Neural Network Audio Channel Parameter Prediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing audio coding technologies face limitations in compression performance and quality due to structural restrictions in audio signal conversion.
Innovation Solution
A method and device that predict channel parameters of an original signal from a downmix signal using a machine learning-based algorithm, generating input and label feature maps, and applying them to a neural network to improve compression performance while maintaining signal quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If existing audio coding technology is used, then audio signal transmission is achieved, but compression performance is limited due to structural restrictions
Solution Approach 1:
The patent replaces traditional mechanical audio coding structures with a neural network-based machine learning system. The neural network learns optimal channel parameter predictions from training data, substituting rigid structural constraints with adaptive computational models that achieve superior compression performance without structural limitations.
Solution Approach 2:
The invention changes the approach from fixed structural parameter conversion to dynamic parameter prediction. By using neural networks to predict channel parameters (such as inter-channel level difference and inter-channel time difference) from downmix signals, the system adapts parameters based on learned patterns rather than structural restrictions, improving compression efficiency.
2Productivity
If compression is increased to improve transmission efficiency, then bandwidth usage improves, but audio signal quality deteriorates
Solution Approach 1:
The patent employs a feedback mechanism where the neural network is trained using ground truth channel parameters from original audio signals. During training, the predicted parameters are compared with actual parameters, and the network adjusts its weights to minimize prediction errors. This feedback loop ensures high prediction accuracy, maintaining audio quality even at high compression ratios.
Solution Approach 2:
The system performs preliminary training action offline where the neural network learns from extensive audio data to master parameter prediction. This preliminary learning phase enables the network to accurately predict channel parameters during actual transmission without requiring high bitrates, thus maintaining quality while improving transmission efficiency.
Data Source
AI summary
A method of predicting a channel parameter of an original signal from a downmix signal is disclosed. The method may include generating an input feature map to be used to predict a channel parameter of the original signal based on a downmix signal of an original signal, determining an output feature map including a predicted parameter to be used to predict the channel parameter by applying the input feature map to a neural network, generating a label map including information associated with the channel parameter of the original signal, and predicting the channel parameter of the original signal by comparing the output feature map and the label map.