the codes of the upper layer once, only the lower layer code is decoded in real time by a small and economical decoder with a small delay time, or when it is desired to reproduce high-quality sound, the upper layer code is used. In addition, decoding can be performed over a long time by a large-sized decoder having a large delay time, and thereafter, can be reproduced at once.
【０００７】前述のように復号品質や符号化圧縮率に選
択性をもたせる符号化方法はスケーラブルな階層符号化
方法と称せられる。スケーラブルな階層符号化方法とし
ては図１Ａに示したサブバンド符号化方法が考えられ
る。すなわち符号化方法１によってサブバンドＳＢ1 の
周波数帯域を符号化し、同様にして帯域ＳＢ2 ，ＳＢ3
を各々独立した符号化方法２，３により符号化を実行す
る。図１Ｂに示すように、復号化の際には、例えば、広
帯域の復号音を必要としない時には、サブバンドＳＢ1
の符号のみを符号化方法１の復号器により復号化して、
サブバンドＳＢ1 の帯域のみの音の復号信号１２1 を
得、また広帯域復号音を必要とする場合はサブバンドＳ
Ｂ1 ，ＳＢ2 ，ＳＢ3 の各符号をそれぞれ符号化方法
１，２，３と対応した復号量により復号して復号信号１
２1 ，１２2 ，１２3 を得てこれらの合成信号１２を出
力する。[0007] As described above, an encoding method that gives selectivity to the decoding quality and the encoding compression ratio is called a scalable hierarchical encoding method. As a scalable hierarchical encoding method, the subband encoding method shown in FIG. 1A can be considered. That encodes the frequency band of the subband SB 1 by the coding method 1, band SB 2 in the same manner, SB 3
Are encoded by independent encoding methods 2 and 3, respectively. As shown in FIG. 1B, when decoding, for example, when a wideband decoded sound is not required, the sub-band SB 1
Is decoded by the decoder of the encoding method 1,
A decoded signal 12 1 of the sound only in the band of the sub-band SB 1 is obtained.
Each of the codes B 1 , SB 2 , and SB 3 is decoded by a decoding amount corresponding to each of the encoding methods 1, 2, and 3, and the decoded signal 1
2 1, 12 2, 12 3 to obtain outputs these synthetic signals 12.
【０００８】[0008]
【発明が解決しようとする課題】しかし、このようなサ
ブバンド符号化方法による階層符号化では、各帯域（す
なわち各層）に発生する量子化誤差、すなわち符号器の
入力信号とその局部復号器の出力信号、つまり伝送路な
どの影響を受けていない復号信号との誤差が図１Ｃに示
すように各帯域ＳＢ1 ，ＳＢ2 ，ＳＢ3 にそれぞれ量子
化誤差１３1 ，１３2 ，１３3 として保存され、よって
全周波数帯域の復号信号１２には各帯域毎に独立に歪み
や雑音が発生してしまう。従って、全帯域を復号する場
合（すなわち上位層までの復号化）でさえも、下位層の
大きな量子化誤差１３1 も、そのまま発生するため、高
品質のものは得られない。広帯域復号信号を高品質に得
るには各サブバンドＳＢ1 ，ＳＢ2 ，ＳＢ3 での各符号
化圧縮率を小さくしなければ、量子化雑音を低減させる
ことができない。従ってこのような階層符号化方法で
は、スケーラブルな符号化を実現できない。[SUMMARY OF THE INVENTION However, in the hierarchical coding by such a sub-band coding method, the quantization error that occurs each band (i.e. each layer), i.e., the input signal of the encoder and its local decoder 1C, that is, an error from a decoded signal that is not affected by a transmission path or the like, as quantization errors 13 1 , 13 2 , and 13 3 in each of the bands SB 1 , SB 2 , and SB 3 as shown in FIG. 1C. Thus, the decoded signal 12 in the entire frequency band is subjected to distortion and noise independently for each band. Therefore, even when decoding the entire band (i.e. decoding up to the upper layer), a large quantization error 13 1 of the lower layer also, since it occurs, it can not be obtained of high quality. If in order to obtain a wideband decoded signal with high quality by reducing the respective coding compression rate in each subband SB 1, SB 2, SB 3 , it is impossible to reduce the quantization noise. Therefore, scalable encoding cannot be realized by such a hierarchical encoding method.
【０００９】従来のサブバンド符号化方法によってスケ
ーラブルな符号化ができないことを図３を参照して更に
具体的に説明する。即ち原音響信号１１の帯域を２分割
し、第１階層（低域領域）をＣＥＬＰ方式で符号化し、
第２階層（高域領域）を変換符号化方法により符号化し
ている。第１階層では、音声の圧縮効率の高いＣＥＬＰ
符号化が実行されているため、その局部復号信号１
２ 1 、（図３Ｂ）の量子化誤差信号１３1 は図３Ｃに示
すように比較的大きい。一方第２階層では様々な波形に
対して符号化可能な変換符号化が実行されているため、
その局部復号信号１２2 は図３Ｂに示すように原音信号
１１に近く、量子化誤差信号１３2 は図３Ｃに示すよう
に小さい。しかし第１階層の符号及び第２階層の符号を
それぞれ復号して広域復号信号を得ても、図３Ｄに示す
ようにその復号信号の量子化誤差の低域部分１４1 は第
１階層の量子化誤差１３1 と変わらない。すなわち、第
２階層までの復号品質は低周波数の帯域においてＣＥＬ
Ｐ符号化方法の符号化性能に依存してしまう。 [0009] more specifically described with reference to FIG. 3 that the conventional sub-band coding method thus can not schedule <br/> Raburu coding. That is, the band of the original audio signal 11 is divided into two, and the first layer (low-frequency region) is encoded by the CELP method.
The second layer (high-frequency region) is encoded by a transform encoding method. In the first layer, CELP with high audio compression efficiency
Since the encoding has been performed, the local decoded signal 1
2 1, the quantization error signal 13 1 (FIG. 3B) is relatively large as shown in FIG. 3C. On the other hand, in the second layer, since transform coding capable of coding various waveforms is performed,
The station unit decoded signal 12 2 is close to the original signal 11 as shown in FIG. 3B, the quantization error signal 13 2 is small as shown in FIG. 3C. But to obtain a wide decoded signal the sign of the first layer code and second layer <br/>, respectively which decodes the low frequency portion 14 1 of the quantization error of the decoded signal, as shown in FIG. 3D It does not change the quantization error 13 1 of the first hierarchy. That is, the decoding quality up to the second layer is CEL in the low frequency band.
It depends on the encoding performance of the P encoding method .
【００１０】この発明の目的は、下位層での復号品質の
影響を上位層でも受けない高品質の復号品質が得られる
スケーラブルな符号化方法及びその復号化方法を提供す
ることにある。[0010] An object of the present invention is to improve decoding quality in a lower layer.
Effect is to provide a scalable encoding method and decoding method that a high-quality decoding quality which is not subject to obtain the upper layer.
【００１１】[0011]
【課題を解決するための手段】請求項１の発明によれ
ば、楽音や音声などの最高周波数がｆn の音響入力信号
を周波数ｆ1 ，ｆ2 ，……，ｆn-1 （ｆ1 ＜ｆ2 ＜，…
…，＜ｆn-1 ＜ｆn ）のｎ個の区分（ｎは２以上の整
数）に分割して符号化する符号化方法において、入力信
号から周波数がｆ1 以下の第１帯域信号を選出し、その
第１帯域信号を第１符号化方法で符号化して第１符号を
出力し、第ｉ−１以下の各符号（ｉ＝２，３，……，
ｎ）から周波数がｆi-1 以下の第ｉ−１復号信号を得、
上記入力信号から周波数ｆi 以下の第ｉ帯域信号を選出
し、その第ｉ帯域信号から上記第ｉ−１復号信号を差し
引いて第ｉ差信号を得、その第ｉ差信号を第ｉ符号化方
法で符号化して第ｉ符号を出力する。According to the Summary of the invention of claim 1, the frequency f 1 the highest frequency of the acoustic input signal f n, such as tone and voice, f 2, ......, f n-1 (f 1 <F 2 <, ...
, <F n-1 <f n ) In a coding method of coding by dividing into n sections (n is an integer of 2 or more), a first band signal having a frequency of f 1 or less is input from an input signal. , The first band signal is encoded by a first encoding method, and a first code is output, and each code of i−1 or less (i = 2, 3,...,
n) to obtain an (i-1) th decoded signal having a frequency of f i-1 or less;
An i-th band signal having a frequency f i or lower is selected from the input signal, and the i-th decoded signal is subtracted from the i-th band signal to obtain an i-th difference signal. And output the i-th code.
