Open AccessFrom audio to information: Learning topics from audio transcripts
Author(s) -
João Pedro Rodrigues,
Emerson Cabrera Paraíso
Publication year - 2020
Language(s) - English
Resource type - Conference proceedings
DOI - 10.5753/kdmile.2020.11967
Subject(s) - computer science , latent dirichlet allocation , audio signal processing , coherence (philosophical gambling strategy) , topic model , speech recognition , artificial intelligence , ideal (ethics) , multimedia , natural language processing , speech coding , audio signal , philosophy , epistemology , physics , quantum mechanics
In this work, the technical feasibility of working with audio transcriptions from Youtube is analyzed, as well as presenting a method that allows data acquisition, pre-processing, and post-processing to work with this type of data. A topic modeling approach with the latent dirichlet allocation algorithm is used. An approach is also presented to dynamically determine the ideal number of topics that make up a given corpus. In the experiments, a database of 250 audio transcriptions was used, obtaining a model with coherence in the range of 40%.