Loudness Equalization Using Temporal Smoothing and Look-Ahead
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Audio coding technologies face challenges in accurately quantifying and maintaining consistent loudness across different audio signals due to the subjective nature of loudness perception, which affects the quality of recreated audio.
Innovation Solution
A method for loudness equalization that involves converting input loudness data to a linear scale, applying temporal smoothing and look-ahead processing, and using non-linear weights to distribute energy across frequency bands, while maintaining constant loudness and preserving audio dynamics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If audio coding is applied to compress audio data, then data transmission efficiency is improved, but audio quality deteriorates due to loss of loudness information
Solution Approach 1:
The audio signal is divided into multiple frequency bands using a filter bank, allowing independent processing of each band. This segmentation enables targeted loudness equalization without affecting the entire audio spectrum, preserving quality while maintaining compression efficiency.
Solution Approach 2:
The patent transforms loudness data from a logarithmic scale to a linear scale, and applies gain adjustments based on reciprocal relationships. These parameter transformations enable more effective loudness control and equalization, improving audio quality after compression.
2Stability of the object's composition
If gain adjustment is applied to equalize loudness, then loudness consistency is improved, but audible noise is generated
Solution Approach 1:
The system applies temporal smoothing to gain adjustments, dynamically adapting the equalization over time rather than applying static gains. This dynamic approach prevents abrupt changes that cause audible noise while maintaining loudness consistency across different audio segments.
Solution Approach 2:
The patent introduces an intermediate processing stage that converts gain data between logarithmic and linear scales, and applies look-ahead processing to anticipate noise-generating situations. This intermediary processing prevents direct application of harsh gain adjustments that would create audible noise.
3Measurement precision
If loudness equalization is applied across frequency bands, then loudness perception is improved, but computational complexity increases
Solution Approach 1:
The audio spectrum is divided into a limited number of frequency bands using a filter bank, reducing the computational burden compared to processing the entire spectrum continuously. Each band can be processed independently with simpler algorithms, lowering overall complexity while maintaining perception accuracy.
Solution Approach 2:
Different processing approaches are applied to different frequency bands based on their characteristics. This local quality approach allows more aggressive processing where needed while using simpler methods in other bands, optimizing the balance between perception accuracy and computational complexity.
Data Source
AI summary
A method for loudness equalization is provided that includes receiving input loudness data at an audio processing system. Converting gain data of the input loudness data to a linear scale at the audio processing system. Determining a reciprocal of a gain-linear loudness value as a function of the converted gain data using the audio processing system. Determining a compression ratio using the audio processing system. Performing temporal smoothing and look ahead processing using the audio processing system. Outputting gain data as a function of the temporal smoothing and look ahead processing using the audio processing system.