Audio Bit Allocation Using Psychoacoustic Projection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing bit allocation processes in audio encoding systems, such as those used in MPEG-audio Layer-3 and AAC, are inefficient due to the large number of loops required to determine quantization parameters, leading to poor encoding efficiency and excessive operations.
Innovation Solution
A method that includes scalefactor and stepsize factor projection techniques to predict and adjust these parameters, reducing the number of loops and operations by using absolute thresholds of hearing and psychoacoustic masking values to generate projection values and determine optimal scalefactors and stepsize factors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional bit allocation processes with multiple loops are used to determine quantization parameters, then audio encoding quality can be maintained, but encoding efficiency deteriorates due to excessive computational operations
Solution Approach 1:
The patent applies preliminary action by pre-calculating and storing offset values for different bit rate conditions before actual audio encoding. The offset values are determined in advance based on psychoacoustic models and stored in a lookup table, allowing the encoder to quickly retrieve and apply appropriate offsets without performing complex iterative calculations during real-time encoding, thus resolving the contradiction between maintaining audio quality and improving encoding efficiency
Solution Approach 2:
The patent uses copying by creating a simplified model of the complex bit allocation process. Instead of executing multiple loops to determine quantization parameters, the system copies pre-computed offset values from a lookup table that represents the results of complex psychoacoustic analysis. This copied data allows rapid determination of quantization parameters while maintaining the quality characteristics of the full computational process
2Productivity
If the number of loops for determining quantization parameters is reduced, then encoding efficiency improves, but the precision of parameter determination may deteriorate
Solution Approach 1:
The patent replaces the mechanical iterative loop system with a direct lookup mechanism. Instead of mechanically executing multiple loops to converge on optimal quantization parameters, the system substitutes this with a direct memory access operation that retrieves pre-computed offset values. This substitution maintains parameter determination accuracy by using pre-calculated values from psychoacoustic models while dramatically improving encoding speed by eliminating iterative computation
Solution Approach 2:
The patent applies parameter changes by transforming the complex multi-parameter optimization problem into a simpler parameter retrieval problem. The system changes the approach from dynamically calculating multiple parameters through iterative loops to statically retrieving offset parameters from pre-computed tables. This parameter transformation maintains accuracy by preserving the essential relationships captured in the pre-computed offsets while simplifying the computational process
Data Source
AI summary
An audio processing method utilized to generate an audio stream. An audio frame includes N frequency subbands. An Ith frequency subband among the N frequency subbands includes M audio samples and has an Ith psychoacoustic masking value. First, an Ith offset of the Ith frequency subband is calculated. Then, the Ith psychoacoustic masking value and the Ith offset are inputted into a projection formula to generate an Ith projection value. According to the Ith projection value and a limit range, an Ith scale factor is determined. Subsequently, the M audio samples in the Ith frequency subband are adjusted according to the Ith scale factor.