Reconhecimento de fala em idosos: elaboração e aplicação de um teste considerando a previsibilidade da palavra

• Ano de defesa: 2011
• Autor(a) principal: Calais, Lucila Leal [UNIFESP]
• Tipo de documento: Tese
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal de São Paulo (UNIFESP)
• Palavras-chave em Português: Cognição; Envelhecimento; Presbiacusia; Percepção de fala
• Link de acesso: http://repositorio.unifesp.br/handle/11600/10014

Objective: To create and to verify the effectiveness of speech recognition assessment instrument, when the predictability of the word is considered in elderly with and without hearing loss. Methods: To create the instrument, initially, the words were taken from the Portuguese Language corpus. The concreteness and a association test were applied to assist the development of the instrument. It was determined that the sentences would be declaratives, affirmatives, with a simple period structure and extension control. Finally, it was applied a predictability test, in order to achieve sentences with high and low predictability. Following, four lists of sentences were chosen with low and high predictability. The speech recognition test using sentences with high and low predictability was recorded in a professional studio and applied in 36 elderly people with normal hearing thresholds up to 4 kHz or with sensorineural hearing loss mild to moderate degree. All elderly people were submitted to a clinical history, audiological conventional evaluation, as well as the elaborated speech recognition test. Participants were distributed in four groups, and two took the test under the condition of silence and masking ipsilateral noise in the signal/noise ratio + 5 dB, and the other two groups only in the noise condition in relations + 5 dB and 0 dB . Results: The concreteness test revealed a tri-modals distribution, with words grouped in categories of low, medium and high concreteness. In the semantic association test all words generated, at least, two associated words, both to verbs and nouns groups evoked. The predictability sentences test generated, in 191 sentences in the categories of low and high predictability. Considering the hearing evaluation and the Test of Speech Recognition with High and Low Predictability Sentences in Noise, the results showed the differences among the groups with worst thresholds and performance in the groups with hearing loss. In silence, the speech recognition test resulted in similar performance to the participants. In general, it was noticed that words scores was greater in high predictability sentences. The noise had a negative influence on speech recognition, meanly when more intense. The predictability support benefited the elderly mostly. There was no significant difference comparing the obtained responses for the right and left ears, according to the order of presentation of the test for the first and second ears. Conclusions: The instrument elaborated permitted to evaluate the speech recognition in the elderly people considering the predictability support, observing the negative impact of peripheral hearing loss and the competitive noise.