【００１２】第ｉ−１復号信号は例えば、第ｉ−１符号
を復号化した信号と、第ｉ−２復号信号とを加算し、そ
の加算信号を標本化周波数が２ｆi の信号に変換して得
る。請求項３の発明の符号化方法によれば楽音、音声な
どの最高周波数がｆn の音響入力信号を、周波数ｆ1 ，
ｆ2 …，ｆn-1 （ｆ1 ＜ｆ2 ＜，…＜ｆn-1 ＜ｆn）
（ｎ＝２以上の整数）で区分してそれぞれ符号化する符
号化方法において、上記音響入力信号より標本化周波数
が２ｆ1 の第１帯域音響信号を得、その第１帯域信号を
第１符号化法により符号化して第１符号を出力し、その
第１符号の符号化誤差を第ｉ−１誤差信号を得（ｉ＝
２，３，…，ｎ）、その第ｉ−１誤差信号を標本化周波
数が２ｆi の第ｉ−１変換誤差信号に変換し、上記音響
入力信号より周波数帯域がｆi-1 〜ｆi 、標本化周波数
が２ｆi の第ｉ帯域信号を得、上記第ｉ−１変換誤差信
号と上記第ｉ帯域信号とを加算して第ｉ加算信号を得，
その第ｉ加算信号と第ｉ符号化法により符号化して第ｉ
符号を出力する。[0012] The i-1 decoded signal, for example, a signal obtained by decoding the first i-1 code, adds the first i-2 decoded signal, the addition signal sampling frequency into a signal of 2f i Get it. According to the encoding method of the third aspect of the present invention, an audio input signal having a maximum frequency of f n , such as a musical sound or a voice, is converted into a frequency f 1 ,
f 2 ..., f n-1 (f 1 <f 2 <, ... <f n-1 <f n)
(N = integer of 2 or more) In the encoding method, a first band audio signal having a sampling frequency of 2f 1 is obtained from the audio input signal, and the first band signal is converted to a first code. A first code is output by encoding according to a coding method, and an encoding error of the first code is obtained as an (i−1) th error signal (i =
2, 3,..., N), the (i-1) th error signal is converted into an (i-1) th conversion error signal having a sampling frequency of 2f i , and the frequency band is f i-1 to f i based on the acoustic input signal. , Obtain an i-th band signal having a sampling frequency of 2f i , add the (i−1) -th conversion error signal and the i-th band signal to obtain an i-th added signal,
The i-th addition signal and the i-th encoding method are used to encode the
Output sign.
【００１３】請求項４の発明によれば、請求項１乃至３
の何れかの発明において上記第１符号化法として符号駆
動線形予測符号化法を用い、上記第ｎ符号化法として変
換符号化法を用いる。請求項５の発明によれば、請求項
１乃至３の何れかの発明において上記第１乃至第ｎ符号
化法として何れも変換符号化法を用いる。請求項６の発
明では請求項１乃至５の何れかの発明において、上記音
響入力信号中の周波数ｆi 以下のほぼ全域の成分のスペ
クトル包絡を重みの基準として、上記第ｉ符号の符号化
過程において心理聴覚重み付け量子化を行う。According to the invention of claim 4, claims 1 to 3 are provided.
Either using a code excited linear predictive coding method as the first coding method in the invention, using varying <br/>換符-coding method as the first n coding method. According to the invention of claim 5, claim
The first to n-th codes according to any one of the first to third aspects,
In each case, a transform coding method is used. According to a sixth aspect of the present invention, in the invention of any one of the first to fifth aspects, the coding process of the i-th code is performed by using a spectral envelope of components of substantially the entire region below the frequency f i in the audio input signal as a weight reference. Performs psychoacoustic weighting quantization.
【００１４】請求項７の発明の復号化方法によれば、入
力符号を第１乃至第ｎ符号（ｎは２以上の整数）に分離
し、上記第１符号を復号して、標本化周波数２ｆ1 の第
１復号信号を出力し、上記第ｉ−１復号信号（ｉ＝２，
３，…，ｎ）を標本化周波数が２ｆi の第ｉ−１変換復
号信号に変換し、上記第ｉ符号を復号して標本化周波数
２ｆi の第ｉ復号信号を得、その第ｉ復号信号と上記第
ｉ−１変換復号信号とを加算して第ｉ加算信号を出力す
る。According to the decoding method of the invention of claim 7, the input code first to n symbols (n is an integer of 2 or more) separated into decodes the first code, the sampling frequency 2f 1 to output the first decoded signal, and the (i-1) th decoded signal (i = 2,
,..., N) are converted into an (i−1) -th converted decoded signal having a sampling frequency of 2f i , and the i-th code is decoded to obtain an i-th decoded signal having a sampling frequency of 2f i . The i-th conversion signal is added to the i-th conversion decoded signal to output an i-th addition signal.
【００１５】[0015]
【実施例】図４Ａに請求項１の発明の符号化方法の実施
例を適用した符号化器の例を示す。この例では原音信号
を２つの周波数帯域に分けて符号化、つまり２階層符号
化する場合である。入力端子２１からの原音入力信号１
１は標本化周波数が２４ｋＨｚ、つまり最高周波数ｆ2
が１２ｋＨｚのデジタル信号であり、この入力信号は第
１帯域選択手段としてのサンプルレート変換器２２1 で
標本化周波数が１６ｋＨｚの信号に変換されて第１帯域
信号２３が取出される。このサンプルレート変換はいわ
ゆるダウンサンプリングであり、例えば変換標本化周波
数比に応じた間隔でサンプルが除去された後、デジタル
低域通過フィルタを通されて実行される。このサンプル
レート変換器２２1 よりの周波数がｆ1 ＝８ｋＨｚ以下
の第１帯域信号２３が取出され、この第１帯域信号２３
は第１符号化方法による第１符号器２４1 で符号化され
る。この例では第１符号器２４1 としてＣＥＬＰ（符号
駆動線形予測符号化）符号方法により符号化する。この
符号化の結果である第１符号Ｃ1 が出力される。FIG. 4A shows an example of an encoder to which an embodiment of the encoding method according to the first aspect of the present invention is applied. In this example, the original sound signal is divided into two frequency bands and encoded, that is, two-layer encoding is performed. Original sound input signal 1 from input terminal 21
1 indicates that the sampling frequency is 24 kHz, that is, the highest frequency f 2
There is a 12kHz digital signal, the input signal is the sampling frequency in the sample rate converter 22 1 of the first band selection unit first band signal 23 is converted into a 16kHz signal is taken out. This sample rate conversion is so-called downsampling, for example, after a sample is removed at an interval corresponding to the conversion sampling frequency ratio, and is then passed through a digital low-pass filter. A first band signal 23 having a frequency f 1 = 8 kHz or less is taken out from the sample rate converter 22 1 , and the first band signal 23
It is encoded by the first encoder 24 1 according to the first coding method. In this example, the first encoder 24 1 is encoded by a CELP (Code-Driven Linear Predictive Encoding) encoding method. First code C 1 is the result of the coding is output.
【００１６】この実施例では局部復号器２５1 で復号さ
れ、周波数がｆ1 以下の第１復号信号１２1 が得られ、
その復号信号１２1 は第１サンプルレート変換器２６1
で標本化周波数が２４ｋＨｚの変換復号信号２７に変換
される。このサンプルレート変換器２６1 はいわゆるア
ップサンプリングを行うものであり、例えば、変換周波
数比に応じた間隔でゼロサンプルを加えた後デジタル低
域通過フィルタに通せばよい。差回路２８で入力信号１
１からこの変換復号信号２７が差引かれ、その差信号２
９が第２符号化方法による第２符号器２４2 で符号化さ
れる。この実施例では第２符号器２４2 で変形離散コサ
イン変換などの変換符号化（Ｔｒａｎｓｆｏｒｍ ｃｏ
ｄｉｎｇ）により符号化される。この符号化結果の第２
符号Ｃ2は出力される。第１符号Ｃ1 と第２符号Ｃ2 は
多重化回路３１で、例えば図４ｂに示すように符号化フ
レームごとに時分割的に多重化され、符号化符号Ｃとし
て出力される。利用者の要求によっては第１符号Ｃ1 の
みを出力してもよい。In this embodiment, a first decoded signal 12 1 having a frequency of f 1 or less is obtained by decoding by a local decoder 25 1 ,
The decoded signal 12 1 is supplied to the first sample rate converter 26 1
Is converted into a converted decoded signal 27 having a sampling frequency of 24 kHz. This sample rate converter 26 1 performs so-called upsampling. For example, it is sufficient to add a zero sample at intervals according to the conversion frequency ratio and then pass through a digital low-pass filter. Input signal 1 in the difference circuit 28
1 is subtracted from the converted decoded signal 27 and the difference signal 2
9 is encoded by the second encoder 24 2 by the second encoding method. In this embodiment transformation coding such as modified discrete cosine transform by the second encoder 24 2 (Transform co
ding). The second of this encoding result
Code C 2 is output. The first code C 1 and the second code C 2 are multiplexed in a multiplexing circuit 31 in a time-division manner for each coded frame, for example, as shown in FIG. May output only the first code C 1 is the user's request.
【００１７】標本化周波数２４ｋＨｚの原音入力信号１
１の周波数スペクトルは例えば図５Ａに示され、この信
号１１中の８ｋＨｚ以下の信号が標本化周波数１６ｋＨ
ｚの信号２３（図５Ｂ）として下位層の第１符号器２４
1 に入力され、高い圧縮効率で符号化される。その符号
化符号Ｃ1 の局部復号器２５1 により復号された復号信
号１２1 は図５Ｂに示すように、下位層入力信号２３に
対しては少なからず量子化誤差１３1 が図５Ｃに示すよ
うに生じる。差回路２８からこの誤差信号１３ 1 と、原
音入力信号１１の８ｋＨｚ以上の高域信号３３とよりな
る信号２９が上位層の第２符号器２４2 に入力され、あ
らゆる性質の入力信号を高品質に符号化可能な変換符号
化方法により符号化される。Original sound input signal 1 having a sampling frequency of 24 kHz
1 is shown in FIG. 5A, for example.
The signal of 8 kHz or less in No. 11 has a sampling frequency of 16 kHz.
z signal 23 (FIG. 5B) as the first encoder 24 in the lower layer.
1And encoded with high compression efficiency. Its sign
Encoding code C1Local decoder 251Decrypted signal decoded by
No. 121Is applied to the lower-layer input signal 23 as shown in FIG. 5B.
On the other hand, the quantization error 131Is shown in Figure 5C
Will occur. From the difference circuit 28, the error signal 13 1And Hara
The high frequency signal 33 of 8 kHz or more of the sound input signal 11
Is transmitted to the second encoder 24 of the upper layer.TwoIs entered in
A transform code that can encode input signals of any quality with high quality
Is encoded by the conversion method.
