CN1345031A - Subband filtering and delaying estimation and correction method for audio data wave packet encoder - Google Patents
Subband filtering and delaying estimation and correction method for audio data wave packet encoder Download PDFInfo
- Publication number
- CN1345031A CN1345031A CN01134555A CN01134555A CN1345031A CN 1345031 A CN1345031 A CN 1345031A CN 01134555 A CN01134555 A CN 01134555A CN 01134555 A CN01134555 A CN 01134555A CN 1345031 A CN1345031 A CN 1345031A
- Authority
- CN
- China
- Prior art keywords
- delay
- filter
- signal
- sub
- sequence
- 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.)
- Granted
Links
Images
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
The invention relates to a rapid developing branch in modern information processing area, i.e., data compression technique, especially a method and equipment used to evaluate and modify the filtering delay of subband in audio data wave packet encoder. The characteristics of the inventino are as follows. The decomposed sequence signal gets across relative reconstruction filter for full wave-let packet. The subband filter delay of wave-let packet in relative layer can be obtained by comparing reconstructed sequence with original sequence. The maximum delay can be determined, after the delay of each subband filter is determined. The modification can be made by inserting some 'zero signal' values into the front end of input signal sequence of sub band filter based on the difference value between the delay the subband filter and the maximum delay.
Description
Technical field
The present invention relates to the branch-data compression technique of a fast development in the present information process field, the method and apparatus that the subband filter delay is estimated and revised in particularly a kind of voice data wavelet packet scrambler.
Background technology
The present invention is primarily aimed at the improvement of wavelet package reconstruction algorithm, in the compression algorithm such as MPEG of existing sound signal, all adopted sub-band division to carry out compressed encoding, obtain the frequency spectrum of signal by rapid fourier change, utilize the multiphase filter group (multiple phase filter bands) of equiband that signal is handled again, obtain 32 sample of signal on the wide subband, and then each sub-band samples is compressed by psychoacoustic model.Lot of experiments to human auditory characteristics shows, human auditory system can mark off the combinations of bands of a non-equiband by frequency distribution, people's ear has tangible difference to the sensitivity of sound in each frequency band range, and these frequency band ranges are called critical subband (critical subbands).The advantage of equiband multiphase filter group is that the exponent number of each sub-filter is identical, does not have delay issue in the calculating, and wave filter has anti-aliasing preferably character; Its problem is the sub-band division of sound signal and people's ear to bring adverse influence to utilizing psychoacoustic model to compress to different these features of sensitivity of different frequency scope sound signal and be not suitable for.Wavelet analysis has been the very fast number credit branch of development since the fifties in this century, and small echo and wavelet package transforms have signal time-frequency localization to be analyzed and the multiresolution analysis ability, has greatly promoted the development of Digital Signal Processing.
Content of the present invention
How based on the wavelet packet technical construction based on the non-wide sub-filter of human auditory's feature to replace the equiband sub-filter in the existing algorithm, after the sub-band division of more being pressed close to critical subband, can utilize psychoacoustic model to compress better, guarantee further to improve ratio of compression under the transparent situation of reconstruct tonequality, this is the wound purpose of grinding of the present invention.
The exponent number that a key issue of the non-wide multiphase filter group of small echo is each sub-filter might be inequality, source signal is producing different delays by meeting behind each sub-filter like this, if can not accurately obtain the length of delay of each sub-filter and revise, this wavelet packet compression algorithm just can't realize accurate reconstruct, influences acoustical quality and ratio of compression.
Because the core procedure of wavelet packet compression algorithm is to adopt small echo multiphase filter group that source signal is carried out Filtering Processing, how the delay of each sub-band filter is estimated and is revised just to become a key issue that must solve, the purpose of this invention is to provide the method and apparatus that subband filter delay in a kind of voice data wavelet packet scrambler is estimated and revised, accurately estimation and correction provide a simple and reliable method for the delay of sub-filter in the wavelet packet compression.
