Magnitude-Phase Softmask Optimization for Stereo Source Separation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing audio source separation techniques struggle with improving the perceptual quality of extracted sources, particularly in cases where amplitude panning is used, leading to interference, reduced source levels, and improper frequency constraints, especially in stereo mixes.

Innovation Solution

Implementing perceptual optimizations such as bulk reduction, expansion and limiting, frequency gating, phase and panning adjustments, and smoothing of softmask values to enhance the quality of source separation by addressing these issues.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Ease of manufacture

If softmask values are applied directly to frequency bins for source separation, then the source extraction process is simple, but the perceptual quality of extracted sources deteriorates due to interference and reduced source levels

Engineering Contradiction:

Improvesimplicity of source extraction processVSAvoidperceptual quality of extracted sources

Core Design Contradiction:

Ease of manufactureVSManufacturing precision

Solution Approach 1:

The patent applies bulk reduction to modify softmask values by multiplying them with a reduction factor, and applies expansion and limiting to adjust the dynamic range of softmask values. These parameter changes optimize the perceptual quality of extracted sources by reducing interference and enhancing source levels while maintaining processing simplicity

Inventive Principle:

Principle #35Parameter changes

Solution Approach 2:

The patent applies smoothing operations to softmask values before they are applied to frequency bins. This preliminary action reduces temporal and spectral variations in the softmask, preventing artifacts and improving the perceptual quality of extracted sources before the main separation process occurs

Inventive Principle:

Principle #10Preliminary action

2Adaptability or versatility

If amplitude panning is used in audio mixes, then the mixing flexibility is improved, but the source separation quality deteriorates due to interference and improper frequency constraints

Engineering Contradiction:

Improvemixing flexibilityVSAvoidsource separation quality

Core Design Contradiction:

Adaptability or versatilityVSManufacturing precision

Solution Approach 1:

The patent applies frequency gating to set softmask values to zero outside a specified frequency range for the target source. This local quality approach ensures that only frequency bins within the relevant range contribute to the extracted source, eliminating interference from out-of-band frequencies while preserving the benefits of amplitude panning in the valid frequency range

Inventive Principle:

Principle #3Local quality

3Object-affected harmful factors

If softmask values are reduced to minimize background interference, then the background suppression is improved, but the source level and clarity deteriorate

Engineering Contradiction:

Improvebackground interference suppressionVSAvoidsource level and clarity

Core Design Contradiction:

Object-affected harmful factorsVSManufacturing precision

Solution Approach 1:

The patent applies expansion and limiting to counterbalance the effect of bulk reduction. By multiplying softmask values with an expansion factor and applying a limiter, the system compensates for the attenuation of source levels caused by background suppression, maintaining both background rejection and source clarity

Inventive Principle:

Principle #8Anti-weight (Counterweight)

Solution Approach 2:

The patent applies smoothing to softmask values, which provides temporal and spectral feedback by averaging values over time and frequency. This feedback mechanism prevents excessive reduction in any single bin while maintaining overall background suppression, preserving source levels and clarity

Inventive Principle:

Principle #23Feedback

Data Source

PatentUS12382234B2Perceptual optimization of magnitude and phase for time-frequency and softmask source separation systems

Publication Date: 2025.08.05 DOLBY LABORATORIES LICENSING CORP

AI summary

A method comprises: obtaining softmask values for frequency bins of time-frequency tiles representing an audio signal; reducing, or expanding and limiting, the softmask values; and applying the reduced, or expanded and limited, softmask values to the frequency bins to create a time-frequency representation of an estimated target source. An alternative method comprises, for each time-frequency tile: obtaining softmask values; applying the softmask values to the frequency bins to create a time-frequency domain representation of an estimated target source; obtaining a panning parameter and a source concentration estimates for the target source; determining, using the panning parameter estimate and the softmask values, a magnitude for the time-frequency representation of the estimated target source; determining, using the panning parameter estimate and the source phase concentration estimate, a phase for the time-frequency representation of the estimated target source; and combining the magnitude and the phase.