Compressió basada en ADPCM de senyals estéreo utilitzant codificació de suma i diferència.

[EN] This Bachelor's Thesis sets out to design and develop a method for compressing stereo audio signals using Adaptive Differential Pulse Code Modulation (ADPCM). The project is planned with the goal of finding a balance between reducing the amount of data needed for audio transmission and storage, while maintaining a level of quality that remains acceptable for the listener. The proposed approach is to build a MATLAB-based ADPCM codec that allows testing with different quantization depths (6, 7, and 8 bits per sample). Instead of compressing the left (L) and right (R) channels separately, the work will focus on converting them into sum (S) and difference (D) signals. This method is expected to take advantage of the redundancy usually found in stereo recordings, where the difference signal tends to carry less information and can therefore be encoded with fewer bits. The objectives are to explore how varying the number of bits for the difference channel affects the overall quality of the reconstructed audio, and to examine whether this approach can provide an efficient and practical compression strategy. The study aims to contribute to the field of audio signal processing, particularly in applications where limited bandwidth or memory makes compression techniques essential.