【００１８】このようにこの実施例では下位層の符号化
符号Ｃ1 は原音をそれ程忠実には符号化しないが、上位
層では下位層の量子化誤差も含めて符号化されるため、
後述で明らかにするように、上位層まで復号する場合
に、下位層をも高い忠実度で復号再生することが可能と
なる。つまり下位層では高い圧縮効率で符号化し、しか
も上位層をも復号する場合は、高品質の復号信号を得る
ことができる。As described above, in this embodiment, the lower layer coding code C 1 does not encode the original sound so faithfully, but the upper layer codes the lower layer coding error including the lower layer quantization error.
As will be described later, when decoding is performed up to the upper layer, the lower layer can be decoded and reproduced with high fidelity. That is, when encoding is performed with high compression efficiency in the lower layer and decoding is also performed in the upper layer, a high-quality decoded signal can be obtained.
【００１９】特に前記実施例では下位層の符号化にＣＥ
ＬＰ方式を用いているため符号化対象が音声の場合、下
位層の第１符号Ｃ1 のみを復号しても比較的良好な品質
が得られ、また演算量が少なく、実時間処理が容易であ
る。下位層の第１符号Ｃ 1 と上位層の第２符号Ｃ 2 をそ
れぞれ復号すれば、上位層の変換符号Ｃ 2 の復号信号
が、下位層のＣＥＬＰ符号Ｃ 1 への符号化誤差を補償す
るので、符号化対象が楽音であっても広帯域にわたり、
品質の高い復号信号が得られる。In particular, in the above-described embodiment, CE is used for the encoding of the lower layer.
If encoded is a voice due to the use of LP mode, and decodes only the first code C 1 of the lower layer relatively good quality can be obtained even, also calculation amount is small, easy real time processing is there. Second code C 2 to its first code C 1 and the upper layer of the lower layer
If each of them is decoded, the decoded signal of the upper-layer transform code C 2 is obtained.
But to compensate for the coding error to the CELP code C 1 of the lower layer
Therefore, even if the encoding target is a musical tone ,
A high quality decoded signal is obtained.
【００２０】符号化を行う場合に、人間の心理聴覚、例
えば大きいレベルのスペクトルによるマスキング特性な
どを考慮して、心理聴覚重み付けをして符号化すること
により聴覚的に量子化誤差を抑圧した効率的な符号化を
することがよくある。例えば符号器２４1 のＣＥＬＰ符
号化方法においては図６に示すように、制御部３５によ
り指定される周期（ピッチ）のベクトルが適応符号帳３
６から取出され、また指定された雑音符号帳３７から雑
音ベクトルが取出され、これらはそれぞれ利得が付与さ
れた後、合成されて線形予測合成フィルタ３８に励振ベ
クトルとして入力される。一方図４Ａのサンプルレート
変換器２２1 よりの入力信号は符号化フレーム周期で線
形予測分析部３９で線形予測分析され、その線形予測係
数が量子化部４１で量子化され、その量子化線形予測係
数に応じて合成フィルタ３８のフィルタ係数が設定され
る。また聴覚重み付け係数演算部４３で線形予測係数よ
り求めたスペクトル包絡に基づいて心理聴覚重み付けの
ためのフィルタ係数を求めて、聴覚重み付けフィルタ４
２に設置する。サンプルレート変換器２２1 よりの入力
信号から合成フィルタ３８よりの合成信号が差し引か
れ、その差信号が聴覚重み付けフィルタ４２へ通され、
その出力のエネルギーが最小になるように制御部３５に
より適応符号帳３６、雑音符号帳３７に対する選択が行
われる。In the case of encoding, in consideration of the psychological auditory perception of human beings, for example, masking characteristics by a large level spectrum, etc., the encoding is performed with psychological auditory weighting, so that the quantization error is suppressed acoustically and the efficiency is reduced. Often, the encoding is done. For example, in the CELP encoding method of the encoder 24 1 , as shown in FIG.
6 and a noise vector from a designated noise codebook 37, which are added with a gain and then combined and input to a linear prediction synthesis filter 38 as an excitation vector. One input signal from the sample rate converter 22 1 of FIG. 4A is a linear prediction analysis in the linear prediction analysis unit 39 in the encoding frame period, the linear prediction coefficients are quantized by the quantization unit 41, the quantized linear prediction The filter coefficient of the synthesis filter 38 is set according to the coefficient. Further, a filter coefficient for psychological auditory weighting is obtained based on the spectral envelope obtained from the linear prediction coefficient by the auditory weighting coefficient calculator 43, and the auditory weighting filter 4
2 Sample rate converter 22 combined signal from the input signal from the synthesis filter 38 from 1 is subtracted, and the difference signal is passed to the perceptual weighting filter 42,
The control unit 35 selects the adaptive codebook 36 and the noise codebook 37 so that the output energy is minimized.
【００２１】変換符号器２４2 の変換符号化方法におい
ては、例えば図７に示すように差回路器２８の出力が離
散コサイン変換器４５で直交コサイン変換されて周波数
領域の係数に変換され、そのスペクトル包絡成分が線形
予測分析部４６で線形予測分析され、これよりスペクト
ル包絡を得、そのスペクトル包絡で変換器４５の出力係
数が割算されて正規化され、その平均化された係数が聴
覚重み付け部４７で聴覚重み付けがなされ、更に量子化
部４８で例えばベクトル量子化される。聴覚重み付け係
数を得るため、この実施例について入力端子２１から原
音入力信号１１が離散コサイン変換器４９で直交コサイ
ン変換して、周波数領域に変換され、その変換係数のス
ペクトル包絡にもとづいて聴覚重み付け係数が係数演算
部５１で演算されて聴覚重み付け部４７に与えられ、正
規化係数の対応する成分に対する乗算がなされる。In the transform coding method of the transform encoder 24 2 , for example, as shown in FIG. 7, the output of the difference circuit 28 is subjected to orthogonal cosine transform by the discrete cosine transformer 45 to be transformed into frequency domain coefficients. The spectral envelope component is subjected to linear predictive analysis in the linear predictive analysis unit 46, thereby obtaining a spectral envelope. The output coefficient of the converter 45 is divided and normalized by the spectral envelope, and the averaged coefficient is weighted by the auditory weight. Perceptual weighting is performed by the unit 47, and further, the quantization unit 48 performs, for example, vector quantization. In order to obtain an auditory weighting coefficient, in this embodiment, the original sound input signal 11 from the input terminal 21 is orthogonally cosine-transformed by the discrete cosine transformer 49 and converted into the frequency domain. Is calculated by the coefficient calculation unit 51 and given to the auditory weighting unit 47, and the corresponding component of the normalization coefficient is multiplied.
【００２２】つまり、上位層の第２の符号器２４2 では
図５Ｃに示すスペクトルの信号２９を符号化するが、こ
の信号２９のスペクトル包絡にもとづいて聴覚重み付け
を行うのではなく、原音入力信号１１のスペクトル包絡
（図５Ｄ）を求め、これに基づいて聴覚重み付け符号化
を行う。次にこの発明の復号化方法の実施例を図８を参
照して説明する。この実施例は図４に示した符号化法に
よる符号化符号の復号化に適用した場合である。入力端
子５５より入力された入力符号に分離回路５６で第１符
号Ｃ1 と第２符号Ｃ2 とに分離され、第１符号Ｃ1 は第
１復号器５７1 によりこの例ではＣＥＬＰ復号化方法に
より最高信号周波数ｆ1 （標本化周波数１６ｋＨｚ）の
第１復号信号５８ 1 に復号されて下位層（低域）復号化
出力６３1 として出力される。That is, the second encoder 24 in the upper layerTwoThen
The signal 29 having the spectrum shown in FIG. 5C is encoded.
Weighting based on the spectral envelope of the signal 29
Instead of performing the spectral envelope
(FIG. 5D) and, based on this, auditory weighted coding
I do. Next, an embodiment of the decoding method of the present invention will be described with reference to FIG.
It will be described in the light of the above. This embodiment is based on the encoding method shown in FIG.
This is a case where the present invention is applied to decoding of a coded code. Input end
The first code is input to the input code input from the
Issue C1And the second code CTwoAnd the first code C1Is the
1 decoder 571In this example, the CELP decoding method
Higher signal frequency f1(Sampling frequency 16kHz)
First decoded signal 58 1And lower layer (low-frequency) decoding
Output 631Is output as
【００２３】この第１復号化出力５８1 はサンプルレー
ト変換器５９により最高信号周波数ｆ2 （標本化周波数
が２４ｋＨｚ）の変換復号信号６１1 に変換される。一
方分離回路５６よりの第２符号Ｃ2 は第２復号器５７2
によりこの例では変換符号復号化がなされ、最高信号周
波数ｆ2 （標本化周波数が２４ｋＨｚ）の第２復号信号
５８2 が得られて、この第２復号信号５８2 は第１変換
復号信号６１1 と加算器６２2 で加算されて上位層（全
帯域）復号化出力６３2 として出力される。The first decoded output 58 1 is converted by the sample rate converter 59 into a converted decoded signal 61 1 having the highest signal frequency f 2 (the sampling frequency is 24 kHz). On the other hand, the second code C 2 from the separation circuit 56 is supplied to the second decoder 57 2
In this example, transform code decoding is performed to obtain a second decoded signal 58 2 having the highest signal frequency f 2 (sampling frequency is 24 kHz), and the second decoded signal 58 2 is the first transformed decoded signal 61 1 and it is added by the adder 62 2 is outputted as an upper layer (the entire band) decoded output 63 2.
