Classifications

• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
  • G10L19/04—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using predictive techniques
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
  • G10L19/04—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using predictive techniques
  • G10L19/16—Vocoder architecture
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
  • G10L19/04—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using predictive techniques
  • G10L19/16—Vocoder architecture
  • G10L19/18—Vocoders using multiple modes
  • G10L19/24—Variable rate codecs, e.g. for generating different qualities using a scalable representation such as hierarchical encoding or layered encoding
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
  • G10L19/02—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis using spectral analysis, e.g. transform vocoders or subband vocoders
  • G10L19/022—Blocking, i.e. grouping of samples in time; Choice of analysis windows; Overlap factoring
  • G10L19/025—Detection of transients or attacks for time/frequency resolution switching
• • G—PHYSICS
  • G10—MUSICAL INSTRUMENTS; ACOUSTICS
  • G10L—SPEECH ANALYSIS TECHNIQUES OR SPEECH SYNTHESIS; SPEECH RECOGNITION; SPEECH OR VOICE PROCESSING TECHNIQUES; SPEECH OR AUDIO CODING OR DECODING
  • G10L19/00—Speech or audio signals analysis-synthesis techniques for redundancy reduction, e.g. in vocoders; Coding or decoding of speech or audio signals, using source filter models or psychoacoustic analysis
  • G10L2019/0001—Codebooks
  • G10L2019/0013—Codebook search algorithms
  • G10L2019/0014—Selection criteria for distances

