Teste de fala comprimida: adaptação e validação / Time-compressed speech test: adaptation and validation

RESUMO Objetivos: adaptar o Teste de Fala Comprimida e verificar a aplicabilidade do novo teste, correlacionando-o com o original. Métodos: a adaptação do Teste de Fala Comprimida se deu por meio de critério de balanceamento fonético. Foi utilizado um programa de edição (Free audio editor) para recortar as palavras do teste original e gerar um novo arquivo. Mantiveram-se 25 palavras das 50 já existentes no teste original. Para validação foram avaliados 73 indivíduos com faixa etária de 16 a 30 anos, os quais realizaram os procedimentos de Anamnese Audiológica e de Processamento Auditivo, Inspeção Visual do Meato Acústico Externo, Audiometria Tonal Liminar, Logoaudiometria, Medidas de Imitância Acústica, Testes Dióticos, Teste de Fala Comprimida e Teste de Fala Comprimida adaptado. Resultados: foi gerada a nova lista com 25 palavras dissilábicas. Houve diferença entre o número de erros e entre as porcentagens de acertos por orelha, entre o Teste de Fala Comprimida e o adaptado. A força de correlação entre os testes foi moderada, não podendo assim ser utilizados os valores de referência do teste original no instrumento adaptado. Conclusão: foi possível adaptar o Teste de Fala Comprimida constatando uma correlação moderada entre os testes, e verificar a aplicabilidade do Teste de Fala Comprimida Adaptado. Sendo assim, o teste adaptado necessita de novos valores de referência, isto é, de critérios de normalidade.

ABSTRACT Purpose: to adapt the Time-Compressed Speech Test and verify the applicability of the new test, correlating it with the original. Methods: the adaptation of Time-Compressed Speech Test was possible through the criterion of phonetic balance. It was used an editing software called Free Audio Editor to cut the words from the original test and generate the new file. From fifty words already presented in the original test, twenty-five were preserved in the new test. For the validation, were evaluated 73 subjects aged from 16 to 30 years-old, who performed the procedures of Hearing Anamnesis and of Auditory Processing, Visual Inspection of the External Acoustic Canal, Pure Tone Audiometry, Speech Audiometry, Acoustic Immittance Measurement, Diotic Testing, original Time-Compressed Speech Test and adapted Time-Compressed Speech Test. Results: a new list of 25 two-syllable words was created. There were differences between the number of errors and the percentages of correct answers by ear, between the original and the adapted Time-Compressed Speech Test. The strength correlation between the tests was moderate, thus the reference values of the original test could not be used in the adapted instrument. Conclusion: it was possible to adapt the Time-Compressed Speech Test observing a moderate correlation between both tests and verify the applicability of the Adapted Time-Compressed Speech Test. Therefore, the adapted test needs new reference values, in other words, normality criteria are required.

Benchmarks for the Dichotic Sentence Identification test in Brazilian Portuguese for ear and age / Valores de referência para o teste de identificação de sentenças dicóticas em português brasileiro segundo orelha e idade

ABSTRACT INTRODUCTION: Dichotic listening tests should be used in local languages and adapted for the population. OBJECTIVE: Standardize the Brazilian Portuguese version of the Dichotic Sentence Identification test in normal listeners, comparing the performance for age and ear. METHODS: This prospective study included 200 normal listeners divided into four groups according to age: 13-19 years (GI), 20-29 years (GII), 30-39 years (GIII), and 40-49 years (GIV). The Dichotic Sentence Identification was applied in four stages: training, binaural integration and directed sound from right and left. RESULTS: Better results for the right ear were observed in the stages of binaural integration in all assessed groups. There was a negative correlation between age and percentage of correct responses in both ears for free report and training. The worst performance in all stages of the test was observed for the age group 40-49 years old. CONCLUSIONS: Reference values for the Brazilian Portuguese version of the Dichotic Sentence Identification test in normal listeners aged 13-49 years were established according to age, ear, and test stage; they should be used as benchmarks when evaluating individuals with these characteristics.

