| 摘要 |
<p>The invention concerns a method of detecting and, if necessary, correcting or masking errors in a digitized data-reduced audio signal consisting of a sequence of data blocks which are to be transmitted or stored, the individual data blocks being, for instance, bit-allocation data, scale-factor selection data, scale factors or encoded sampling values from part-band filtering or transformation in the spectral region, such as Fourier transforms. In this method, one or more received bit patterns for the data units of a data block are graded by their probability of occurrence and, if the probability is low, declared to be erroneous and, if necessary, corrected or masked.</p> |