Definitions

• the present invention relates to predictive coding of information signals, e.g. Audio signals, and especially adaptive to predictive coding.
• a predictive encoder - or transmitter - encodes signals by predicting a current or current value of the signal to be coded by the past or previous values of the signal. In the case of linear prediction, this prediction or presumption is made about the current value of the signal by a weighted sum of the past values of the signal.
• the prediction weights or also prediction coefficients are continuously adapted or adapted to the signal, so that the difference between the predicted signal and the actual signal is minimized in a predetermined manner.
• the prediction coefficients are optimized, for example, with respect to the square of the prediction error.
• the error criterion when optimizing the predictive encoder or predictor can also be selected differently. Instead of using the least squares criterion, the spectral flatness of the error signal, i.e. differences or residuals.
• the prediction weights for the prediction can be adapted to the signal at a predetermined speed. There is a parameter for this in the so-called least mean squares (LMS) algorithm.
• LMS least mean squares
• This parameter determines how quickly the prediction coefficients adapt to an optimal set of prediction coefficients, a non-optimally adapted set of prediction coefficients leading to the prediction being less accurate and therefore the prediction errors being larger, which in turn results in an increased bit rate for the transmission of the signal, since small values or small prediction errors or differences can be transmitted with fewer bits than larger ones.
• the predictors i.e. the prediction algorithms, on the transmitter side and on the reception side, given the same times for both sides, are reset to a specific state, which is referred to as a reset.
• the object of the present invention is therefore to create a scheme for predictive coding of an information signal which on the one hand has a sufficient robustness against errors in the difference values or residuals of the coded information signal and on the other hand a lower increase associated therewith the bit rate or reduction in signal quality.
• the present invention is based on the knowledge that the previous fixed setting of the speed parameter of the adaptive prediction algorithm, which is the basis of a predictive coding, has to be abandoned, towards a variable setting of this parameter.
• an adaptive prediction algorithm is assumed which can be controlled by a speed coefficient in order to work and in with a first adaptation speed and a first adaptation precision and an associated first prediction precision in the event that the speed coefficient has a first value
• the adaptation time periods occurring after the reset times can be in which the prediction errors are initially increased due to the as yet unadapted prediction coefficients, decrease by the speed parameter initially is set to the first value and after a while to the second value.
• the speed parameter is set to the second value again after a predetermined period of time after the reset times, the prediction errors and thus the residuals to be transmitted are optimized or smaller than would be possible with the first speed parameter value.
• the present invention is based on the knowledge that prediction errors after reset times can be minimized by the speed parameters, such as e.g.
• the step size parameter of an LS algorithm is changed for a certain period after the reset times in such a way that the speed of the adaptation of the weights increases for this period - with, of course, reduced precision.
• FIG. 1 is a block diagram of a predictive encoder according to an embodiment of the present invention.
• Fig. 2 is a block diagram illustrating the operation of the encoder of Fig. 1;
• FIG. 3 shows a block diagram of a decoder corresponding to the encoder of FIG. 1 according to an embodiment of the present invention
• Fig. 4 is a flow chart illustrating the operation of the decoder of Fig. 3;
• FIG. 5 shows a block diagram of the prediction device of FIGS. 1 and 3 according to an exemplary embodiment of the present invention
• FIG. 6 is a block diagram of the transversal filter of FIG. 5 according to an embodiment of the present invention
• FIG. 7 shows a block diagram of the adaptation control from FIG. 5 according to an exemplary embodiment of the present invention.
• FIG. 8 shows a diagram to illustrate the behavior of the prediction device from FIG. 5 for two different fixed speed parameters.
• the encoder 10 has an input 12 at which it receives the information signal s to be encoded and an output 14 at which it outputs the encoded information signal ⁇ .
• the information signal can be any signal, such as an audio signal, a video signal, a measurement signal or the like.
• the information signal s consists of a sequence of information values s (i) with ielN, that is to say audio values, pixel values, measured values or the like.
• the coded information signal ⁇ is composed of a sequence of difference values or residuals ⁇ (i) with ielN, which correspond to the signal values s (i) in the manner described below.
• the encoder 10 comprises a prediction device 16, a subtractor 18 and a control device 20.
• the prediction device 16 is connected to the input 12 in order, as will be described in more detail below, to predict a current signal value s (n) Determine value s' (n) from previous signal values s (m) with m ⁇ n and melN and output it at an output, which in turn is connected to an inverting input of subtractor 18.
• a non-inverting input of the subtractor 18 is also connected to the input 12 to subtract the predicted value s' (m) from the actual signal value s (n) - or simply to make the difference between the two values - and the result at the output 14 to output as difference value ⁇ (n).
• the prediction device 16 implements an adaptive prediction algorithm. In order to be able to carry out the adaptation, it therefore receives the difference value ⁇ (n) - also called prediction error - via a feedback path 22 at a further input.
• the prediction device 16 comprises two control inputs which are connected to the control device 20. Via these control inputs, the control device 20 is able to initialize prediction coefficients or filter coefficients ⁇ i of the prediction device 16 at specific points in time, as will be described in the following, and to change a speed parameter of the prediction algorithm on which the prediction device 16 is based, which is described below is denoted by ⁇ .
• the control device 20 first initializes the prediction or filter coefficients ⁇ ⁇ of the prediction device 16.
• the initialization after step 40 takes place at predetermined reset times.
• the reset times or more precisely the signal value numbers n r at which a reset was carried out after step 40 can occur, for example, at fixed time intervals from one another.
• the reset times can be reconstructed on the decoder side, for example by incorporating information about them into the coded information signal ⁇ or by standardizing the fixed time interval or the fixed number of signal values between them.
• the coefficients (Di are set to any values which, for example, are the same at every reset time, ie each time step 40 is carried out.
• the prediction coefficients are initialized in step 40 to values which heuristically have been derived from typical representative information signals and in this regard, on average, that is, via the representative set of information signals, such as, for example, a mixture of jazz, classical, rock, etc. music pieces, gave an optimal set of prediction coefficients.