RESUMO Introdução: Os testes de escuta dicótica devem ser utilizados na língua nativa e adaptados para a população alvo. Objetivo: Estabelecer critérios de referência para o teste DSI em indivíduos normouvintes segundo a orelha, faixa etária e etapa do teste. Método: Estudo prospectivo transversal com 200 indivíduos normouvintes, separados em quatro grupos: 13 a 19 anos (GI), 20 a 29 anos (GII), 30 a 39 anos (GIII) e 40 a 49 anos (GIV). O teste DSI foi aplicado em quatro etapas: Treino, integração binaural, escuta direcionada direita e esquerda. Resultados: Foram observados melhores resultados para a orelha direita nas etapas de integração binaural em todos os grupos avaliados. Houve correlação negativa entre a porcentagem de acertos e a idade, bilateralmente, para as etapas de treino e integração binaural. O pior desempenho, em todas as etapas do teste, foi observado para a faixa etária de 40 a 49 anos de idade. Conclusões: Os valores de referência para a versão em português brasileiro do teste DSI em indivíduos normouvintes de 13 a 49 anos de idade foram estabelecidos segundo a idade, orelha e etapa do teste e devem ser utilizados como padrões de referência na avaliação dos indivíduos com essas características.

Elaboração da versão em Português Brasileiro do teste de identificação de sentenças dicóticas (DSI) / Development of the Brazilian Portuguese version of the dichotic sentence identification test (DSI)

OBJETIVO: Elaborar a versão em Português Brasileiro do teste de identificação de sentenças dicóticas (DSI). MÉTODOS: As frases utilizadas na versão em Português do teste DSI foram extraídas do teste de identificação de sentenças sintéticas (SSI) em Português. Foi realizado um método de combinação onde formaram-se 45 pares de frases, sendo cada par de frases correspondente a um item de teste. Os itens foram distribuídos em cinco grupos, sendo que cada grupo corresponde a uma condição do teste. Para realizar essa disposição, criou-se um programa na plataforma computacional Matlab 7.0, que utilizou o método Monte Carlo ou método da força bruta para obter a distribuição dos itens em cada grupo. As frases foram editadas, gravadas e finalizadas; para a manipulação dos dados utilizou-se o programa CaKewalk Sonar 2.2. Para a finalização, acabamento e padronização da gravação utilizou-se o programa Sound Forge. RESULTADOS: Como resultado obteve-se um compact disc com seis faixas para serem aplicadas na avaliação comportamental do processamento auditivo, nas seguintes etapas: calibração, treino, integração binaural, escuta direcionada à direita e escuta direcionada à esquerda. Para o uso na reabilitação dos distúrbios do processamento auditivo recomenda-se o uso das faixas de calibração e de treinamento auditivo. CONCLUSÃO: O compact disc com a versão em Português Brasileiro do teste DSI constou de seis faixas: calibração, treino, integração binaural, escuta direcionada à direita, escuta direcionada à esquerda e treinamento auditivo.

PURPOSE: To develop the Brazilian Portuguese version of the dichotic sentence idenfication test (DSI). METHODS: The sentences used in the Portuguese version of DSI test were extracted from the Synthetic Sentece Identification (SSI) test in Portuguese. A method of combination was used in order to create 45 pairs of sentences, each pair corresponding to a test item. The items were divided into five groups, each group corresponding to a specific test condition. For this purpose, a computer program was designed in the Matlab 7.0 computing platform, using the Monte Carlo method, or the brute force method, for the distribution of items in each group. The sentences were edited, recorded and finalized; for manipulation of data it was used the CaKewalk Sonar 2.2 program. Finally, the Sound Forge program was used for finalization and standardization of the recording. RESULTS: As a result, a compact disc with six tracks was obtained to be used as part of the behavioral assessment of auditory processing, including the following tracks: calibration, practice, binaural integration, directed attention to the right and to the left ears. The use of the calibration and auditory training tracks is recommended for use in the rehabilitation of auditory processing disorders. CONCLUSION: The compact disc with the Brazilian Portuguese version of the DSI test consisted of six tracks: calibration, practice, binaural integration, directed attention to the right and to the left ears and auditory training.

