Deep Neural Network Models for Audio Quality Assessment

With the proliferation of new and more complex multimedia and network services, measuring the perceived quality of audio signals has become crucial. Providers strive to deliver high quality and reliable services to their customers in order to assure them a good quality of experience (QoE). In this work, a new audio quality dataset labelled by humans is presented. Moreover, results of two DNN-based models for estimating the quality experienced by users of audio transmission and communication services are compared.

Speaker Bios

Anderson Avila received his BSc in Computer Science from the Federal University of Sao Carlos, in 2004, and his MSc in Information Engineering at Federal University of ABC, both in Brazil. He is now pursuing his PhD at the Institute National de la Recherche Scientifique (INRS-EMT), Canada, where he works on the robustness of speaker verification and speech emotion recognition systems, under the supervision of Professor Tiago H. Falk.