Audio Frame Coding with LPC Noise Shaping for Low Bitrates
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional audio coding methods fail to achieve an optimal balance between strong compression and good acoustic quality due to insufficient combination of parametric and non-parametric coding tools, particularly in low-bitrate scenarios, as they do not adequately distinguish between zero-quantized and non-zero-quantized portions of audio signals.
Innovation Solution
An audio decoder and encoder that differentiate between zero-quantized and non-zero-quantized portions of audio signals by applying distinct noise shaping and spectral modification techniques based on linear prediction coefficient envelopes, ensuring improved acoustic quality and compression efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If conventional audio coding methods use a single quantization approach for all spectral portions, then device complexity is reduced and ease of operation is improved, but acoustic quality deteriorates due to insufficient distinction between zero-quantized and non-zero-quantized portions
Solution Approach 1:
The spectrum is segmented into zero-quantized portions and non-zero-quantized portions based on the quantized spectrum values. Different coding tools are applied to each segment: parametric coding tools (noise filling, spectral band replication, intelligent gap filling) are applied to zero-quantized portions, while non-parametric coding tools are applied to non-zero-quantized portions. This segmentation allows optimized processing for each type of spectral content.
Solution Approach 2:
Different coding strategies are applied to different portions of the spectrum based on their characteristics. Zero-quantized portions (where spectral coefficients are zero or near-zero) receive parametric coding treatment with noise shaping, while non-zero-quantized portions receive non-parametric coding treatment. This local differentiation improves overall acoustic quality by matching the coding approach to the local spectral characteristics.
2Loss of substance
If parametric coding tools are applied to all spectral portions, then compression ratio is improved, but acoustic quality deteriorates due to inappropriate application in non-zero-quantized portions
Solution Approach 1:
The coding approach dynamically adapts to the spectral characteristics of each portion. The system determines whether each spectral portion is zero-quantized or non-zero-quantized and selectively applies parametric or non-parametric coding tools accordingly. This dynamic adaptation ensures optimal compression is achieved in appropriate regions while maintaining acoustic quality in regions where parametric coding would be detrimental.
3Manufacturing precision
If different coding tools are combined for zero-quantized and non-zero-quantized portions, then acoustic quality is improved, but device complexity increases due to multiple processing paths
Solution Approach 1:
The spectrum is pre-segmented into zero-quantized and non-zero-quantized portions before applying different coding tools. This preliminary classification simplifies the subsequent processing by establishing clear boundaries for where parametric and non-parametric tools should be applied, reducing the complexity of managing multiple coding paths during the actual encoding process.
Data Source
AI summary
Embodiments comprise an audio decoder configured to, for a predetermined frame among consecutive frames, decode, from a data stream, a quantized spectrum and a linear prediction coefficient based envelope representation; to locate, in the quantized spectrum, zero-quantized portions and non-zero-quantized portions and to derive a dequantized spectrum using in zero-quantized portions of the quantized spectrum, filling the quantized spectrum with a synthesized spectral data modified depending, according to a first manner, on the linear prediction coefficient based envelope representation, and in non-zero-quantized portions of the quantized spectrum, modifying the quantized spectrum depending, in a second manner, on the linear prediction coefficient based envelope representation; and to reconstruct the predetermined frame using the dequantized spectrum; so that, for a predetermined portion, the modification according to the first and the second manner cause a spectral quantization noise shaping which comprises different smoothness characteristics.Corresponding encoders and methods are also disclosed.