The combined method for audio stream optimization based on fragmentation and minimal information content assessment

This paper presents a method for optimizing segments of an audio stream with minimal informational content (such as silence, speech pauses, and other weakly informative sounds) based on fragmentation and structural analysis of the signal. The method involves preliminary processing of audio data by dividing the signal into fragments and assigning each a descriptor containing information about quantitative and qualitative characteristics (e.g, amplitude fluctuations and the assumed level of informational content). This approach enables automatic identification of low-informative segments, which are replaced with compact descriptors, while informative fragments are preserved unchanged. The proposed optimization is complemented by lossless compression algorithms, including dictionary-based coding (Lempel—Ziv—Welch algorithm) and entropy-based methods (Huffman or arithmetic coding), allowing a significant reduction in audio data size while retaining the ability to reconstruct the original structure. The experimental results are compared with the performance of widely used lossy compression formats (MP3 and AAC).