Audio Encoding Auto-Regressive Coefficients Mirroring
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional quantization schemes for auto-regressive (AR) coefficients are not effective when dealing with high audio frequencies and limited bit-budgets, leading to large perceptual errors in signal reconstruction.
Innovation Solution
A method involving weighted averaging of quantized elements flipped around a mirroring frequency, using a frequency grid codebook in a closed-loop search procedure, to efficiently encode and decode AR coefficients, separating low-frequency and high-frequency parts of the audio signal.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional quantization schemes are used for AR coefficients, then the quantization process is simple and fast, but large perceptual errors occur in high-frequency regions and at limited bitrates
Solution Approach 1:
The spectrum is divided into low-frequency and high-frequency parts, with different quantization strategies applied to each. The low-frequency part uses conventional quantization while the high-frequency part uses a modeling approach based on spectral mirroring, allowing efficient representation at limited bitrates while maintaining accuracy where needed
Solution Approach 2:
The invention changes the representation parameters for high-frequency coefficients by using a model based on low-frequency coefficients and spectral mirroring properties. Instead of directly quantizing high-frequency coefficients, the system derives them from low-frequency data through mathematical relationships, improving efficiency at low bitrates
2Measurement precision
If full-spectrum quantization is applied to all AR coefficients, then reconstruction accuracy is maintained across all frequencies, but computational complexity increases significantly
Solution Approach 1:
The invention extracts and processes only the essential low-frequency coefficients that carry the most important spectral information. High-frequency coefficients are derived from these extracted low-frequency coefficients through spectral mirroring relationships, eliminating the need to quantize and transmit all coefficients while maintaining reconstruction accuracy
Solution Approach 2:
Instead of applying quantization to the entire spectrum, the invention applies full quantization only to the low-frequency part and uses a modeling approach for the high-frequency part. This partial action approach maintains necessary accuracy while significantly reducing computational complexity
Data Source
AI summary
An encoder for encoding a parametric spectral representation (ƒ) of auto-regressive coefficients that partially represent an audio signal. The encoder includes a low-frequency encoder configured to quantize elements of a part of the parametric spectral representation that correspond to a low-frequency part of the audio signal. It also includes a high-frequency encoder configured to encode a high-frequency part (ƒH) of the parametric spectral representation (ƒ) by weighted averaging based on the quantized elements ({circumflex over (ƒ)}L) flipped around a quantized mirroring frequency ({circumflex over (ƒ)}m), which separates the low-frequency part from the high-frequency part, and a frequency grid determined from a frequency grid codebook in a closed-loop search procedure. Described are also a corresponding decoder, corresponding encoding/decoding methods and UEs including such an encoder/decoder.