Neural Audio Window Equalization for Perceived Spectrum Balance
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is a need for computer-implemented aids in the audio production process, specifically in recording, mixing, and mastering, to automatically balance the perceived spectrum of audio content, which existing digital audio workstations (DAWs) do not effectively address.
Innovation Solution
A method and electronic device that determine feature values and model parameters from an input audio window using a neural network, allowing for automatic profile equalization by matching the audio window's spectrum to a target spectrum defined by model parameters, which are set to ideal values based on training data, and utilizing equal-loudness-level contours to weight the audio content for optimal perceived balance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual mastering process is used to balance the perceived spectrum, then audio quality can be optimized, but the process is time-consuming and requires expert knowledge
Solution Approach 1:
The system performs automatic spectrum balancing without requiring human operators. The neural network analyzes the input audio and automatically adjusts the spectrum to match target contours, enabling the system to serve itself in the mastering process rather than requiring external expert intervention
Solution Approach 2:
Equal-loudness-level contours are pre-calculated and stored for different reference loudness levels. These pre-computed contours serve as target models that guide the automatic equalization process, eliminating the need for real-time manual adjustment while maintaining perceptual accuracy
2Productivity
If automatic spectrum balancing is implemented, then processing time is reduced, but accuracy in matching perceived spectrum may deteriorate
Solution Approach 1:
The system uses a feedback mechanism where the actual spectrum of the audio is compared against the target equal-loudness-level contour, and equalization adjustments are made to minimize the difference. This closed-loop approach ensures that automatic processing achieves accurate perceptual matching
Solution Approach 2:
The system dynamically adjusts equalization parameters based on the input audio characteristics and selected reference loudness level. By changing the target contour parameters according to the specific audio content, the system maintains high accuracy across different scenarios while operating automatically
3Adaptability or versatility
If multiple equal-loudness-level contours are stored for different reference loudness levels, then adaptability to different listening conditions is improved, but memory requirements increase
Solution Approach 1:
Instead of storing complete spectral contours for every possible loudness level, the system stores a limited set of reference contours at key loudness levels. The target contour for intermediate levels is obtained by interpolating between stored references, reducing memory requirements while maintaining adaptability across the full range of reference loudness levels
Data Source
AI summary
A method comprising determining feature values of an input audio window and determining model parameters for the input audio window based on processing of feature values using a neural network.