• the control device 20 sets the speed parameter ⁇ to a first value, the steps 40 and 42 preferably being carried out essentially at the same time as the reset times.
• the setting of the speed parameter to the first value has the consequence that the prediction device 16 carries out a rapid adaptation of the prediction coefficients ⁇ > i initialized in step 40 - with admittedly reduced adaptation precision.
• the prediction device 16 and the subtractor 18 then act together as a prediction device in order to encode the information signal s and in particular the current signal value s (n) by predicting the same by adapting the prediction coefficients C ⁇ I.
• step 44 comprises several sub-steps, namely the determination by prediction device 16 of a predicted value s' (n) for the current signal value s (n) using previous signal values s (m) with m ⁇ n using the current prediction coefficient cöi, subtracting the predicted value s' (n) from the actual signal value s (n) by the subtractor 18, outputting the resulting difference value ⁇ (n) at the output 14 as part of the coded information signal ⁇ and that Adaptation or adaptation of the coefficients ⁇ > i by the prediction device 16 on the basis of the prediction error or difference value ⁇ (n), which it receives via the feedback path 22.
• the prediction device 16 uses the speed parameter ⁇ predetermined or set by the control device 20 for the adaptation or adaptation of the prediction coefficients C ⁇ I , which, as will be described in more detail below with reference to the exemplary embodiment of an LMS algorithm, determines how strongly the feedback prediction errors ⁇ (n) per adaptation iteration, here n, are involved in the adaptation or update of the prediction coefficients ⁇ > i or how strongly the prediction coefficients cöi depend on the prediction error ⁇ (n) per adaptation iteration, ie per feedback ⁇ (n), can change.
• step 46 the control device 20 then checks whether the speed parameter ⁇ should be changed or not.
• the determination in step 46 can be carried out in several ways. For example, the control device 20 determines that a speed parameter change should be carried out if since a predetermined period of time has passed after the initialization or setting in steps 40 or 42. Alternatively, the control device 20 evaluates in step 46 an adaptation degree, such as the approximation to an optimal set of coefficients c ⁇ i with a corresponding low mean prediction error, of the prediction device 16, as will be explained in more detail below.
• step 46 the control device 20 checks in a step 48 whether there is a reset time again, ie a time at which the prediction coefficients should be reinitialized for reasons of resynchronization. First, it is again assumed that there is no reset time. If there is no reset time, the prediction device 16 continues with the coding of the next signal value, as indicated by "n-» n + l ⁇ in Fig. 2.
• the coding of the information signal s is adapted the prediction coefficient ⁇ > i continues with the adaptation speed, as set by the speed parameter ⁇ , until finally, when the loop 44, 46, 48 passes through in step 46, the control device 20 determines that a change in the speed parameter should be carried out In this case, the control device 20 sets the speed parameter ⁇ to a second value in a step 50.
• the setting of the speed parameter ⁇ to the second value has the result that the prediction device 16 adapts the step 44 when the loop 44-48 is run through Prediction coefficients coi from now on with a lower adaptation rate , but with an increased adaptation precision, so that in these runs following the time of the speed parameter change, which refer to subsequent signal values of the information signal s, the resulting residuals become smaller than ⁇ (n), which in turn enables an increased compression rate when integrating the values ⁇ (n) into the coded signal.
• control device 20 After further runs of the loop 44-48, the control device 20 recognizes a reset point in step 48 at some point, whereupon the functional sequence begins again in step 40.
• the decoder includes an input 62 for receiving the coded information signal ⁇ consisting of the difference values or residuals ⁇ (n), an output 64 for outputting the decoded information signal s which, except for rounding errors in the representation of the difference values ⁇ (n), corresponds to the original information signal s (n) and accordingly from a sequence of decoded signal values s (n), a prediction device 66 which is identical to or that of the encoder 10 of FIG.
• An input of the prediction device 66 is connected to the output 64 in order to obtain already decoded signal values s (n). From these already decoded signal values s (m) with m ⁇ n, the prediction device 66 determines a predicted value s' (n) for a signal value s (n) to be currently decoded and outputs this predicted value to a first input of the adder 68.
• a second input of adder 68 is connected to input 62 to sum the predicted value s' (n) with the difference value ⁇ (n) and the result or sum to output 64 as part of the decoded signal s and an output the input of the prediction device 66 for predicting the next signal value.
• Another input of the prediction device 66 is connected to the input 62 in order to obtain the difference value ⁇ (n), which uses this value to adapt the current prediction coefficients cöi.
• the prediction coefficients ⁇ x can be initialized by the control device 70, just as the speed parameter ⁇ can be varied by the control device 70.
• the operation of the decoder 60 will now be described below with reference to FIGS. 3 and 4.
• the control device 70 first initializes the prediction coefficients coi of the prediction device 66 and sets the speed parameter ⁇ thereof to a first value which corresponds to a higher adaptation speed but a reduced adaptation precision.
• step 94 the prediction device 66 then decodes the coded information signal ⁇ or the current difference value ⁇ (n) by predicting the information signal by adapting the prediction coefficients cöi. More specifically, step 94 comprises several substeps. First, the prediction device 66, which is aware of the already decoded signal values s (m) with m ⁇ n, predicts the signal value to be currently determined from them in order to obtain the predicted value s' (n). The prediction device 66 uses the current prediction coefficients cöi. The difference value ⁇ (n) currently to be decoded is added by the adder 68 to the predicted value s' (n) in order to output the sum thus obtained as part of the decoded signal s at the output 64.
• the prediction device 66 uses this value s (n) in the next predictions.
• the prediction device 66 uses the difference value ⁇ (n) from the coded signal stream to adapt the current prediction coefficients coi, the adaptation speed and the adaptation precision being predetermined by the currently set speed parameter ⁇ . In this way, the prediction coefficients ⁇ > i are updated or adapted.
• step 96 the control device checks whether a speed change in parameters. If this is not the case, the control device 70 determines in a step 98 corresponding to step 48 whether there is a reset time. If this is not the case, the loop of steps 94-98 is run through again, this time for the next signal value s (n) or the next difference value ⁇ (n), as indicated by "n—» n + l ⁇ in FIG 4 is indicated.
• step 96 the control device 70 sets the speed parameter ⁇ to a second value in a step 100, which corresponds to a lower adaptation speed but a higher adaptation precision, as has already been discussed with regard to the coding ,
