Audio Encoding Bit Allocation in Split Multi-Rate Vector Quantization

The patent applies local quality by differentiating the treatment of sub-vectors based on their energy characteristics. High-energy sub-vectors receive larger codebooks for better quantization accuracy, while low-energy sub-vectors use smaller codebooks to save bits. This localized adaptation resolves the contradiction by optimizing the balance between quantization accuracy and bit rate for each sub-vector individually rather than uniformly across all sub-vectors.

The patent dynamically changes the codebook size parameter based on the energy level of each sub-vector. By adjusting the codebook indication value bits according to sub-vector energy characteristics, the system optimizes the trade-off between quantization precision and bit rate. This parameter adaptation allows the system to achieve high quantization accuracy where needed while minimizing bit rate in regions where it is less critical.

The patent implements local quality by assigning different bit allocation strategies to different sub-vectors based on their energy characteristics. Instead of uniform treatment, each sub-vector receives an optimized bit allocation matching its energy level, improving overall bit rate efficiency without significantly increasing encoding complexity through automated energy-based classification.

The patent introduces dynamic bit allocation that adapts to the energy characteristics of each sub-vector. The system dynamically determines the appropriate codebook size and bit allocation for each sub-vector based on real-time energy analysis, allowing flexible optimization of bit rate efficiency while maintaining manageable encoding complexity through systematic adaptation rules.