Audio Mixing with Magnitude Equalization for Coloration Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional audio signal mixing techniques often result in loss of spectral energy and loudness due to phase differences between input channels, leading to audible coloration effects, especially when down-mixing or up-mixing audio signals for different playback systems.
Innovation Solution
The proposed solution involves transforming input audio signals into the frequency domain, applying short-term energy or magnitude equalization to adjust for energy losses, and then inverting the transform to generate output signals with equalized loudness, using a combination of short-term energy equalization above 1 kHz and magnitude equalization below 1 kHz to minimize phase dependencies and coloration effects.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If conventional time-domain mixing techniques are used to mix audio channels, then the mixing process is simple and computationally efficient, but spectral energy and loudness are lost due to phase differences between input channels
Solution Approach 1:
The patent transforms the mixing process from the time domain to the frequency domain, where spectral components can be processed independently. By changing the domain parameter from time to frequency, the system preserves spectral energy that would otherwise be lost due to phase cancellation in time-domain mixing. This is achieved through Fourier transformation of input channels, independent processing of spectral components, and inverse transformation back to time domain.
Solution Approach 2:
The invention moves the mixing operation from the time dimension to the frequency dimension. Instead of mixing signals as they evolve over time, the patent analyzes and mixes the spectral content at different frequencies. This dimensional change allows energy preservation because phase relationships that cause cancellation in the time domain are resolved by processing each frequency component separately in the frequency domain.
2Object-affected harmful factors
If conventional time-domain mixing techniques are used, then the implementation is straightforward, but audible coloration effects occur due to energy loss from phase misalignment
Solution Approach 1:
By transforming to the frequency domain, the patent changes the parameter space in which mixing occurs. This allows independent control of spectral components at different frequencies, enabling the system to preserve energy and eliminate coloration effects that arise from indiscriminate time-domain mixing of all frequency components together.
Solution Approach 2:
The mixing process is segmented by frequency, with each spectral component processed independently rather than all frequencies being mixed together as a single time-domain signal. This segmentation allows the system to identify and preserve energy in each frequency band separately, preventing the phase-related coloration effects that occur when all frequencies are processed uniformly.
3Reliability
If frequency-domain processing with equalization is applied, then spectral energy and loudness are preserved, but the computational complexity and processing requirements increase
Solution Approach 1:
The patent changes the processing domain from time to frequency, enabling reliable preservation of spectral energy and perceived quality. The frequency-domain representation allows for energy-preserving mixing operations that maintain the integrity of the audio signal across all frequency components, significantly improving reliability and perceived quality compared to conventional time-domain methods.
Data Source
AI summary
According to one embodiment, during mixing of an N-channel input signal to generate an M-channel output signal, in at least one frequency sub-band, magnitude equalization is applied to the mixed channel signals such that an amplitude sum magnitude for the N input channels (e.g., the magnitude of a sum of estimated amplitudes of the N input channels) is approximately equal to an amplitude sum magnitude for the M output channels (e.g., the magnitude of a sum of estimated amplitudes of the M output channels). In one implementation, magnitude equalization is applied to one or more sub-bands (e.g., those below 1 kHz), and power equalization is applied to one or more other sub-bands (e.g., those above 1 kHz) to reduce coloration effects in the output signal.