Spectral Envelope Bit Allocation for Efficient Audio Encoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional encoding apparatuses face inefficiencies due to envelope unevenness in MDCT coefficient sequences, which affects the encoding efficiency when using smoothed amplitude spectral envelopes for normalization.
Innovation Solution
An encoding apparatus that estimates a spectral envelope for frequency domain sample sequences raised to a power other than 2, allowing bit allocation to change based on this estimation, and a decoding apparatus that generates an unsmoothed spectral envelope sequence to adjust bit allocation accordingly.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If smoothed amplitude spectral envelope is used for normalization, then auditory distortion is reduced, but encoding efficiency deteriorates due to envelope unevenness
Solution Approach 1:
The patent changes the power parameter from the conventional value of 2 to other values (η≠2) when calculating the spectral envelope. This parameter modification allows the system to reduce auditory distortion while achieving better encoding efficiency by optimizing the spectral envelope characteristics for the specific application.
Solution Approach 2:
The patent introduces dynamic bit allocation that adapts to the spectral envelope characteristics. By making the bit allocation dynamic and coefficient-dependent rather than static, the system can efficiently represent the normalized MDCT coefficients while maintaining high audio quality.
2Device complexity
If bit allocation is fixed, then device complexity is reduced, but encoding efficiency deteriorates
Solution Approach 1:
The patent applies different bit allocation strategies to different frequency coefficients based on their individual characteristics and the spectral envelope. This local optimization allows each coefficient to be encoded with appropriate precision, improving overall encoding efficiency without requiring complex global optimization.
Solution Approach 2:
The patent performs preliminary normalization of the MDCT coefficients using the spectral envelope before the actual encoding process. This preliminary action simplifies the subsequent bit allocation by pre-compensating for spectral variations, making the encoding process more efficient.
Data Source
AI summary
There are provided an encoding or decoding technique capable of performing more efficient encoding or decoding than before. An encoding apparatus for encoding a time-series signal in a frequency domain is provided with: a spectral envelope estimating portion 2A estimating a spectral envelope, regarding absolute values of a frequency domain sample sequence corresponding to the time-series signal raised to the power of η as a power spectrum, on the assumption that η is a predetermined positive number other than 2; and an encoding portion 2B performing such encoding that changes bit allocation or that bit allocation substantially changes, for each coefficient of the frequency domain sample sequence corresponding to the time-series signal, based on the estimated spectral envelope.