Fraunhofer-Gesellschaft Zur Foerderung Der Angewandten Forschung E.V.

Pierre Lauber

Martin Dietz

Ralph Sperschneider

Juergen Herre

Michael Schug

Karlheinz Brandenburg

Heinz Gerhaeuser

Daniel Homm

Roland Bitto

Reinhold Boehm

Ali Nowbakht-Irani

In a method for producing a data stream of code words of variable lengths which are divided up into a plurality of sets of code words, wherein a raster having segments is specified for the data stream, wherein two adjacent raster points define a segment, code words of the first sets are written into the data stream starting at raster points. Subsequently, code words of the second set are written into the data stream pursuant to a predetermined assignment rule, wherein each code word of the second set is assigned to a different segment. Entire code words or parts of code words which cannot be written according to their assignment, are stored and entered into the data stream in further attempts, wherein the assignment by a predetermined regulation is changed from attempt to attempt. This procedure is analogously repeated for any further set which may be present. Thus, the ends of code words of the second set are decoupled from the starts of following code words of the second set, since the corresponding code words of a set are written segment by segment, which results in a reduction of error propagation.

Method and device for generating a data flow from variable-length code words and a method and device for reading a data flow from variable-length code words

In a method for concealing an error in an encoded audio signal a set of spectral coefficients is subdivided into at least two sub-bands (), whereupon the sub-bands are subjected to a re-verse transform (). A specific prediction is performed () for each quasi time signal of a sub-band to obtain an estimated temporal representation for a sub-band of a set of spectral coefficients following the current set. A forward transform () of the time signal of each sub-band provides estimated spectral coefficients which can be used () instead of erroneous spectral coefficients of a following set of spectral coefficients, e.g. in order to conceal transmission errors. Transforming at the sub-band level provides independence from transform characteristics such as block length, window type and MDCT algorithm while at the same time preserving spectral processing for error concealment. Thus the spectral characteristics of audio signals can also be taken into account during error concealment.

Method and device for error concealment in an encoded audio-signal and method and device for decoding an encoded audio signal

In a method for coding an audio signal to obtain a coded bit stream, discrete-time samples of the audio signal are transformed into the frequency domain to obtain spectral values. The spectral values are coded with a code table having a limited number of code words of different lengths to obtain spectral values coded by code words, the length of a code word assigned to a spectral value being that much shorter the higher the probability of occurrence of the spectral value is. A raster is then specified for the coded bit stream, the raster having equidistant raster points and the distance between the raster points depending on the code table(s) used. In order to obtain error-tolerant Huffman coding, priority code words, which represent particular spectral values which are psychoacoustically more important than other spectral values, are so arranged in the raster that the start of each priority code word coincides with a raster point.

Methods and devices for coding or decoding an audio signal or bit stream

Growing community of inventors

Nuremberg, Germany

Pierre Lauber