Stereo Mid-Side Extraction for Moving and Off-Center Sources
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional mid and side signal calculation techniques for stereo audio signals fail to effectively separate audio sources that are not center-panned, particularly those that are moving, have reverberation, or are spatially dominant at various points in time-frequency space.
Innovation Solution
A method for extracting target mid and side audio signals by obtaining target panning and phase difference parameters for each time segment and frequency band, using weighted sums and differences of left and right audio signals to create target mid and side signals that can capture audio sources varying in time and frequency, allowing for simultaneous extraction of multiple sources.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional mid and side signal calculation is used, then center-panned sources are effectively separated, but non-center-panned sources cannot be separated
Solution Approach 1:
The patent applies dynamics by making the mid/side extraction adaptive to the actual panning configuration of the audio sources. Instead of using fixed conventional mid/side calculation, the system dynamically determines whether sources are center-panned or non-center-panned and adjusts the extraction method accordingly, enabling effective separation for both cases
Solution Approach 2:
The patent changes the parameters of the mid/side extraction process based on the detected panning configuration. By modifying the extraction parameters according to whether sources are center-panned or non-center-panned, the system achieves accurate separation for different panning scenarios without being limited to a single configuration
2Measurement precision
If panning-specific mid and side signal calculation techniques are used, then stationary non-center panned sources can be targeted, but moving sources and sources with reverberation cannot be separated
Solution Approach 1:
The patent makes the mid/side extraction dynamically adaptive to track moving sources and accommodate reverberation. By continuously adjusting the extraction parameters based on the detected panning configuration at different time points, the system can effectively separate moving sources and sources with reverberation, overcoming the limitation of static panning-specific techniques
Solution Approach 2:
The patent employs feedback by using the detected panning configuration information to adjust the mid/side extraction process. The system analyzes the panning of audio sources and uses this information to optimize the extraction parameters, creating a closed-loop system that adapts to changing source positions and reverberation characteristics
3Ease of manufacture
If conventional mid and side signal extraction is used, then the process is simple, but source separation fails for non-center-panned sources
Solution Approach 1:
The patent introduces dynamic adaptation to the mid/side extraction process while maintaining relative simplicity. By automatically detecting the panning configuration and adjusting the extraction method accordingly, the system achieves effective separation for non-center-panned sources without requiring complex manual configuration or processing
Data Source
Figure 1~2
Figure 3~4a
Figure 4b~4c
AI summary
The present disclosure relates to a method and audio processing arrangement for extracting a target mid (and optionally a target side) audio signal from a stereo audio signal. The method comprises obtaining (S1) a plurality of consecutive time segments of the stereo audio signal and obtaining (S2), for each of a plurality of frequency bands of each time segment of the stereo audio signal, at least one of a target panning parameter (Θ) and a target phase difference parameter (Φ). The method further comprises extracting (S3), for each time segment and each frequency band, a partial mid signal representation (211, 212) based on at least one of the target panning parameter (Θ) and the target phase difference parameter (Φ) of each frequency band and forming (S4) the target mid audio signal (M) by combining the partial mid signal representations (211, 212) for each frequency band and time segment.