Method for the transmission of a signal

1. A signal transmission method in which an analog signal is converted into a digital signal, transmitted in digital form and reconverted into an analog signal and whereby said digital signal is partitioned by means of overlapping time windows in temporally successive blocks which are each converted into a signal sequence representing a short-time spectrum, said method comprising the steps of

a) windowing said digital signal and forming temporal blocks of block length TB with overlapping regions of relative size N-1/N where N=[2 to the power of n] 2.sup.n for whole numbers of n, whereby overlapping blocks are selected each having a fixed number of samples;

b) transforming said samples of each block by subjecting individual ones of said blocks offset by TB/2*N in the time domain to a sine or cosine transformation respectively to produce transform output values;

c) sampling said transform output values at different positions, causing aliasing, said sampling comprising sampling said transform output values from N consecutive values according to a sampling scheme: C1: 0, N, 2N, 3N . . . , or a subsampling scheme C2: N/2, 3N/2 [/ whereby four combinations with respect to transformation and sub-scanning forms result:] resulting in four combinations of

transformation and sampling forms K1-K4 as follows

K1: cosine transformation+selection scheme C1

K2: cosine transformation+selection scheme C2

K3: sine transformation+selection scheme C1

K4: sine transformation+selection scheme C2;

d) arranging sampled transform output values so that said aliasing arises at predetermined positions, by applying one of said combinations K1 . . . K4 in any arbitrary permutation on each of overlapping blocks, whereby four groups of values with entries for overlapping regions of blocks results, said groups of values being differentiated by their preceding signs;

e) said arranging including selecting a combination from K1 . . . K4 so that after inverse transformation at a receiver and summation of the components in the signal segments of the blocks involved in the overlap, all signals not originating from the same segment of the original signal are compensated whereby said aliasing is cancelled in a receiver;

f) coding, transmission and decoding said sampled/sub-sampled transform output values of individual blocks after said arranging;

g) subjecting decoded individual blocks to inverse sine or cosine transformation, producing inverse transform values;

h) arranging said inverse transform values so that non-aliased components corresponding to original input blocks are located in respective original positions, said arranging of said inverse transform values including segmentation of continuous-time inverse transform signal values into successive signal segments Ni for i=1, 2, 3 . . . , which, depending on the combination K1 . . . K4 used, contain components Ni . . . and temporally reflected alias components Si . . . ; and

i) summing said signal segment components Ni . . . and said temporally reflected alias components Si . . . in the signal segments Ni of overlapping blocks.

2. A method according to claim 1, wherein:

a. said blocks are evaluated using analysis windows prior to transformation and synthesis windows after transformation, and said windows form segments of length TB/N equal to said signal segments;

b. said evaluation is performed by multiplying said signal components Ni . . . or, respectively, said temporally reflected components Si . . . corresponding to said signal components by components of analysis window ani . . . or, respectively, temporally reflected components asi . . . corresponding to these components and components of synthesis window sni . . . ; and

c. said analysis and synthesis windows fulfill the following conditions in block overlap regions:

I. the sum of temporal regular wanted signal components of an analysis window and synthesis window superimposed in signal segments of a block is equal to one;

II. the sum of temporal regular aliasing components of an analysis window and synthesis window superimposed in signal segments of a block is equal to zero,

3. A method according to claim 2, wherein with said windows symmetric about a block center and with double overlapping, a synthesis window function is determined from a pre-determined analysis window function according to the following equation: ##EQU7## for [0 less or x less or 0.5] 0.ltoreq.x.ltoreq.0.5 where s(x) is a synthesis window function,

a(x) is an analysis window function,

x represents a standardized time with a value of 0 at the start of a block and a value of 1 at the end of said block.

4. A method according to claim 2, wherein with said windows symmetric about a block center and with quadruple overlapping, an analysis window function is first determined which fulfills the following equations:

a(0.5-x)*a(x)=a(0.25-x)*a(0.25-x)

a(1-x)*a(x-0.5)=a(0.75-x)*a(0.75-x)

for 0.ltoreq.x.ltoreq.0.25

where a(x) is an analysis window function, and a synthesis window function is determined from a previously determined analysis window function according to the following equation: ##EQU8## where sni are components of a synthesis window function, ani are components of an analysis window function, x represents a standardized time with a value of 0 at the start of a block and a value of 1 at the end of said block.

5. A method according to claim 2, wherein with said windows symmetric about a block center and with eightfold overlapping, an analysis window function is first determined which fulfills the following equations:

a(x)*(0.25-x)=a(0.125)+x)*a(0.125)-x)

a(x)*a(0.5-x)+a(0.125+x)*a(0.375-x)+a(0.25+x)*a(0.25-x)+a(0.375-x)*a(0.125-x)=0 ##EQU9## for 0.ltoreq.x.ltoreq.0.125 where a(x) is an analysis window function, and a synthesis window function is determined from a previously determined analysis window function according to the following system of equations: ##EQU10## where sni are components of a synthesis window function, ani are components of an analysis window function, x represents a standardized time with a value of 0 at the start of a block and a value of 1 at the end of said block.