Ultimate principle of the present invention is: the delay of a wave filter is the character decision by this wave filter, and it is irrelevant with signal, in wavelet packet compression because the needs of non-wide sub-band division, must adopt inclined to one side wavelet packet binary tree, based on aforementioned starting point, can think that the delay that is in the pairing sub-filter of leafy node on the same level in the wavelet packet binary tree is identical, its value is basically by its exponent number decision, accurate estimation can obtain completely setting from setting the identical wavelet packet y-bend of the degree of depth partially with this wavelet packet y-bend: it is the length of delay that the corresponding wave filter of full binary tree of N produces that the length of delay of the sub-filter that the leafy node on a certain level N of inclined to one side binary tree that is promptly adopted in the wavelet packet compression process is corresponding equals the degree of depth just, for this reason, as long as after the length of delay of full binary tree wave filter determined, just can obtain the length of delay of the sub-filter of the leafy node correspondence on a certain level N of inclined to one side binary tree, sub-filter to all leafy node correspondences on the inclined to one side binary tree carries out same processing, can obtain the length of delay of each sub-filter in the wavelet packet multiphase filter group, make the accurate reconstruct that can guarantee source signal after the corresponding correction.
Purpose of the present invention can realize by following mode, the present invention passes through full wavelet decomposition wave filter with signal, then the sequence of decomposing is passed through corresponding full wavelet reconstruction wave filter, since reconstruction signal compare with original signal except the reconstruction signal head end more zero, other is identical, the sequence and the original signal contrast of reconstruct just can be obtained the delay of the wavelet sub-band wave filter of the corresponding number of plies; After having determined the delay of each sub-filter, determine maximum-delay, and determine the delay of each sub-filter and the difference of maximum-delay that the zero-signal numerical value of the difference same number input signal sequence front end that inserts this sub-filter is revised therewith.
Compare with modification method with delay estimation commonly used, it is simple that method of the present invention has principle, and it is convenient to realize, the accurate and reliable advantage of the length of delay of resultant each sub-filter.Prove that through a large amount of numerical experiments the reconstruct to the WAVELET PACKET DECOMPOSITION coefficient after the correction of each sub-filter length of delay has reached extraordinary effect, reconstruction value can reach 10 to approaching of source signal value under the double precision datum type
-7Precision owing to postpone to estimate and the solution of the problem of correction, realized the accurate decomposition and reconstruction that source signal and Selection of Wavelet Basis have nothing to do, have great practical significance.
Below exemplify embodiment and further illustrate the present invention.
Description of drawings
Fig. 1 is the inclined to one side binary tree synoptic diagram of wavelet packet of the present invention.
Fig. 2 is the wavelet packet full binary tree synoptic diagram corresponding with Fig. 1.
Specific implementation of the present invention
Biorthogonal wavelet base with 10 rank is that example describes, establish the inclined to one side binary tree of the wavelet packet that adopts as shown in Figure 1.With the corresponding wavelet packet full binary tree of the inclined to one side binary tree of last wavelet packet as shown in Figure 2:
Fig. 1 is three layers of inclined to one side binary tree of wavelet packet, and the filter delay corresponding with the full tree of N=3 among the length of delay of 1,2 respective filter of subband and Fig. 2 is identical; Subband 3,4 is identical with the corresponding filter delay of the full tree of N=2 among Fig. 2 with 5 among Fig. 1, and the delay of the small echo binary tree of other different depth asks method identical therewith.
With DB10 rank small echo is example: it is 1243 points that the reconstruction signal of its 2 rank full binary tree postpones; It is 1207 points that the reconstruction signal of its 3 rank full binary tree postpones.
Then for the wavelet reconstruction tree of Fig. 1, the delay of its subband 1,2 is 1135 points, the delay of its subband 3,4,5 is 1207 points, then when reconstruct, add that at the front end of reconstruct small echo zero of respective number forms new wavelet filter and just can obtain perfect reconstruct effect.
