Audio Decoder Loss Concealment Using Spectral and Temporal Cues
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio decoders lack an effective mechanism to select the most suitable loss concealment tool for different audio signal characteristics during packet loss, leading to noticeable artifacts in the reconstructed audio signal.
Innovation Solution
An audio decoder that selects loss concealment tools based on two measures: the spectral position of the spectral centroid and the temporal predictability of the audio signal, assigning tools such as pitch-based or tonal frequency domain concealment methods to minimize annoyance from packet loss.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If a single loss concealment tool is used in audio decoders, then device complexity is reduced, but loss concealment quality deteriorates for signals with characteristics mismatching the fixed tool
Solution Approach 1:
The patent implements dynamic selection of loss concealment tools based on real-time analysis of audio signal characteristics. The system transitions from a static single-tool approach to a dynamic multi-tool approach where the appropriate concealment method is automatically selected according to the detected signal type (speech, music, noise), ensuring optimal concealment quality for varying audio content while maintaining manageable complexity through automated decision-making
Solution Approach 2:
The patent changes the operational parameters of the loss concealment system by introducing multiple concealment tools with different characteristics and selecting among them based on signal parameters. The system analyzes spectral characteristics, temporal properties, and other signal parameters to determine which concealment tool best matches the current audio signal, thereby adapting the concealment approach to match the specific parameters of the lost signal
2Reliability
If multiple loss concealment tools are added to handle different signal types, then loss concealment quality improves, but device complexity increases due to the need for signal analysis and tool selection mechanisms
Solution Approach 1:
The patent segments the loss concealment function into multiple specialized tools, each optimized for specific signal types (speech concealment, music concealment, noise concealment). By dividing the overall concealment task into specialized sub-tasks handled by different tools, the system achieves high concealment quality for each signal type while organizing complexity into manageable, purpose-specific modules rather than requiring one complex universal solution
Solution Approach 2:
The patent implements feedback mechanisms where the system continuously analyzes the audio signal characteristics and uses this information to select the appropriate concealment tool. The analysis of spectral properties, temporal patterns, and other signal features provides feedback that drives the selection decision, creating a closed-loop system that adapts to the actual signal content and maintains optimal concealment performance
Solution Approach 3:
The system performs self-service by automatically analyzing signal characteristics and selecting the appropriate concealment tool without external intervention. The built-in signal analysis capabilities enable the decoder to autonomously determine the signal type and choose the matching concealment method, reducing the need for complex external control mechanisms and simplifying the overall system architecture
3Measurement precision
If loss concealment tools are selected based on comprehensive signal analysis, then concealment accuracy improves, but processing time increases
Solution Approach 1:
The patent applies partial action by performing signal analysis to the extent necessary for accurate tool selection without over-analyzing. The system identifies the key distinguishing features of different signal types and performs analysis only on those critical parameters needed to differentiate between speech, music, and noise, avoiding unnecessary computational steps while maintaining sufficient accuracy for reliable concealment tool selection
Data Source
Figure 1
Figure 2
Figure 3
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.