【００２４】つまり下位層復号化出力６３1 としては理
想的な場合は図５Ｂ中の復号信号１２1 が得られる。一
方第２復号器５７2 の復号信号５８2 は理想的には図５
Ｅに示すように、下位層（低域）の量子化誤差信号１３
1 の復号信号６０1 と、高域信号３３の復号信号６４2
とである。よって加算器６２2 よりの復号化出力６３2
には低域の復号信号５８1 に対し、その量子化誤差１３
1 と対応する復号信号６０1 が加算され、量子化誤差が
著しく軽減され、かつ高域復号信号６４2 に高い忠実度
のものであるから、加算器６２から得られる上位層まで
の復号化出力６３2 は原音入力信号１１に著しく近く、
その量子化誤差信号は例えば図５Ｆに示すように全帯域
にわたり、著しく小さなものとなる。That is, in the ideal case, the decoded signal 12 1 in FIG. 5B is obtained as the lower layer decoded output 63 1 . Meanwhile decoded signal 58 of the second decoder 57 2 is ideally 5
As shown in E , the quantization error signal 13 of the lower layer (low band)
A first decoded signal 60 1, the decoded signal 64 and second high-frequency signal 33
And Thus the adder 62 decoding than 2 output 63 2
Has a quantization error of 13 for the low-frequency decoded signal 58 1.
1 and the decoded signal 60 1 corresponding to 1 are added to significantly reduce the quantization error and have high fidelity to the high-frequency decoded signal 64 2. 63 2 is close significantly original input signal 11,
The quantization error signal becomes extremely small over the entire band, for example, as shown in FIG. 5F.
【００２５】次にこの発明の符号化方法をｎ階層（ｎ帯
域）分割符号化に適用した例としてｎ＝４の場合につい
て図９を参照して説明する。図９において図４Ａと対応
する部分に同一符号を付けてある。この例では原音入力
信号１１は最高周波数がｆn＝ｆ4 でその標本化周波数
が２ｆ4 であり、第１サンプルレート変換器（第１帯域
選択手段）２２1 で標本化周波数が２ｆ1 （但しｆ1 ＜
ｆ2 ＜ｆ3 ＜ｆ4 ）の入力信号２３1 に変換され、つま
り周波数ｆ1 以下の第１帯域信号２３1 が選出され、そ
の第１帯域信号２３1 は第１符号器２４1 で符号化さ
れ、第１符号Ｃ1として出力されると共にその第１符号
Ｃ1 は第１復号器２５1 で標本化周波数２ｆ1 の信号に
復号され、その復号信号１２1 は第１サンプルレート変
換器２６1で標本化周波数が２ｆ2 の第１変換復号信号
に変換される。一方入力信号１１が第２帯域選択手段と
してのサンプルレート変換器２２2 で標本化周波数が２
ｆ2の信号に変換されて、周波数ｆ2 以下の第２帯域信
号２３2 が取出される。この第２帯域信号２３2 から第
１サンプルレート変換器２６1 よりの第１変換復号信号
が第２差回路２８2 で引算され、その第２差信号２９2
が第２符号器２４2 で符号化され、第２符号Ｃ2 が出力
される。Next, a case where n = 4 will be described with reference to FIG. 9 as an example in which the encoding method of the present invention is applied to n-layer (n-band) division encoding. In FIG. 9, parts corresponding to those in FIG. 4A are denoted by the same reference numerals. In this example, the original sound input signal 11 has the highest frequency f n = f 4 and the sampling frequency is 2f 4 , and the first sample rate converter (first band selecting means) 22 1 has the sampling frequency of 2f 1 ( However, f 1 <
f 2 <is converted to an input signal 23 1 f 3 <f 4), i.e. the frequency f 1 is the first band signal 23 1 is selected following the first band signal 23 1 code by the first encoder 24 1 ized, its first code C 1 is outputted as the first code C 1 is decoded in the sampling frequency 2f 1 of the signal at a first decoder 25 1, the decoded signal 12 1 is the first sample rate converter At 26 1 , the sampling frequency is converted to a first converted decoded signal having a frequency of 2f 2 . On the other hand the input signal 11 is sampled frequency sample rate converter 22 2 of the second band selection means 2
is converted into f 2 of the signal, the frequency f 2 or less of the second band signal 23 2 is taken out. The first converted decoded signal from the first sample rate converter 26 1 is subtracted from the second band signal 23 2 by the second difference circuit 28 2 , and the second difference signal 29 2 is obtained.
There is encoded in the second encoder 24 2, the second code C 2 is output.
【００２６】以下同様の処理を行うが、第３符号Ｃ3 を
得る処理を、ｉ＝３（ｉ＝２，３，……，ｎ、この例で
はｎ＝４）を例として説明する。第ｉ−１（＝第２）符
号Ｃi-1 （＝Ｃ2 ）が第ｉ−１（＝第２）復号器２５2
で復号されて標本化周波数２ｆi-1 （＝２ｆ2 ）の第ｉ
−１（＝第２）復号信号を得、この第ｉ−１（＝第２）
復号信号と第ｉ−２（＝第１）サンプルレート変換器２
６i-2 （＝２６1 ）よりの第ｉ−２（＝第１）変換復号
信号との和が加算器６０i-1 （＝６０2 ）でとられ、そ
の和信号は第ｉ−１（＝第２）サンプルレート変換器２
６i-1 （＝２６2 ）で標本化周波数２ｆi （＝２ｆ
3 ）、周波数がｆ i （＝ｆ 3 ）以下の第ｉ−１（＝第
２）変換復号信号に変換される。一方、第ｉ（＝第３）
帯域選択手段としてのサンプルレート変換器２２i （＝
２２3 ）により入力信号１１から、周波数がｆi （＝ｆ
3 ）、標本化周波数が２ｆi （＝２ｆ3 ）の第ｉ（＝第
３）帯域信号２３i （＝２３3 ）が取出され、その第ｉ
（＝第３）帯域信号２３i （＝２３3 ）は第ｉ−１（＝
第２）サンプルレート変換器２６i-1 （＝２６2 ）より
の変換復号信号が第ｉ（＝第３）差回路２８i （２８
3 ）で減算され、その第ｉ（＝第３）減算信号２９3 が
第ｉ（＝第３）符号器２４i （＝２４3 ）で符号化さ
れ、第ｉ（＝第３）符号Ｃi （＝Ｃ3 ）を出力する。な
お、第ｉ−１（＝第２）復号器２５i-1 （＝２５2 ）
と、加算器６０i-1 （＝６０2 ）と第ｉ−１（＝第２）
サンプルレート変換器２６i-1 （＝２６2 ）は第ｉ−１
（＝第２）復号化手段４０i-1 （＝４０2 ）を構成す
る。ただ第１復号化手段４０1 は第ｉ−２層が存在せ
ず、つまり更に低域の信号を扱わないので加算器６０ 1
は省略される。また最上位層の帯域信号２３ n （＝２３
4 ）は周波数ｆ n （＝ｆ 4 ）以下の信号であるため第ｎ
（＝４）帯域選択手段としてのサンプルレート変換器２
２ n （＝２２ 4 ）を備える必要はない。 [0026] performs the same processing, the process of obtaining the third code C 3, i = 3 (i = 2,3, ......, n, in this example n = 4) will be described as an example. The i-1 (= second) code C i-1 (= C 2 ) is the i-1 (= second) decoder 25 2
And the i-th sampling frequency 2f i-1 (= 2f 2 )
-1 (= second) decoded signal, and the i-1 (= second) decoded signal is obtained.
Decoded signal and i-2 (= first) sample rate converter 2
The sum with the i-2 (= first) converted decoded signal from 6 i-2 (= 26 1 ) is obtained by the adder 60 i-1 (= 60 2 ), and the sum signal is the i-1 ( i-1 ) th signal. (= Second) sample rate converter 2
6 i-1 (= 26 2 ) and the sampling frequency 2f i (= 2f)
3 ) and the frequency is f i (= f 3 ) Is converted into the following i-1 (= second) converted decoded signal. On the other hand, the i-th (= third)
The sample rate converter 22 i (=
22 3 ), the frequency f i (= f
3 ), an i -th (= third) band signal 23 i (= 23 3 ) having a sampling frequency of 2f i (= 2f 3 ) is extracted, and the i-th band signal is extracted.
The (= third) band signal 23 i (= 233 3 ) is the i−1 (=
( 2 ) The converted decoded signal from the sample rate converter 26 i-1 (= 26 2 ) is converted into an ith (= third) difference circuit 28 i (28)
Is subtracted by 3), the first i (= 3) subtraction signal 29 3 is encoded in the i (= 3) encoder 24 i (= 24 3), the i (= 3) code C i (= C 3 ) is output. Note that the (i−1) th (= second) decoder 25 i−1 (= 25 2 )
When the adder 60 i-1 (= 60 2 ) and the i-1 (= second)
The sample rate converter 26 i-1 (= 26 2 ) is the i-1
The (= second) decoding means 40 i-1 (= 40 2 ). Only the first decoding means 40 is absent is the i-2 layers, that is so further not handle signals of low frequency adder 60 1
Is omitted. Also, the band signal 23 n (= 23) of the uppermost layer
4 ) is the frequency f n (= f 4 ) N-th
(= 4) Sample rate converter 2 as band selection means
2 n (= 22 4) need not be provided with.