Claims (1)
1, the method that the subband filter delay is estimated and revised in a kind of voice data wavelet packet scrambler, it is characterized in that signal by full wavelet decomposition wave filter, then the sequence of decomposing is passed through corresponding full wavelet reconstruction wave filter, since reconstruction signal compare with original signal except the reconstruction signal head end more zero, other is identical, the sequence and the original signal contrast of reconstruct just can be obtained the delay of the wavelet sub-band wave filter of the corresponding number of plies; After having determined the delay of each sub-filter, determine maximum-delay, and determine the delay of each sub-filter and the difference of maximum-delay that the zero-signal numerical value of the difference same number input signal sequence front end that inserts this sub-filter is revised therewith.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN01134555A CN1123864C (en) | 2001-11-02 | 2001-11-02 | Subband filtering and delaying estimation and correction method for audio data wave packet encoder |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN01134555A CN1123864C (en) | 2001-11-02 | 2001-11-02 | Subband filtering and delaying estimation and correction method for audio data wave packet encoder |
Publications (2)
Publication Number | Publication Date |
---|---|
CN1345031A true CN1345031A (en) | 2002-04-17 |
CN1123864C CN1123864C (en) | 2003-10-08 |
Family
ID=4672585
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN01134555A Expired - Fee Related CN1123864C (en) | 2001-11-02 | 2001-11-02 | Subband filtering and delaying estimation and correction method for audio data wave packet encoder |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN1123864C (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN100463465C (en) * | 2002-12-25 | 2009-02-18 | 日本电信电话株式会社 | Estimation method and apparatus of overall conversational speech quality, program and recording medium for realizing the method |
WO2010108315A1 (en) * | 2009-03-24 | 2010-09-30 | 华为技术有限公司 | Method and device for switching a signal delay |
-
2001
- 2001-11-02 CN CN01134555A patent/CN1123864C/en not_active Expired - Fee Related
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN100463465C (en) * | 2002-12-25 | 2009-02-18 | 日本电信电话株式会社 | Estimation method and apparatus of overall conversational speech quality, program and recording medium for realizing the method |
WO2010108315A1 (en) * | 2009-03-24 | 2010-09-30 | 华为技术有限公司 | Method and device for switching a signal delay |
Also Published As
Publication number | Publication date |
---|---|
CN1123864C (en) | 2003-10-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Sinha et al. | Audio compression at low bit rates using a signal adaptive switched filterbank | |
US7953605B2 (en) | Method and apparatus for audio encoding and decoding using wideband psychoacoustic modeling and bandwidth extension | |
KR100896737B1 (en) | Device and method for robustry classifying audio signals, method for establishing and operating audio signal database and a computer program | |
DE60000412T2 (en) | DATA FRAME STRUCTURE FOR ADAPTIVE BLOCK LENGTH CODING | |
EP0797313B1 (en) | Switched filterbank for use in audio signal coding | |
DE69833834T2 (en) | Scalable audio coding and decoding method and device | |
CN100508030C (en) | Improving quality of decoded audio by adding noise | |
EP1080542B1 (en) | System and method for masking quantization noise of audio signals | |
JP3277682B2 (en) | Information encoding method and apparatus, information decoding method and apparatus, and information recording medium and information transmission method | |
CN101149925B (en) | Space parameter selection method for parameter stereo coding | |
CN101894560B (en) | Reference source-free MP3 audio frequency definition objective evaluation method | |
CN102201242A (en) | Encoder, decoder, encoding method, and decoding method | |
KR20090110244A (en) | Method for encoding/decoding audio signals using audio semantic information and apparatus thereof | |
KR100389487B1 (en) | Information encoding method and apparatus, Information decoding method and apparatus, and information transmission method | |
CN101436407A (en) | Method for encoding and decoding audio | |
CN101436406B (en) | Audio encoder and decoder | |
CN1123864C (en) | Subband filtering and delaying estimation and correction method for audio data wave packet encoder | |
JPH09101799A (en) | Signal coding method and device therefor | |
CN1318904A (en) | Practical sound coder based on wavelet conversion | |
Sinha et al. | A fractal self-similarity model for the spectral representation of audio signals | |
JP3413691B2 (en) | Information encoding method and device, information decoding method and device, and information recording medium and information transmission method | |
JP3297238B2 (en) | Adaptive coding system and bit allocation method | |
Gunjal et al. | Traditional Psychoacoustic Model and Daubechies Wavelets for Enhanced Speech Coder Performance | |
Sablatash et al. | Compression of high-quality audio signals, including recent methods using wavelet packets | |
KR101333162B1 (en) | Tone and speed contorol system and method of audio signal using imdct input |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C06 | Publication | ||
PB01 | Publication | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant | ||
PP01 | Preservation of patent right |
Effective date of registration: 20051209 Pledge (preservation): Preservation |
|
PD01 | Discharge of preservation of patent |
Pledge (preservation): Preservation Release effective date: 20060609 |
|
PP01 | Preservation of patent right |
Effective date of registration: 20061102 Pledge (preservation): Preservation |
|
PD01 | Discharge of preservation of patent |
Date of cancellation: 20090812 Pledge (preservation): Preservation registration |
|
C17 | Cessation of patent right | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20031008 |