Audio Noise Level Estimation for Zero-Bit Subband Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current audio encoding techniques, such as G.722.1C, HE-AAC-V1, and G.719, face challenges in reconstructing uncoded frequency domain coefficients due to high calculation complexity and increased encoding delay, with methods like noise filling and spectral band replication either failing to accurately recover spectrum envelopes or requiring excessive parameter information.
Innovation Solution
A method that estimates the noise level of zero bit encoding subbands by calculating the power spectrum of audio signals using MDCT frequency domain coefficients, allowing for controlled energy proportion of noise filling during decoding, thereby improving spectrum reconstruction without excessive parameter transmission or complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If spectral band replication technique is used to reconstruct uncoded subbands, then spectrum envelope recovery is improved, but calculation complexity increases and encoding delay increases
Solution Approach 1:
The patent extracts only the essential spectral envelope information from the coded subbands using a simplified replication approach. Instead of performing full spectral analysis and parameter extraction as in HE-AAC-V1, the invention selectively replicates and shapes spectral components to recover the envelope, removing unnecessary computational steps while maintaining recovery accuracy
Solution Approach 2:
The patent divides the frequency spectrum into multiple subbands and applies different reconstruction strategies to each. Coded subbands are processed with full accuracy while uncoded subbands use simplified replication, segmenting the problem to reduce overall complexity while maintaining quality where it matters most
2Measurement precision
If spectral band replication technique is used to reconstruct uncoded subbands, then spectrum envelope recovery is improved, but encoding delay increases
Solution Approach 1:
The patent performs preliminary spectral envelope estimation and replication parameter preparation during the encoding phase. By pre-calculating the replication factors and spectral shapes that will be needed for uncoded subbands, the decoding process can proceed more quickly without performing complex real-time analysis, thus reducing encoding delay while maintaining recovery accuracy
Solution Approach 2:
The patent transforms the complex spectral replication problem into a parameter-driven approach where a small set of replication parameters (scaling factors, frequency mappings) are computed and transmitted. This parameterization reduces both computational complexity and encoding delay while preserving the ability to accurately reconstruct spectral envelopes at the decoder
3Device complexity
If simple spectral band replication is used to reconstruct uncoded subbands, then calculation complexity is reduced, but tone and noise component recovery deteriorates
Solution Approach 1:
The patent applies different quality levels of spectral replication to different regions of the spectrum based on their characteristics. Tone components receive more precise replication with preserved phase and amplitude relationships, while noise components use stochastic replication methods. This local differentiation maintains recovery accuracy for critical components while keeping overall complexity low
Solution Approach 2:
The patent introduces dynamic adaptation in the spectral replication process by analyzing the local spectral characteristics (tone vs. noise, stationary vs. transient) and adjusting replication parameters accordingly. This dynamic approach allows the system to maintain high recovery accuracy for tone and noise components while using simpler methods where appropriate, balancing complexity and quality
4Device complexity
If noise filling method is used to reconstruct uncoded subbands, then calculation complexity is reduced, but spectrum envelope recovery deteriorates
Solution Approach 1:
The patent merges spectral band replication with noise filling in a hybrid approach. The replication component preserves the spectral envelope structure by copying and shaping frequency components from coded subbands, while the noise filling component adds stochastic elements to model uncorrelated noise. This combination maintains low computational complexity while significantly improving spectrum envelope recovery compared to pure noise filling
Solution Approach 2:
The patent creates a composite spectral reconstruction method that combines deterministic spectral replication (for envelope preservation) with stochastic noise modeling (for realism). This composite approach leverages the strengths of both methods: the structural accuracy of replication and the natural variability of noise filling, achieving good envelope recovery without the high complexity of pure replication techniques
Data Source
AI summary
The present invention relates to a method and system for audio encoding and decoding and a method for estimating a noise level, and the method for estimating a noise level in the present invention comprises: estimating a power spectrum of an audio signal to be encoded according to a frequency domain coefficient of the audio signal to be encoded; and estimating a noise level of a zero bit encoding subband audio signal according to the power spectrum obtained by calculating, and this noise level for controlling an energy proportion of noise filling to spectral band replication during decoding; wherein a zero bit encoding subband refers to an encoding subband of which allocated bit number is zero. The present invention can well reconstruct the uncoded frequency domain coefficients.