【００２７】このようにしてこの発明は入力信号帯域を
ｎ区間に分割して符号化する場合に適用できる。第１〜
第ｎ（＝第４）符号Ｃ1 〜Ｃn （＝Ｃ4 ）は多重化回路
３１でフレームごとに多重化されて符号化符号Ｃとして
出力される。この場合多重化回路３１は第１又は第１〜
第ｉ符号の何れでも選択して出力することができるよう
にされる。第１〜第ｎ（＝第４）符号器２４1 〜２４n
（＝２４4 ）は符号器２４i のｉが大となる程圧縮率が
小さくなる、という使い方をすれば広帯域、高品質の符
号化をする。これを満たせばその符号化方法は、例えば
全てを変換符号化としてもよい。As described above, the present invention can be applied to a case where an input signal band is divided into n sections and encoded. First to first
The n-th (= fourth) codes C 1 to C n (= C 4 ) are multiplexed for each frame by the multiplexing circuit 31 and output as a coded code C. In this case, the multiplexing circuit 31 has the first or first to first
Any of the i-th codes can be selected and output. First to n-th (= fourth) encoders 24 1 to 24 n
If (= 24 4 ) is used such that the compression ratio decreases as the i of the encoder 24 i increases, high-bandwidth, high-quality encoding is performed. If this is satisfied, the encoding method may be, for example, all transform encoding.
【００２８】第１〜第４符号器２４1 〜２４4 において
聴覚重み付け符号化を行う場合はサンプルレート変換器
２２1 ，２２2 ，２２3 よりの各周波数がｆ1 ，ｆ2 ，
ｆ3以下の信号が聴覚重み付け係数演算部７２1 ，７２
2 ，７２3 へそれぞれ供給され、それぞれそのスぺクト
ル包絡に基づく聴覚重み付け係数が演算され、また入力
信号が聴覚重み付け係数演算部７２4 に入力されて同様
に聴覚重み付け係数が演算され、これら聴覚重み付け係
数演算部７２1 〜７２4 でそれぞれ演算された聴覚重み
付け係数が第１〜第４符号器２４1 〜２４4 へ供給さ
れ、前述したように聴覚重み付け符号化が行われる。When perceptual weighting encoding is performed in the first to fourth encoders 24 1 to 24 4 , the frequencies from the sample rate converters 22 1 , 22 2 , and 22 3 are f 1 , f 2 , and 23, respectively.
Signals equal to or smaller than f 3 are output from the auditory weighting coefficient calculators 72 1 , 72
2, 72 3 are supplied to, the perceptual weighting coefficients based on the scan Bae spectrum envelope is calculated respectively, Similarly perceptual weighting coefficient input signal is input to the perceptual weighting coefficient calculating unit 72 4 is calculated, these auditory The perceptual weighting coefficients calculated by the weighting coefficient calculators 72 1 to 72 4 are supplied to the first to fourth encoders 24 1 to 24 4 , and perceptual weighting is performed as described above.
【００２９】この発明の符号化方法をｎ階層分割符号化
への適用例としてｎ＝４の場合を図１０に示す。この例
も原音入力信号１１の最高周波数がｆn ＝ｆ4 でその標
本化周波数が２ｆ4 の場合で、第１サンプルレート変換
器（第１帯域選択手段）２２ 1 で標本化周波数が２ｆ1
（但しｆ1 ＜ｆ2 ＜ｆ3 ＜ｆ4 ）の入力信号２３1 に変
換され、つまり周波数ｆ1 以下の第１帯域信号２３1 が
選出され、その第１帯域信号２３1 は第１符号器２４1
で符号化され、第１符号Ｃ1 として出力されると共にそ
の第１符号Ｃ1 は第１複号器２５1 で標本化周波数２ｆ
1 の信号に復号され、その復号信号１２1 と第１帯域信
号２３1 との差が第１差回路６５1 でとられ、その差信
号（第１誤差信号）１３1 は第１サンプルレート変換器
２６1 で標本化周波数が２ｆ2 の第１変換誤差信号に変
換される。The encoding method according to the present invention is applied to n-layer divisional encoding.
FIG. 10 shows a case where n = 4 as an example of application to the present invention. This example
Also the highest frequency of the original sound input signal 11 is fn= FFourIn that mark
Realization frequency is 2fFour, The first sample rate conversion
Device (first band selecting means) 22 1And the sampling frequency is 2f1
(However, f1<FTwo<FThree<FFour) Input signal 231Strange
The frequency f1The following first band signal 231But
Selected and its first band signal 231Is the first encoder 241
And the first code C1Is output as
The first code C1Is the first duplexer 251And sampling frequency 2f
1And the decoded signal 121And the first band signal
No.231Is different from the first difference circuit 651And the difference
No. (first error signal) 131Is the first sample rate converter
261And the sampling frequency is 2fTwoTo the first conversion error signal of
Is replaced.
【００３０】一方入力信号１１から第２帯域選択手段６
６2 で周波数帯域がｆ1 〜ｆ2 、標本化周波数が２ｆ2
の第２帯域信号２３2 が取出される。例えば入力信号１
１がサンプルレート変換器２２2 で標本化周波数２ｆ2
の信号に変換され、その信号が遮断周波数ｆ1 の高域通
過フィルタ６７2 に通されて第２帯域信号２３2 が得ら
れる。この第２帯域信号２３2 は第１サンプルレート変
換器２６1 よりの第１変換誤差信号と第２加算器６８2
で加算され、その第２加算信号２９2 が第２符号器２４
2 で符号化され、第２符号Ｃ2 が出力される。On the other hand, from the input signal 11, the second band selecting means 6
6 2 at the frequency band f 1 ~f 2, the sampling frequency is 2f 2
2 is taken out second band signal 23. For example, input signal 1
1 is a sampling rate converter 22 2 and a sampling frequency 2f 2
Is converted to the signal, the signal is a second band signal 23 2 is passed through at high pass filter 67 second cut-off frequency f 1 is obtained. The second band signal 23 2 is obtained by combining the first conversion error signal from the first sample rate converter 26 1 with the second adder 68 2
And the second addition signal 29 2 is added to the second encoder 24
2 and a second code C 2 is output.
【００３１】以下同様の処理を行うが、第３符号Ｃ3 を
得る処理を、ｉ＝３（ｉ＝２，３，…，ｎ、この例では
４まで）を例として説明する。第ｉ−１（＝第２）符号
Ｃi- 1 （＝Ｃ2 ）が第ｉ−１（＝第２）復号器２５2 で
復号されて標本化周波数２ｆ i-1 （＝２ｆ2 ）の第ｉ−
１（＝第２）復号信号を得、この第ｉ−１（＝第２）復
号信号と第ｉ−１（＝第２）加算器６８i-1 （＝６
８2 ）より第ｉ−１（＝第２）加算信号２９i-1 （＝２
９2 ）との差が差回路６５i-1 （＝６５2 ）でとられ、
その第ｉ−１（＝第２）誤差信号１３2 は第ｉ−１（＝
第２）サンプルレート変換器２６i-1 （＝２６2 ）で標
本化周波数２ｆi （＝２ｆ3 ）の第ｉ−１（＝第２）変
換誤差信号に変換される。一方、第ｉ（＝第３）帯域選
択手段６６i（＝６６3 ）により入力信号１１から、帯
域がｆi-1 〜ｆi （＝ｆ2 〜ｆ3 ）、標本化周波数が２
ｆi （＝ｆ3 ）の第ｉ（＝第３）帯域信号２３i （＝２
３3 ）が取出され、その第ｉ（＝第３）帯域信号２３i
（＝２３3 ）は第ｉ−１（＝第２）変換誤差信号と第ｉ
（＝第３）加算器６８i （＝６８3 ）で加算され、その
第ｉ（＝第３）加算信号２９3 が第ｉ（＝第３）符号器
２４i （＝２４3 ）で符号化され、第ｉ（＝第３）符号
Ｃi （＝Ｃ3 ）を出力する。Hereinafter, the same processing is performed, but the third code CThreeTo
The process to obtain is i = 3 (i = 2, 3,..., N, in this example,
4) will be described as an example. The (i-1) th (= second) code
Ci- 1(= CTwo) Is the (i−1) th (= second) decoder 25Twoso
Decoded and sampling frequency 2f i-1(= 2fTwo)-
1 (= second) decoded signal, and the i-1 (= second) decoding signal is obtained.
Signal and the (i-1) th (= second) adder 68i-1(= 6
8Two) From the (i-1) th (= second) addition signal 29i-1(= 2
9Two) Is the difference circuit 65i-1(= 65Two),
The i-1 (= second) error signal 13TwoIs the i-1 (=
2) Sample rate converter 26i-1(= 26Two)
Realization frequency 2fi(= 2fThree) -Th (= second) transformation of
It is converted into a conversion error signal. On the other hand, the i-th (= third) band selection
Selection means 66i(= 66Three), The input signal 11
Area is fi-1~ Fi(= FTwo~ FThree), Sampling frequency is 2
fi(= FThree) -Th (= third) band signal 23i(= 2
3Three) Is extracted and its i-th (= third) band signal 23i
(= 23Three) Is the (i-1) th (= second) conversion error signal and the ith
(= Third) adder 68i(= 68Three)
The ith (= third) addition signal 29ThreeIs the ith (= third) encoder
24i(= 24Three) And the ith (= third) code
Ci(= CThree) Is output.
【００３２】このようにしてこの発明は入力信号帯域を
ｎ区間に分割して符号化する場合に適用できる。最上位
層、つまり周波数ｆn-1 〜ｆn （＝ｆ3 〜ｆ4 ）の帯域
を選出する第ｎ（＝第４）帯域選択手段６６n （＝６６
4 ）は単なる遮断周波数がｆ n-1 （＝ｆ3 ）の高域通過
フィルタ６７n （＝６７4 ）でよい。第１〜第ｎ（＝第
４）符号Ｃ1 〜Ｃn （＝Ｃ4 ）は多重化回路３１でフレ
ームごとに多重化されて符号化符号Ｃとして出力され
る。この場合多重化回路３１は第１又は第１〜第ｉ符号
の何れでも選択して出力することができるようにされ
る。As described above, the present invention reduces the input signal band.
