Audio Decoder Loss Concealment Using Spectral and Temporal Cues
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio codecs lack an effective mechanism to select the most suitable packet loss concealment method based on audio signal characteristics, leading to noticeable artifacts during signal recovery.
Innovation Solution
An audio decoder that determines a spectral position and temporal predictability of the audio signal to assign the most appropriate loss concealment tool from a set, using measures such as spectral centroid and temporal self-similarity to choose between pitch-based and tonal frequency domain concealment methods.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single packet loss concealment method is used, then the device complexity is reduced, but the quality of signal recovery deteriorates because the method cannot adapt to different audio characteristics
Solution Approach 1:
The patent implements dynamic selection of concealment methods by calculating signal characteristics (spectral centroid, temporal predictability) for each lost frame and choosing the most appropriate method based on current signal conditions. This allows the system to adapt to different audio characteristics (speech, music, noise) in real-time, improving recovery quality without requiring a fixed complex structure for all possible signal types
Solution Approach 2:
The patent changes the operational parameters of the concealment system by using measurable signal features (spectral centroid position, temporal predictability values) to determine which concealment method to apply. This parameter-based selection mechanism enables the system to optimize recovery quality for different signal types by adjusting the choice of method based on quantifiable signal properties
2Adaptability or versatility
If multiple packet loss concealment methods are supported, then the adaptability to different audio characteristics is improved, but the device complexity increases due to the need for multiple concealment tools and selection mechanisms
Solution Approach 1:
The patent segments the concealment system into distinct specialized methods (pitch-based, frame repetition, tonal component prediction) each optimized for specific signal types. By dividing the overall concealment function into these specialized segments and selecting only the appropriate one based on signal characteristics, the system achieves high adaptability while managing complexity through modular design
Solution Approach 2:
The patent introduces intermediary measurement mechanisms (spectral centroid calculation, temporal predictability assessment) that act as mediators between the audio signal and the concealment method selection. These intermediaries provide objective criteria for choosing the appropriate method, reducing the complexity of direct signal analysis while maintaining high adaptability to different audio characteristics
3Ease of operation
If pitch-based PLC is used for all signals, then the ease of operation is improved, but the quality of concealment deteriorates for non-periodic and polyphonic signals
Solution Approach 1:
The patent applies local quality by matching specific concealment methods to local signal characteristics. Instead of uniformly applying pitch-based PLC to all signals, the system calculates local features (spectral centroid, temporal predictability) for each lost frame and selects the concealment method that is locally optimal for that specific signal segment, thereby maintaining simplicity where pitch-based methods work well while improving quality where they don't
Data Source
AI summary
An assignment of one of phase set of different loss concealment tools of an audio decoder to a portion of the audio signal to be decoded from a data stream, which portion is affected by loss, that is the selection out of the set of different loss concealment tools, may be made in a manner leading to a more pleasant loss concealment if the assignment/selection is done based on two measures: A first measure which is determined measures a spectral position of a spectral centroid of a spectrum of the audio signal and a second measure which is determined measures a temporal predictability of the audio signal. The assigned or selected loss concealment tool may then be used to recover the portion of the audio signal.