• FIGS. 5-7 After a predictive coding scheme according to an exemplary embodiment of the present invention has been described in general terms with reference to FIGS. 1-4, a special exemplary embodiment for the prediction device 16 is described with reference to FIGS. 5-7, according to which exemplary embodiment namely the prediction device 16 works according to an LMS adaptation algorithm.
• the prediction device 16 comprises an input 120 for signal values s (n), an input 122 for prediction errors or difference values ⁇ (n), two control inputs 124 and 126 for the initialization of the coefficients öi or the setting of the speed parameter ⁇ and an output 128 for outputting the predicted value s' (n).
• prediction device 16 includes a transversal filter 130 and an adaptation control 132.
• Transversal filter 130 is connected between input 120 and output 128.
• the adaptation controller 132 is connected to the two control inputs 124 and 126 and furthermore to the inputs 120 and 122 and furthermore comprises an output in order to forward correction values ⁇ i for the coefficients (Ü ⁇ to the transversal filter 130.
• the LMS algorithm which is implemented by the prediction device 16, possibly in conjunction with the subtractor 18 (FIG. 1), is a linear adaptive filter algorithm which, generally speaking, consists of two basic processes:
• a filtering process which (a) calculates the output signal s' (n) of a linear filter in response to an input signal s (n) by the transversal filter 130 and (b) generates an estimation error ⁇ (n) by comparing the output signal s '(n) with a desired response s (n) by the sub-emitter 18 or the obtaining of the estimation error ⁇ (n) from the coded information signal ⁇ .
• An adaptive process which is carried out by the adaptation controller 132 and has an automatic adaptation of the filter coefficients Cöi of the transversal filter 130 in accordance with the estimation error ⁇ (n).
• the transversal filter 130 receives the sequence of signal values s (n) at an input 140.
• the input 140 is followed by a series connection of m delay elements 142, so that the signal values s (nl) ... s (nm) which precede the current signal value s (n) are present at connection nodes between the m delay elements 142.
• the weighting devices 144 output their result to a respective one of a plurality of summers 146, which are connected in series, so that at an output 148 of the transversal filter 130, the sum of the last summer of the series connection the estimated value or predicted value s' (m)
• the estimate s' (n) in an environment that is stationary in the broader sense comes close to a value predicted by the Vienna solution when the number of iterations reaches n infinitely.
• the adaptation controller 132 is shown in more detail in FIG. 7.
• the adaptation controller 132 accordingly comprises an input 160, at which the sequence of difference values ⁇ (n) is received. These are multiplied in a weighting device 162 by the speed parameter ⁇ , which is also referred to as step size parameter.
• the result is fed to a plurality of m multiplication devices 164, which multiply the same by one of the signal values s (nl) ... s (nm).
• the results of the multipliers 164 form correction values ⁇ cöi ... ⁇ m . Consequently, the correction values ⁇ > i ... ⁇ m represent a scalar version of the inner product of the estimation error ⁇ (n) and the vector of signal values s (nl) ...
• the scaling factor ⁇ which is used in the adaptation controller 132 and, as already mentioned, also referred to as the step width parameter, can be regarded as a positive quantity and should meet certain conditions relative to the spectral content of the information signal so that the LMS Algorithm implemented by means 16 of Figures 5-7 is stable. Stability here means that with increasing n, that is to say if the adaptation is carried out for an infinitely long time, the mean square error which is generated by the filter 130 reaches a constant value. An algorithm that fulfills this condition is said to be stable on a quadratic average.
• a change in the speed parameter ⁇ causes a change in the adaptation precision, ie in the precision, since the coefficients (can be adapted to an optimal set of coefficients.
• a mismatch of the filter coefficients leads to an increase in the mean square error or the energy in the difference values ⁇ in the steady state n—> Imony.
• the feedback loop which acts on the weights öi, behaves like a low-pass filter whose determination time constant is inversely proportional to the parameter ⁇ . Consequently, by setting the parameter ⁇ to a small value, the adaptive process slows down, with most of the effects of gradient noise on the weights cöi being filtered out, which in turn has the effect of reducing the mismatch.
• FIGS. 5-7 shows the influence of the setting of the parameter ⁇ on different values ⁇ i and ⁇ 2 on the adaptation behavior of the prediction device 16 from FIGS. 5-7 on the basis of a graph in which the number of iterations n or the number along the x-axis the predictions and adaptations n and the mean energy of the residual values ⁇ (n) or the mean square of the error is plotted along the y axis.
• a solid line refers to a speed parameter ⁇ i.
• a dashed curve results, whereby, as can be seen, fewer iterations, namely n 2 , are required until the steady state is reached, but the steady state with a higher energy E 2 the residual values is linked.
• the steady state at Ei or E 2 is not characterized by a settling of the mean square of the error of the residual values or residuals to an asymptotic value, but also by a settling of the filter coefficients ö with a certain, in the case of ⁇ i higher and in the case of ⁇ 2 lower, precision based on the optimal set of filter coefficients.
• the present invention can also be used in connection with other adaptive prediction algorithms in which a speed parameter is used the adjustment between the adaptation speed on the one hand and the adaptation precision on the other hand can be adjusted. Since the adaptation precision in turn influences the energy of the residual values, the speed parameter can always be set so that the adaptation speed is high, whereupon the same is set to a value at which the adaptation speed is low but the adaptation precision and therefore the energy of the residual values is lower. With such prediction algorithms, for example, there would be no connection between input 120 or adaptation fault 132.
• triggering can also be carried out depending on the degree of adaptation, e.g. a triggering of a change in speed parameters if the coefficients are correct ⁇ , e.g. a sum of the absolute values thereof falls below a certain value, which indicates an approximation to the quasi-stationary state, as shown in FIG. 8, to a certain degree of approximation.
• the scheme according to the invention can also be implemented in software.
• the implementation can be on a digital storage medium, in particular a floppy disk or a CD with electronically readable control signals that can interact with a programmable computer system in such a way that the corresponding method is carried out.
• the invention thus also consists in a computer program product with program code stored on a machine-readable carrier for carrying out the method according to the invention when the computer program product runs on a computer.
• the invention can thus be implemented as a computer program with a program code for carrying out the method if the computer program runs on a computer.