This is applicable when encoding is performed by dividing into n sections. Top
Layer, frequency fn-1~ Fn(= FThree~ FFour) Band
N-th (= fourth) band selecting means 66 for selectingn(= 66
Four) Means that the mere cutoff frequency is f n-1(= FThree) High pass
Filter 67n(= 67Four). 1st to nth (=
4) Code C1~ Cn(= CFour) Is a frame in the multiplexing circuit 31.
Multiplexed for each frame and output as encoded code C
You. In this case, the multiplexing circuit 31 has the first or first to i-th codes.
Can be selected and output.
You.
【００３３】第１〜第ｎ（＝第４）符号器２４1 〜２４
n （＝２４4 ）は符号器２４i のｉが大となる程圧縮率
が小さくなる、という使い方を行えば広帯域、高品質の
符号化をする。これを満せばその符号化方法は、例えば
全てを変換符号化としてもよい。第１〜第４符号器２４
1 〜２４4 において聴覚重み付け符号化を行う場合はサ
ンプルレート変換器７１1 ，７１2 ，７１3 により入力
信号がそれぞれ標本化周波数が２ｆ1 ，２ｆ2 ，２ｆ3
の信号により変換されることにより、入力信号１１から
それぞれ周波数がｆ1 ，ｆ2 ，ｆ3 以下の信号が取出さ
れて聴覚重み付け係数演算部７２1 ，７２2 ，７２3 へ
それぞれ供給され、それぞれそのスペクトル包絡に基づ
く聴覚重み付け係数が演算され、また入力信号が聴覚重
み付け係数演算部７２4 に入力されて同様に聴覚重み付
け係数が演算され、これら聴覚重み付け係数演算部７２
1 〜７２4 でそれぞれ演算された聴覚重み付け係数が第
１〜第４符号器２４1 〜２４4 へ供給され、前述したよ
うに聴覚重み付け符号化が行われる。The first to n-th (= fourth) encoders 24 1 to 24
If n (= 24 4 ) is used in such a manner that the compression ratio decreases as the i of the encoder 24 i increases, wideband and high-quality encoding is performed. If this is satisfied, the encoding method may be, for example, all transform coding. First to fourth encoders 24
1 when performing perceptual weighting coding in 24 4 Sample rate converter 71 1, 71 2, 71 3 by the input signal is the sampling frequency, respectively 2f 1, 2f 2, 2f 3
By being converted by the signal, the frequency respectively from the input signal 11 is supplied to the f 1, f 2, f 3 the following signals are taken out perceptually weighted coefficient calculation unit 72 1, 72 2, 72 3, respectively its perceptual weighting coefficients based on the spectral envelope is calculated, also the input signal is perceptually weighted coefficients similarly is calculated is input to the perceptual weighting coefficient calculating unit 72 4, these perceptual weighting coefficient calculator 72
1-72 4 perceptual weighting coefficients computed respectively is supplied to the first to fourth encoder 24 1-24 4, perceptual weighting coding is performed as described above.
【００３４】この発明による復号化方法を適用した復号
化器の例として、ｎ＝４、つまり入力符号が第１〜第４
符号Ｃ1 〜Ｃ4 が入力される場合を図１１に、図８と対
応する部分に同一符号を付けて示す。符号分離手段５６
で入力符号Ｃは第１〜第４符号Ｃ1 〜Ｃ4 に分離され
て、それぞれ第１〜第４復号器５７1 〜５７4 へ供給さ
れる。第１復号器５７1 の第１復号信号５８1 は第１復
号化出力６３1 として出力されると共にサンプルレート
変換器５９1 で標本化周波数がそれぞれ２ｆ2 、第１変
換復号信号６１1 に変換され、その第１変換復号信号６
１1 は第２復号器５７2 より第２復号信号５８2 に第２
加算器６２2 で加算されて第２復号化出力６３2 として
出力されると共に第２サンプルレート変換器５９2 で標
本化周波数が２ｆ3 の変換復号信号に変換される。一般
には第ｉ−１（ｉ＝２，３，…，ｎ、例えばｉ＝３）加
算器６２i-1 （＝６２2 ）よりの第ｉ−１（＝第３）復
号化出力６３i-1 （＝６３2 ）が第ｉ−１（第２）サン
プルレート変換器５９i-1 （＝５９2 ）で標本化周波数
が２ｆi （＝２ｆ3 ）の第ｉ−１（＝第２）変換復号信
号６１i-1 （＝６１2 ）に変換され、その第ｉ−１（＝
第２）変換復号信号６１i-1 （＝６１2 ）と第ｉ（＝第
３）復号器５７i （＝５７3 ）からの第ｉ（＝第３）復
号信号５８i （＝５８3 ）とが第ｉ（＝第３）加算器６
２i （＝６２3）で加算されて第ｉ（＝第３）復号化出
力６３i （＝６３3 ）を得ることができ、これが出力さ
れる。[0034] Examples of the decoder according to the decoding how according to the present invention, n = 4, i.e. the input code is the first to fourth
FIG. 11 shows a case where the codes C 1 to C 4 are input , and the parts corresponding to those in FIG. Code separation means 56
In input code C is separated to the first to fourth code C 1 -C 4, it is supplied to the first to fourth decoders 57 1 to 57 4, respectively. The first decoded signal 58 of the first decoder 57 1 conversion with sample rate converter 59 1 sampling frequency respectively 2f 2, in the first transform decoding signal 61 1 is output as 1 first decoded output 63 And the first transformed decoded signal 6
1 1 second to 2 second decoded signal 58 from the second decoder 57 2
Adder 62 second sample rate converter 59 2 at the sampling frequency is outputted as a second decoded output 63 2 is added in 2 is converted into converted decoded signal 2f 3. Generally the first i-1 (i = 2,3, ..., n, for example, i = 3) adder 62 i-1 (= 62 2) (i-1) th than (= 3) decoded output 63 i- 1 (= 63 2) is the (i-1) (second) sample rate converter 59 i-1 (= 59 2) at a sampling frequency 2f i (= 2f 3) (i-1) th of the (= second) The converted decoded signal 61 i-1 (= 61 2 ) is converted into the converted decoded signal 61 i-1 (= 61 2 ).
Second) converting the decoded signal 61 i-1 (= 61 2 ) and the i (= 3) first i (= 3 from the decoder 57 i (= 57 3)) decoded signal 58 i (= 58 3) Is the ith (= third) adder 6
2 i (= 62 3) the i (= 3) is added in can Rukoto obtain the decoded output 63 i (= 63 3), which is output.
【００３５】[0035]
【発明の効果】以上説明したように、この発明によれ
ば、階層符号化方法において下位層の量子化誤差を上位
層で符号化しているため、ＣＥＬＰ符号化方法と変換符
号化方法などの、圧縮方法の異なる符号化方法によって
階層を構成しても、上位層までの復号信号において符号
化品質を低下させない、という効果がある。また、下位
層の量子化誤差を上位層で符号化する、という操作を繰
り返すことにより、複数階層化において量子化誤差を階
層数に応じて減少させることが可能となる。更に、この
ような符号化方法によって、どの階層で復号しても聴感
上の復号品質が最適となり、スケーラブルな階層符号化
を実現できる。As described above, according to the present invention, the quantization error of the lower layer is coded by the upper layer in the hierarchical coding method, so that the CELP coding method and the transform coding method can be used. Even if a hierarchy is configured by an encoding method having a different compression method, there is an effect that encoding quality is not deteriorated in a decoded signal up to an upper layer. In addition, by repeating the operation of encoding the quantization error of the lower layer in the upper layer, it is possible to reduce the quantization error in multiple layers according to the number of layers. Furthermore, with such an encoding method, the audible decoding quality is optimized regardless of which layer is decoded, and scalable hierarchical encoding can be realized.
【図１】サブバンド符号化方法を３つの周波数帯域に分
割する方法によって実現した場合の原音（Ａ）と符号化
再生音（Ｂ）、および量子化誤差（Ｃ）の例を示す図。FIG. 1 is a diagram showing an example of an original sound (A), an encoded reproduced sound (B), and a quantization error (C) when a subband encoding method is realized by a method of dividing into three frequency bands.
【図２】スケーラブルな階層構造を持つ階層符号化方法
の特徴を説明するための図。FIG. 2 is a diagram illustrating characteristics of a hierarchical encoding method having a scalable hierarchical structure.
【図３】サブバンド符号化方法によって階層符号化を実
現した場合の原音、復号信号、量子化誤差の様子を示す
図。FIG. 3 is a diagram showing a state of an original sound, a decoded signal, and a quantization error when hierarchical encoding is realized by a subband encoding method.
【図４】Ａはこの発明による符号化方法を２階層符号化
法に適用した場合の符号化器の例を示すブロック図、Ｂ
は多重化された符号の例を示す図である。FIG. 4A is a block diagram showing an example of an encoder when the encoding method according to the present invention is applied to a two-layer encoding method;
FIG. 3 is a diagram illustrating an example of a multiplexed code.
【図５】Ａ〜Ｄは図４Ａの符号化動作における原音、復
号信号、上位層符号化入力、上位層聴覚重み付けの基準
の各例を示す図、Ｅ，Ｆは上位層の復号信号、上位層ま
での復号の量子化誤差の例を示す図である。5A to 5D are diagrams showing examples of an original sound, a decoded signal, an upper layer coded input, and a reference for upper layer auditory weighting in the encoding operation of FIG. 4A; E and F are upper layer decoded signals; It is a figure which shows the example of the quantization error of decoding to a layer.
