Method of reducing crosstalk in processing of acoustic or optical signals

1. Method for reducing frequency crosstalk in processing of digitized input signals, in which input signals are subdivided into partial frequency bands by means of series connected transformation stages, for subsequent encoding, transmission or storage, followed by decoding, the signals from the partial bands being again combined to form a signal, said method comprising the steps of:

prior to said encoding, for each partial band performing a weighted summation of a partial band signal thereof, which signal contains spectral components of specific frequencies as signal components, and other partial band signals from partial bands in which said spectral components of said specific frequencies occur as crosstalk components in a stop range;

processing the partial band signals by means of an inverse operation to the weighted summation, following decoding.

2. Method according to claim 1, wherein a first transformation stage of said series connected transformation stages subdivides the input signal into an even number M of partial bands 0 to M-1; and means are provided for causing an output signal of uneven numbered partial bands of said stage to undergo a correction operation before entering a second transformation stage.

3. Method according to claim 2, wherein said correction operation comprises multiplying every second time consecutive value of output signals of said uneven numbered partial bands by -1, while the output signals of the even numbered partial bands remain unchanged.

4. Method according to claim 3, wherein weighted signals y.sub.i in the partial band i are obtained in accordance with the following: ##EQU5## in which x.sub.i is a partial band signal of partial band i, and c.sub.m and d.sub.m represent weight factors, d.sub.m being determined from the weight factors c.sub.m by the above equation.

5. Method according to claim 2, wherein weighted signals y.sub.i in the partial band i are obtained in accordance with the following: ##EQU6## in which x.sub.i is a partial band signal of partial band i, and c.sub.m and d.sub.m represent weight factors, d.sub.m being determined from the weight factors c.sub.m by the above equation.

6. Method according to claim 1, wherein weighted signals y.sub.i i in the partial band i are obtained in accordance with the following: ##EQU7## in which x.sub.i is a partial band signal of partial band i, c.sub.m and d.sub.m represent weight factors, d.sub.m being determined from the weight factors c.sub.m by the above equation, N is the number of partial bands in said second transformation stage, k is a partial band index for partial bands of the first transformation stage and m is a partial band index for partial bands of the second transformation stage.

7. Method according to claim 6, wherein said inverse operation is performed according to the following equation:

x.sub.i '=d.sub.m .multidot.(y.sub.i '-c.sub.m .multidot.y.sub.j ')

x.sub.j '=d.sub.m .multidot.(y.sub.j '-c.sub.m .multidot.y.sub.i ')

in which x.sub.i ' is the partial band signal of partial band i after performing inverse operation y.sub.i ', the weighted partial band signal of partial band i, following encoding and decoding.

8. Method according to claim 6, wherein the weight factors c.sub.m for summation are optimized with respect to the frequency response of the series-connected stages.

9. Method according to claim 8, wherein the number of weight factors c.sub.m to be used for summation is half as large as a number of partial bands produced by a following stage.

10. Method according to claim 6, wherein the number of weight factors c.sub.m to be used for summation is half as large as a number of partial bands produced by a following stage.

11. Method according to claim 1, wherein said inverse operation is performed according to the following equation:

x.sub.i '=d.sub.m .multidot.(y.sub.i '-c.sub.m .multidot.y.sub.j ')

x.sub.j '=d.sub.m .multidot.(y.sub.j '-c.sub.m .multidot.y.sub.i ')

in which x.sub.i ' is the partial band signal of partial band i after performing the inverse operation, and y.sub.i ' is the weighted partial band signal of partial band i, following encoding and decoding.

12. Method according to claim 11, wherein the number of weight factors c.sub.m to be used for summation is half as large as a number of partial bands produced by a following stage.