Avaliação de inteligibilidade da fala em salas de aula utilizando aurilizações simuladas

• Ano de defesa: 2022
• Autor(a) principal: Mahl, Ana Luiza
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de Santa Maria; Brasil; Engenharia Civil; UFSM; Programa de Pós-Graduação em Engenharia Civil; Centro de Tecnologia
• Palavras-chave em Português: Acústica de salas; Inteligibilidade da fala; Salas de aula; Simulação computacional; Aurilização; RAIOS; Room acoustics; Speech intelligibility; Classrooms; Computer simulation; Auralization; CNPQ::ENGENHARIAS::ENGENHARIA CIVIL
• Link de acesso: http://repositorio.ufsm.br/handle/1/26364

The present research aims to evaluate speech intelligibility in higher education classrooms, specifically two classrooms at the Technology Center of the Federal University of Santa Maria, in Santa Maria, RS, through articulation indices (AI) obtained with real articulation tests (TAR) and articulation tests with simulated signals (TASS), contributing to the validation of the methodology for using TASS to obtain AI. To this end, acoustic measurements are carried out to survey the monaural impulsive response in both classrooms, in order to obtain the acoustic parameters: reverberation time (TR), initial decay time, clarity factor for speech and definition for the speaks. Then, the input data are entered into the room acoustics software, computer code RAIOS 7, in which the computer models are calibrated, comparing the measured and simulated TR. With this, binaural impulsive responses (BRIRs) are generated in the same software. Subsequently, the generated BRIRs are convolved with anechoic signals (which, in this case, are monosyllables recorded in the anechoic chamber of Inmetro, in Xerém, Rio de Janeiro), which consists in the generation of aurilizations for each source-receiver pair of each classroom that also takes place in the RAIOS7 software. Finally, the TAR (in classrooms) and TASS (in which the binaural signals are heard through headphones) are applied to volunteers to obtain the AI. The results obtained confirm that the use of aurilizations with simulated signals for the application of articulation tests in order to obtain articulation indices is a reliable methodology, with an absolute mean error lower than 5.0%. In this way, it is possible to evaluate the intelligibility of speech rooms in a virtual way, still in the design phase and for existing rooms, without the need to insert all students in a classroom, which is especially useful in pandemic scenarios, as currently experienced, in addition to avoiding possible distractions during the tests.