Audio Signal Loss Compensation Using Spectral Masking Data

The patent converts the harmful effect of masking-induced data loss into a benefit by analyzing the masking threshold and selectively reconstructing only those lost signals that are perceptible to human hearing. The system uses the masking characteristics to identify which lost components would actually be audible, then prioritizes reconstruction efforts on those specific components, turning the previously harmful data loss into a targeted reconstruction opportunity that improves overall sound quality.

The patent changes the parameter of signal reconstruction by introducing a loss signal compensation parameter that dynamically adjusts the reconstruction process. Instead of uniformly reconstructing all lost signals, the system modifies the reconstruction parameters based on masking threshold analysis, scale factor information, and signal characteristics (speech vs. audio), thereby optimizing the balance between data loss compensation and sound quality enhancement.

The patent applies partial action by selectively compensating for loss signals rather than attempting to reconstruct all lost data. The system performs masking threshold analysis to identify which lost signals are perceptible and prioritizes reconstruction of those specific components. This partial reconstruction approach achieves sufficient signal loss compensation while maintaining low bitrate, as it avoids the excessive computational and bandwidth requirements of full signal reconstruction.

The patent implements preliminary action by performing masking threshold analysis and loss signal detection during the encoding phase. The encoder pre-calculates the masking threshold, identifies lost signals that should be compensated, and transmits compensation parameters to the decoder. This preliminary analysis enables the decoder to efficiently reconstruct only the necessary lost components without requiring excessive bitrate, as the reconstruction strategy is predetermined by the encoder's analysis.

The patent applies local quality by treating different frequency regions and signal types differently in the masking and reconstruction process. The system analyzes masking thresholds locally across the frequency spectrum and applies selective reconstruction only in regions where lost signals are perceptible. Additionally, the patent applies different reconstruction strategies based on local signal characteristics (speech vs. audio), thereby minimizing sound quality loss in critical regions while maintaining aggressive bitrate reduction in less critical regions.

The patent introduces dynamics by making the masking and reconstruction process adaptive rather than static. The system dynamically adjusts the reconstruction strategy based on real-time analysis of signal characteristics, masking thresholds, and user preferences. The decoder can adaptively select between different reconstruction modes (e.g., speech-oriented vs. audio-oriented) based on the detected signal type, thereby optimizing the balance between bitrate reduction and sound quality maintenance for each specific input signal.

The patent implements dynamics by creating a dynamic signal classification and routing system. The decoder continuously analyzes incoming signals to detect whether they exhibit speech characteristics or general audio characteristics, and dynamically switches between speech-coded decoding and audio-coded decoding pathways. This dynamic adaptation enables the system to achieve high decoding accuracy for speech signals while maintaining the capability to process general audio signals, thereby resolving the contradiction between specialization and versatility.

The patent achieves universality by designing a multi-functional decoder architecture that can handle both speech and general audio signals through a single unified system. The decoder incorporates signal classification capabilities that route different signal types to appropriate decoding pathways, and includes loss signal compensation mechanisms that work effectively for both speech and audio. This multi-functional design eliminates the need for separate dedicated decoders for speech and audio, thereby maintaining versatility while preserving decoding accuracy for speech through specialized processing pathways.