Audio Transmission With Neural FEC Selection Under Packet Loss
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The instability of Internet-based voice transmission (VoIP) due to packet loss leads to unstable quality, and existing FEC redundant encoding solutions increase bandwidth consumption, particularly in bandwidth-constrained scenarios, causing network congestion and further packet losses.
Innovation Solution
A packet loss recovery capability prediction neural network model determines the need for redundant encoding based on audio encoding feature parameters, predicting the quality loss of decoded audio when data is lost, and selectively applying FEC to generate redundancy packets.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If FEC redundant encoding is applied to all encoded data to achieve packet loss concealment, then packet loss recovery capability is improved, but network bandwidth consumption increases
Solution Approach 1:
The patent applies different encoding strategies to different portions of encoded data based on their individual characteristics. The neural network evaluates each encoded data packet's vulnerability to packet loss and applies FEC redundant encoding only to packets with high vulnerability (low packet loss recovery capability), while skipping packets with low vulnerability. This selective application optimizes bandwidth usage while maintaining effective packet loss concealment where it matters most.
Solution Approach 2:
Instead of applying FEC redundant encoding to all encoded data packets (excessive action), the patent uses a neural network to identify and apply encoding only to the necessary subset of packets that benefit most from protection (partial action). This partial application reduces overall bandwidth consumption while maintaining adequate packet loss recovery capability for critical data.
2Reliability
If stronger packet loss concealment enhancement is applied, then packet loss recovery capability is improved, but network bandwidth consumption increases leading to network congestion
Solution Approach 1:
The patent implements a dynamic encoding strategy where the neural network continuously evaluates the packet loss recovery capability of each encoded data packet and adjusts the application of FEC redundant encoding in real-time. This dynamic approach allows the system to adapt to varying network conditions and data characteristics, applying strong concealment enhancement only where and when needed, thereby avoiding network congestion while maintaining effective packet loss recovery.
Solution Approach 2:
The system changes the parameter of FEC encoding application from a static all-or-nothing approach to a dynamic selective approach based on neural network evaluation. By changing how the encoding parameter is applied (selectively rather than universally), the system achieves effective packet loss concealment without excessive bandwidth consumption that would cause network congestion.
Data Source
AI summary
An audio transmission method is described. A packet loss recovery capability corresponding to current encoded data of an audio signal can be determined according to a first audio encoding feature parameter corresponding to the current encoded data and a second audio encoding feature parameter corresponding to previous or subsequent encoded data of the current encoded data using a packet loss recovery capability prediction neural network model. Whether to perform redundant encoding for the current encoded data can be determined according to the packet loss recovery capability. In response to the redundant encoding being determined to be performed, the redundant encoding can be performed according to the current encoded data to generate a corresponding redundancy packet.