【図６】ＣＥＬＰ符号化器の概略を示すブロック図。FIG. 6 is a block diagram schematically showing a CELP encoder.
【図７】変換符号化器の概略を示すブロック図。FIG. 7 is a block diagram schematically showing a transform encoder.
【図８】この発明の復号化方法を２階層符号化の復号法
に適用した復号器の例を示すブロック図。FIG. 8 is a block diagram showing an example of a decoder in which the decoding method of the present invention is applied to a two-layer coding decoding method.
【図９】この発明の符号化方法を４階層符号化方法とし
て実現した場合の符号器の例を示すブロック図。FIG. 9 is a block diagram showing an example of an encoder when the encoding method of the present invention is implemented as a four-layer encoding method.
【図１０】この発明による４階層符号化方法を実現する
符号器の他の例を示すブロック図。FIG. 10 is a block diagram showing another example of an encoder for realizing the four-layer encoding method according to the present invention.
【図１１】この発明の復号化方法を４階層符号化方法と
して実現した場合の復号器の例を示すブロック図。FIG. 11 is a block diagram showing an example of a decoder when the decoding method of the present invention is realized as a four-layer coding method.
フロントページの続き (56)参考文献 特開 平８−46517（ＪＰ，Ａ) 特開 昭63−201700（ＪＰ，Ａ) 特開 平４−104617（ＪＰ，Ａ) 特開 平６−197084（ＪＰ，Ａ) 特開 平１−233496（ＪＰ，Ａ) (58)調査した分野(Int.Cl.7，ＤＢ名) G10L 19/02 Continuation of the front page (56) References JP-A-8-46517 (JP, A) JP-A-63-201700 (JP, A) JP-A-4-104617 (JP, A) JP-A-6-19784 (JP) , A) JP-A-1-233496 (JP, A) (58) Fields investigated (Int. Cl. 7 , DB name) G10L 19/02
Claims (7)
響入力信号を周波数ｆ1 ，ｆ2 ，……，ｆn-1 （ｆ1 ＜
ｆ2 ＜，……，＜ｆn-1 ＜ｆn ）のｎ個の区分（ｎは２
以上の整数）に分割して符号化する符号化方法におい
て、 上記入力信号から周波数がｆ1 以下の第１帯域信号を選
出する第１帯域選択過程と、 上記第１帯域信号を第１符号化方法で符号化して第１符
号を出力する第１符号化過程と、 第ｉ−１以下の各符号（ｉ＝２ ，３，……，ｎ）から
周波数がｆi-1 以下の第ｉ−１復号信号を得る第ｉ−１
復号化過程と、 上記入力信号から周波数ｆi 以下の第ｉ帯域信号を選出
する第ｉ選択過程と、 上記第ｉ帯域信号から上記第ｉ−１復号信号を差し引い
て第ｉ差信号を得る第ｉ差過程と、 上記第ｉ差信号を第ｉ符号化方法で符号化して第ｉ符号
を出力する第ｉ符号化過程と、 を有する音響信号符号化方法。1. A frequency f 1 the highest frequency of the acoustic input signal f n, such as tone and voice, f 2, ......, f n -1 (f 1 <
f 2 <, ......, n-number of segments (n of <f n-1 <f n ) 2
A first band selecting step of selecting a first band signal having a frequency equal to or less than f 1 from the input signal, and a first encoding of the first band signal. A first encoding step of encoding the data and outputting a first code, and an i-th code whose frequency is f i-1 or less from each code (i = 2, 3,..., N) or less. I-1 to obtain one decoded signal
A decoding step, an i-th selection step of selecting an i-th band signal having a frequency f i or less from the input signal, and a i-th difference signal obtained by subtracting the (i−1) -th decoded signal from the i-th band signal. An audio signal encoding method comprising: an i-th difference process; and an i-th encoding process of encoding the i-th difference signal by an i-th encoding method and outputting an i-th code.
符号を復号する過程と、その復号された信号と第ｉ−２
復号信号とを加算する過程と、その加算された信号を標
本化周波数が２ｆi の信号に変換して上記第ｉ−１復号
信号を得る過程と、 を有することを特徴とする請求項１記載の音響信号符号
化方法。2. The method according to claim 1, wherein the (i-1) th decoding process is performed in the (i-1) th decoding process.
Decoding the code, the decoded signal and i-2
A step of adding the decoded signal, according to claim 1, characterized in that it comprises the steps of obtaining the (i-1) th decoded signal and converts the summed signal into a signal of sampling frequency is 2f i, the Audio signal encoding method.
響入力信号を、周波数ｆ1 ，ｆ2 …，ｆn-1 （ｆ1 ＜ｆ
2 ＜，…＜ｆn-1 ＜ｆn ）（ｎ＝２以上の整数）で区分
してそれぞれを符号化する符号化方法において、 上記入力信号より標本化周波数が２ｆ1 の第１帯域信号
を得る第１帯域選択過程と、 上記第１帯域信号を第１符号化法により符号化して第１
符号を出力する第１符号化過程と、 上記ｉ−１符号化過程（ｉ＝２，３，…，ｎ）の符号誤
差として第ｉ−１誤差信号を得る第ｉ−１誤差取出し過
程と、 上記第ｉ−１誤差信号を標本化周波数が２ｆi の第ｉ−
１変換誤差信号に変換する第ｉ−１変換過程と、 上記入力音響信号より周波数帯域がｆi-1 〜ｆi 、標本
化周波数が２ｆi の第ｉ帯域信号を得る第ｉ帯域選出過
程と、 上記第ｉ−１変換誤差信号と上記第ｉ帯域信号とを加算
して第ｉ加算信号を得る第ｉ加算過程と、 上記第ｉ加算信号を第ｉ符号化法により符号化して第ｉ
符号を出力する第ｉ符号化過程と、 を有する音響信号符号化方法。3. A musical sound, the acoustic input signal maximum frequency is f n, such as voice, frequency f 1, f 2 ..., f n-1 (f 1 <f
2 <,... <F n-1 <f n ) (n = an integer equal to or greater than 2), and encodes each of the first band signals having a sampling frequency of 2f 1 from the input signal. And a first band selecting step of obtaining the first band signal, and encoding the first band signal by a first encoding method.
A first encoding step of outputting a code, an i-1 error extracting step of obtaining an i-1 error signal as a code error of the i-1 encoding step (i = 2, 3,..., N); The (i-1) -th error signal is converted to an i-th error signal having a sampling frequency of 2f i .
And the i-1 conversion process for converting the 1 conversion error signal, a frequency band from the input audio signal f i-1 ~f i, and the i band selection process sampling frequency to obtain a first i-band signal of 2f i An i-th addition step of adding the i-th conversion error signal and the i-th band signal to obtain an i-th addition signal, and encoding the i-th addition signal by an i-th encoding method to obtain an i-th addition signal.
An i-th encoding step of outputting a code.
号化法であり、上記第ｎ符号化法は変換符号化法である
ことを特徴とする請求項１乃至３の何れかに記載の音響
信号符号化方法。4. The method according to claim 1, wherein the first encoding method is a code-driven linear predictive encoding method, and the n-th encoding method is a transform encoding method. Audio signal encoding method.
までの各々が変換符号化法であることを特徴とする請求
項１乃至３の何れかに記載の音響信号符号化方法。 5. The method according to claim 1, wherein said first encoding method is to said n-th encoding method.
Each of which is a transform coding method
Item 4. The audio signal encoding method according to any one of Items 1 to 3.
ほぼ全域の成分のスペクトル包絡を重みの基準として、
上記第ｉ符号化過程において心理聴覚重み付け量子化を
行うことを特徴とする請求項１乃至５の何れかに記載の
音響信号符号化方法。6. The spectral envelope of components of substantially the entire frequency range below the frequency f i in the acoustic input signal is defined as a weight reference.
Acoustic signal encoding method according to any of claims 1 to 5, characterized in that the perceptual weighting quantization in the i-th encoding process.
