Audio Signal Decoding with IGF for High-Frequency Band Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current audio codecs face limitations in maintaining high audio quality at low bitrates due to restricted bandwidth extension techniques that fail to accurately align tonal harmonics, leading to loss of HF detail and timbre, and require complex domain transformations, increasing computational complexity and memory requirements.
Innovation Solution
The Intelligent Gap Filling (IGF) method allows for full-rate core decoding and encoding across the entire audio signal range, using frequency tiles from lower frequency ranges to fill spectral gaps in higher frequency ranges, eliminating the need for downsampling and upsampling, and operating within the same spectral domain as the core decoder, ensuring accurate alignment of tonal components and reducing computational complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If bandwidth extension techniques are used to reduce bitrate, then coding efficiency is improved, but audio quality deteriorates due to loss of HF detail and timbre
Solution Approach 1:
The patent copies spectral information from lower frequency bands to higher frequency bands through spectral replication. The low-frequency spectral components are copied and transposed to high-frequency regions, creating a realistic high-frequency content that preserves audio quality while reducing bitrate requirements.
Solution Approach 2:
The patent changes spectral parameters by applying spectral shaping and filtering to the replicated high-frequency content. Energy information values are used to adjust the spectral envelope, and filtering operations modify the spectral characteristics to match the original signal's timbre and color, thereby improving audio quality.
2Productivity
If spectral patching is used to reconstruct HF spectral region, then bandwidth extension is achieved, but tonal harmonics misalignment occurs
Solution Approach 1:
The patent employs feedback mechanisms through iterative spectral matching and alignment processes. The spectral replication and shaping operations are refined by comparing the reconstructed high-frequency content with the original signal characteristics, adjusting the spectral parameters to achieve accurate tonal harmonics alignment.
3Manufacturing precision
If domain transformation is applied for bandwidth extension, then HF reconstruction is improved, but computational complexity increases
Solution Approach 1:
The patent uses the existing MDCT transform for multiple purposes: both for the core low-frequency coding and for the high-frequency spectral replication. This multi-functional use of the same transform domain eliminates the need for separate domain transformations, reducing computational complexity while maintaining HF reconstruction quality.
4Productivity
If downsampling and upsampling are used for bandwidth extension, then bitrate is reduced, but full audio bandwidth is lost
Solution Approach 1:
The patent applies partial bandwidth encoding by fully encoding only the essential low-frequency core band while using spectral replication to synthesize the high-frequency content. This partial action approach reduces bitrate by encoding only the necessary information at full resolution, while the remaining high-frequency content is reconstructed through spectral copying and shaping.
Data Source
Figure 1A~1B
Figure 2A
Figure 2B
AI summary
An apparatus for decoding an encoded audio signal comprising an encoded representation of a first set of first spectral portions and an encoded representation of parametric data indicating spectral energies for a second set of second spectral portions, comprises: an audio decoder (900) for decoding the encoded representation (901b) of the first set of the first spectral portions to obtain a first set of first spectral portions (904) and for decoding the encoded representation of the parametric data to obtain a decoded parametric data (902) for the second set of second spectral portions indicating, for individual reconstruction bands, individual energies; a frequency regenerator (906) for reconstructing spectral values in a reconstruction band (920) comprising a second spectral portion (922, 923) using a first spectral portion of the first set of the first spectral portions and an individual energy for the reconstruction band, the reconstruction band comprising a first spectral portion (921) and the second spectral portion; wherein the frequency regenerator (906) is configured for determining (912) a survive energy information comprising an accumulated energy of the first spectral portion having frequency values in the reconstruction band, determining (918) a tile energy information of further spectral portions (922, 923) of the reconstruction band (920) for frequency values different from the first spectral portion (921) having frequencies in the reconstruction band (920), wherein the further spectral portions (922, 923) are to be generated by frequency regeneration using a first spectral portion (302) different from the first spectral portion (921, 306) in the reconstruction band; determining (914) a missing energy in the reconstruction band (920) using the individual energy for the reconstruction band and the survive energy information; and adjusting (916) the further spectral portions in the reconstruction band based on the missing energy information and the tile energy information.