Sub-Band PCM Error Concealment for Residual DAB Bit Errors

The patent divides the audio frame into multiple sub-band signals (e.g., 32 sub-bands) and processes error concealment independently for each sub-band. This segmentation allows the system to apply error concealment at a finer granularity, reducing the need to store entire frames while still effectively masking errors. Each sub-band can be concealed using local prediction from adjacent sub-bands, eliminating the need for large memory buffers required by frame-level techniques.

The patent applies different error concealment strategies to different sub-bands based on their specific characteristics. Instead of uniformly applying frame-level concealment across the entire audio frame, the system independently processes each sub-band, using local prediction from neighboring sub-bands. This localized approach reduces memory requirements while maintaining audio quality, as each sub-band is concealed using only locally available information rather than requiring storage of entire frames.

The patent implements a feedback mechanism where the decoded sub-band signals are used to generate prediction values that are compared with the actual received signals. The prediction error is analyzed to detect residual errors that escaped channel decoding correction. This feedback loop enables the system to identify and conceal errors at the sub-band level, providing an additional layer of error detection beyond what channel decoding alone can achieve.

The patent performs preliminary error detection and concealment at the sub-band level before final audio reconstruction. By analyzing prediction errors in each sub-band and applying concealment techniques proactively, the system prevents residual errors from propagating through the audio stream. This preliminary action complements channel decoding correction by addressing errors that slip through the initial correction stage.

The patent divides the audio signal into multiple sub-bands using efficient filter bank decomposition, which can be implemented using standard DSP techniques. This segmentation enables precise error detection in each sub-band by comparing predicted and actual values independently. The use of sub-band decomposition is a well-established technique in audio processing, allowing the system to achieve high error detection precision without requiring fundamentally new processing architectures.

The patent replaces complex error detection mechanisms with a simpler prediction-based approach. Instead of using elaborate error detection algorithms, the system uses linear prediction to generate expected sub-band values and compares these with actual received values. This substitution of mechanical error detection with a prediction-comparison mechanism reduces device complexity while maintaining high detection precision, as the prediction process naturally highlights anomalies.