Landscapes

• Engineering & Computer Science (AREA)
• Human Computer Interaction (AREA)
• Acoustics & Sound (AREA)
• Signal Processing (AREA)
• Health & Medical Sciences (AREA)
• Audiology, Speech & Language Pathology (AREA)
• Multimedia (AREA)
• Physics & Mathematics (AREA)
• Computational Linguistics (AREA)
• Quality & Reliability (AREA)
• Compression, Expansion, Code Conversion, And Decoders (AREA)
• Management, Administration, Business Operations System, And Electronic Commerce (AREA)
• Compression Or Coding Systems Of Tv Signals (AREA)
• Reduction Or Emphasis Of Bandwidth Of Signals (AREA)
• Train Traffic Observation, Control, And Security (AREA)

Applications Claiming Priority (2)

Publications (2)

Family

ID=34625823

Family Applications (1)

Country Status (17)

Families Citing this family (6)

Family Cites Families (21)

• 2004
  • 2004-02-13 DE DE102004007185A patent/DE102004007185B3/de not_active Expired - Lifetime
  • 2004-12-20 DE DE502004003807T patent/DE502004003807D1/de active Active
  • 2004-12-20 EP EP04804095A patent/EP1700293B1/fr active Active
  • 2004-12-20 PT PT04804095T patent/PT1700293E/pt unknown
  • 2004-12-20 CA CA2556024A patent/CA2556024C/fr active Active
  • 2004-12-20 JP JP2006552473A patent/JP4351260B2/ja active Active
  • 2004-12-20 ES ES04804095T patent/ES2285551T3/es active Active
  • 2004-12-20 AU AU2004316541A patent/AU2004316541B2/en active Active
  • 2004-12-20 KR KR1020067016186A patent/KR100852483B1/ko active IP Right Grant
  • 2004-12-20 CN CN2004800415752A patent/CN1914670B/zh active Active
  • 2004-12-20 AT AT04804095T patent/ATE362169T1/de active
  • 2004-12-20 WO PCT/EP2004/014496 patent/WO2005083683A1/fr active IP Right Grant
  • 2004-12-20 BR BRPI0418389-4A patent/BRPI0418389B1/pt active IP Right Grant
  • 2004-12-20 RU RU2006132731/09A patent/RU2345426C2/ru active
• 2006
  • 2006-07-27 IL IL177124A patent/IL177124A/en active IP Right Grant
  • 2006-08-03 US US11/462,140 patent/US7386446B2/en active Active
  • 2006-09-07 NO NO20064021A patent/NO338722B1/no unknown
• 2007
  • 2007-02-01 HK HK07101128A patent/HK1094080A1/xx unknown

Non-Patent Citations (1)

Also Published As

Similar Documents

Legal Events