Audio Semantic Communication in Heterogeneous Networks using Dedicated Autoencoders

• Grant number: 25/26814-1
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: February 01, 2026
• End date: January 31, 2027
• Field of knowledge: Engineering - Electrical Engineering - Telecommunications
• Principal Investigator: Fernando José von Zuben
• Grantee: Marcos Ferreira Semolini
• Host Institution: Faculdade de Engenharia Elétrica e de Computação (FEEC). Universidade Estadual de Campinas (UNICAMP). Campinas , SP, Brazil
• Company: Universidade Estadual de Campinas (UNICAMP). Faculdade de Engenharia Elétrica e de Computação (FEEC)
• Associated research grant: 21/00199-8 - SMART NEtworks and ServiceS for 2030 (SMARTNESS), AP.PCPE

Abstract

In audio semantic communication using autoencoders, an autoencoder is trained to extract and transmit only the meaningful, high-level features of an audio signal, rather than transmitting the entire raw data. This process significantly reduces the amount of data that needs to be transmitted over a network, enhancing communication efficiency and robustness, particularly in low-bandwidth or noisy conditions.An autoencoder based on a wav2vec-vec2wav architecture has already been implemented following the proposal by Tog et al. (2021). This framework has shown promise in extracting semantic features from audio signals and encoding them in a compact representation.In an audio transmission network where nodes have distinct computational capabilities, it is crucial to design encoders and decoders that are tailored to each node's local computational restrictions. This research project investigates the potential of utilizing autoencoders in heterogeneous transmission networks for audio semantic communication. (AU)