上の整数）に分離する分離過程と、 上記第１符号を復号して、標本化周波数２ｆ1 の第１復
号信号を第１復号化出力として出力する第１復号過程
と、 上記第ｉ−１復号化出力（ｉ＝２，３，…，ｎ）を標本
化周波数が２ｆi の第ｉ−１変換復号化出力に変換する
第ｉ−１変換過程と、 上記第ｉ符号を復号して標本化周波数２ｆi の第ｉ復号
信号を得る第ｉ復号過程と、 上記第ｉ復号信号と上記第ｉ−１変換復号化出力とを加
算して第ｉ復号化出力を出力する第ｉ加算過程と、 を有する音響信号復号化方法。7. A separation process of separating an input code into first to n-th codes (n is an integer of 2 or more), and decoding the first code to convert a first decoded signal of a sampling frequency 2f 1 into a first code. a first decoding step of outputting as a decoded output, said first i-1 decoded output (i = 2,3, ..., n ) converting the sampling frequency to the (i-1) th converted decoded output 2f i I-th conversion process, i-th decoding process of decoding the i-th code to obtain an i-th decoded signal at a sampling frequency of 2f i , And an i-th adding step of adding an i and i to output an i-th decoded output.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP07065622A JP3139602B2 (en) | 1995-03-24 | 1995-03-24 | Acoustic signal encoding method and decoding method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP07065622A JP3139602B2 (en) | 1995-03-24 | 1995-03-24 | Acoustic signal encoding method and decoding method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| JPH08263096A JPH08263096A (en) | 1996-10-11 |
| JP3139602B2 true JP3139602B2 (en) | 2001-03-05 |
Family
ID=13292314
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| JP07065622A Expired - Lifetime JP3139602B2 (en) | 1995-03-24 | 1995-03-24 | Acoustic signal encoding method and decoding method |
Country Status (1)
| Country | Link |
|---|---|
| JP (1) | JP3139602B2 (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009513992A (en) * | 2003-06-25 | 2009-04-02 | ドルビー スウェーデン アクチボラゲット | Apparatus and method for encoding audio signal and apparatus and method for decoding encoded audio signal |
Families Citing this family (32)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3134817B2 (en) | 1997-07-11 | 2001-02-13 | 日本電気株式会社 | Audio encoding / decoding device |
| DE19743662A1 (en) * | 1997-10-02 | 1999-04-08 | Bosch Gmbh Robert | Bit rate scalable audio data stream generation method |
| JP3541680B2 (en) | 1998-06-15 | 2004-07-14 | 日本電気株式会社 | Audio music signal encoding device and decoding device |
| US6549147B1 (en) | 1999-05-21 | 2003-04-15 | Nippon Telegraph And Telephone Corporation | Methods, apparatuses and recorded medium for reversible encoding and decoding |
| US6446037B1 (en) * | 1999-08-09 | 2002-09-03 | Dolby Laboratories Licensing Corporation | Scalable coding method for high quality audio |
| SE0202159D0 (en) | 2001-07-10 | 2002-07-09 | Coding Technologies Sweden Ab | Efficientand scalable parametric stereo coding for low bitrate applications |
| US8605911B2 (en) | 2001-07-10 | 2013-12-10 | Dolby International Ab | Efficient and scalable parametric stereo coding for low bitrate audio coding applications |
| ES2237706T3 (en) | 2001-11-29 | 2005-08-01 | Coding Technologies Ab | RECONSTRUCTION OF HIGH FREQUENCY COMPONENTS. |
| JP4290917B2 (en) | 2002-02-08 | 2009-07-08 | 株式会社エヌ・ティ・ティ・ドコモ | Decoding device, encoding device, decoding method, and encoding method |
| EP1484841B1 (en) | 2002-03-08 | 2018-12-26 | Nippon Telegraph And Telephone Corporation | DIGITAL SIGNAL ENCODING METHOD, DECODING METHOD, ENCODING DEVICE, DECODING DEVICE and DIGITAL SIGNAL DECODING PROGRAM |
| CN100346392C (en) | 2002-04-26 | 2007-10-31 | 松下电器产业株式会社 | Device and method for encoding, device and method for decoding |
| JP3881943B2 (en) * | 2002-09-06 | 2007-02-14 | 松下電器産業株式会社 | Acoustic encoding apparatus and acoustic encoding method |
| SE0202770D0 (en) | 2002-09-18 | 2002-09-18 | Coding Technologies Sweden Ab | Method of reduction of aliasing is introduced by spectral envelope adjustment in real-valued filterbanks |
| KR100513729B1 (en) | 2003-07-03 | 2005-09-08 | 삼성전자주식회사 | Speech compression and decompression apparatus having scalable bandwidth and method thereof |
| US7844451B2 (en) | 2003-09-16 | 2010-11-30 | Panasonic Corporation | Spectrum coding/decoding apparatus and method for reducing distortion of two band spectrums |
| JP4679049B2 (en) | 2003-09-30 | 2011-04-27 | パナソニック株式会社 | Scalable decoding device |
| US7949518B2 (en) | 2004-04-28 | 2011-05-24 | Panasonic Corporation | Hierarchy encoding apparatus and hierarchy encoding method |
| WO2006030865A1 (en) * | 2004-09-17 | 2006-03-23 | Matsushita Electric Industrial Co., Ltd. | Scalable encoding apparatus, scalable decoding apparatus, scalable encoding method, scalable decoding method, communication terminal apparatus, and base station apparatus |
| KR20070061818A (en) * | 2004-09-17 | 2007-06-14 | 마츠시타 덴끼 산교 가부시키가이샤 | Audio encoding apparatus, audio decoding apparatus, communication apparatus and audio encoding method |
| WO2006057626A1 (en) * | 2004-11-29 | 2006-06-01 | National University Of Singapore | Perception-aware low-power audio decoder for portable devices |
| WO2006096099A1 (en) * | 2005-03-09 | 2006-09-14 | Telefonaktiebolaget Lm Ericsson (Publ) | Low-complexity code excited linear prediction encoding |
| KR100818268B1 (en) * | 2005-04-14 | 2008-04-02 | 삼성전자주식회사 | Apparatus and method for audio encoding/decoding with scalability |
| FR2888699A1 (en) * | 2005-07-13 | 2007-01-19 | France Telecom | HIERACHIC ENCODING / DECODING DEVICE |
| CN101385079B (en) * | 2006-02-14 | 2012-08-29 | 法国电信公司 | Device for perceptual weighting in audio encoding/decoding |
| EP2017830B9 (en) * | 2006-05-10 | 2011-02-23 | Panasonic Corporation | Encoding device and encoding method |
| JP5403949B2 (en) * | 2007-03-02 | 2014-01-29 | パナソニック株式会社 | Encoding apparatus and encoding method |
| WO2009093466A1 (en) * | 2008-01-25 | 2009-07-30 | Panasonic Corporation | Encoding device, decoding device, and method thereof |
| US20120041761A1 (en) | 2009-03-13 | 2012-02-16 | Panasonic Corporation | Voice decoding apparatus and voice decoding method |
| JPWO2010103854A1 (en) | 2009-03-13 | 2012-09-13 | パナソニック株式会社 | Speech coding apparatus, speech decoding apparatus, speech coding method, and speech decoding method |
| CA2759914A1 (en) | 2009-05-29 | 2010-12-02 | Nippon Telegraph And Telephone Corporation | Encoding device, decoding device, encoding method, decoding method and program therefor |
| JP5031006B2 (en) * | 2009-09-04 | 2012-09-19 | パナソニック株式会社 | Scalable decoding apparatus and scalable decoding method |
| CN109215670B (en) * | 2018-09-21 | 2021-01-29 | 西安蜂语信息科技有限公司 | Audio data transmission method and device, computer equipment and storage medium |
-
1995
- 1995-03-24 JP JP07065622A patent/JP3139602B2/en not_active Expired - Lifetime
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2009513992A (en) * | 2003-06-25 | 2009-04-02 | ドルビー スウェーデン アクチボラゲット | Apparatus and method for encoding audio signal and apparatus and method for decoding encoded audio signal |
Also Published As
| Publication number | Publication date |
|---|---|
| JPH08263096A (en) | 1996-10-11 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP3139602B2 (en) | Acoustic signal encoding method and decoding method | |
| JP3881943B2 (en) | Acoustic encoding apparatus and acoustic encoding method | |
| JP4934427B2 (en) | Speech signal decoding apparatus and speech signal encoding apparatus | |
| JP4871894B2 (en) | Encoding device, decoding device, encoding method, and decoding method | |
| JP3391686B2 (en) | Method and apparatus for decoding an encoded audio signal | |
| JP3566652B2 (en) | Auditory weighting apparatus and method for efficient coding of wideband signals | |
| KR101303145B1 (en) | A system for coding a hierarchical audio signal, a method for coding an audio signal, computer-readable medium and a hierarchical audio decoder | |
| JP4662673B2 (en) | Gain smoothing in wideband speech and audio signal decoders. | |
| JP4081447B2 (en) | Apparatus and method for encoding time-discrete audio signal and apparatus and method for decoding encoded audio data | |
| JP4958780B2 (en) | Encoding device, decoding device and methods thereof | |
| KR19990077753A (en) | Audio signal coding apparatus, audio signal decoding apparatus, and audio signal coding and decoding apparatus | |
| JP3344962B2 (en) | Audio signal encoding device and audio signal decoding device | |
| US8036390B2 (en) | Scalable encoding device and scalable encoding method | |
| JP5236040B2 (en) | Encoding device, decoding device, encoding method, and decoding method | |
| US9230551B2 (en) | Audio encoder or decoder apparatus | |
| US20100250260A1 (en) | Encoder | |
| JP3186007B2 (en) | Transform coding method, decoding method | |
| JP2002330075A (en) | Subband adpcm encoding/decoding method, subband adpcm encoder/decoder and wireless microphone transmitting/ receiving system | |
| Jayant et al. | Coding of wideband speech | |
| JP2004302259A (en) | Hierarchical encoding method and hierarchical decoding method for sound signal | |
| JPH09127985A (en) | Signal coding method and device therefor | |
| JP4373693B2 (en) | Hierarchical encoding method and hierarchical decoding method for acoustic signals | |
| JPH09127987A (en) | Signal coding method and device therefor | |
| JPH09127998A (en) | Signal quantizing method and signal coding device | |
| JP3504485B2 (en) | Tone encoding device, tone decoding device, tone encoding / decoding device, and program storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20071215 Year of fee payment: 7 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20081215 Year of fee payment: 8 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20091215 Year of fee payment: 9 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20101215 Year of fee payment: 10 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20101215 Year of fee payment: 10 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20111215 Year of fee payment: 11 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20111215 Year of fee payment: 11 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20121215 Year of fee payment: 12 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20121215 Year of fee payment: 12 |
|
| FPAY | Renewal fee payment (event date is renewal date of database) |
Free format text: PAYMENT UNTIL: 20131215 Year of fee payment: 13 |
|
| S531 | Written request for registration of change of domicile |
Free format text: JAPANESE INTERMEDIATE CODE: R313531 |
|
| R350 | Written notification of registration of transfer |
Free format text: JAPANESE INTERMEDIATE CODE: R350 |
|
| EXPY | Cancellation because of completion of term |