Development of Compact Disks in Speech Audiometry and First Stage Verification for the New Speech Materials / 대한이비인후과학회지

BACKGROUND AND OBJECTIVES: To standard test method of Korean speech audiometry, we developed an audio compact disk, the KsoM-07 (Korean Society of Otolaryngolgy, Monosyllabic, 2007) word lists and evaluated them for clinical usage. SUBJECTS AND METHOD: The monitored live voice (MLV) of a female talker was digitally recorded with a Beltone audiometer and the GoldWave program. The calibration for each word was done by calculating the root-mean-square (RMS) at the moving time of 300 ms using MATLAB(R) (7.0 version, The MathWorks, Inc. Natick, Massachusetts, USA) and the GoldWave (v 5.25 GoldWave Inc. St. John's, NL, Canada) program. The discrimination score was measured in 20 normal subjects to obtain psychometric function curve of each word and to compare the homogeneity of two lists. The first verification methods for new speech materials were done by comparing the discrimination scores of the flat type and the descending types at 150 sensorineural hearing loss with similar pure tone threshold. RESULTS: 44% of MLV recording was above +/-3 dB deviation and had to be recalibrated. Two lists showed equal difficulty or homogeneity in normal subjects. Discrimination scores were statistically different among the groups with respect to hearing level and the slope of audiogram. CONCLUSION: KsoM-07 word lists were relatively homogeneous and showed differences between the descending type and the flat type of hearing loss. Therefore, KsoM-07 word lists can be useful in evaluating the impairment found in the descending type of sensorineural hearing loss as well as in serving as a new speech test material for Korean monosyllabic words.

Compressão de freqüências e suas implicações no reconhecimento de fala / Frequency compression and its effects in speech recognition

TEMA: compressão de freqüências. OBJETIVO: avaliar o índice percentual de reconhecimento de fala (IPRF) utilizando compressão de freqüências em três razões diferentes. MÉTODOS: palavras monossílabas foram gravadas utilizando um algoritmo de compressão de freqüências em três razões: 1:1, 2:1, 3:1, gerando três listas de palavras. Dezoito normo-ouvintes realizaram o IPRF utilizando as listas de palavras modificadas. Foram subdivididos em dois grupos, considerando a familiaridade com o material de fala gravado: grupo de fonoaudiólogas (F) e grupo de acompanhante de pacientes (P). RESULTADOS: observou-se uma piora estatisticamente significante no IPRF quando se utilizou compressão de freqüências. O grupo F teve melhor desempenho que o grupo P em todas as razões de compressão aplicadas. CONCLUSÃO: a compressão de freqüências dificulta o reconhecimento da fala, sendo que, quanto maior a razão de compressão, maior é a dificuldade. A familiaridade com as palavras facilita o seu reconhecimento em qualquer condição de escuta.

BACKGROUND: frequency compression. AIM: to evaluate the index of speech recognition (IPRF) using frequency compression in three different ratios. Methods: monosyllabic words were recorded using an algorithm of frequency compression in three ratios: 1:1, 2:1, 3:1, generating three lists of words. Eighteen listeners accomplished the IPRF using the modified words. They were subdivided in two groups, considering familiarity with the speech material: group of audiologists (F) and group of patients (P). RESULTS: a statistically significant decrease in accuray was observed when using frequency compression. Group F presented a better performance than Group P in all of the applied ratio frequency compression ratios. CONCLUSION: Frequency compression hinders speech recognition; as the compression ratio increases, so does the level of difficulty. Familiarity with the words facilitates recognition in any hearing condition.

A Study for the Acoustic Characteristics of PB Word List according to Frequency / 대한이비인후과학회지

BACKGROUND AND OBJECTIVES: The Korean phonetically balanced (PB) word list (Ham's Korean PB Word List) is currently the most commonly used word list for speech discrimination test. The aim of this study was to investigate the frequency characteristics of the Ham's Korean PB word list and to compare frequency characteristics of the Ham's Korean PB word list with that of the 21st Century Sejong words, which represent characteristics of the contemporary spoken Korean language. SUBJECTS AND METHOD: The 300 syllables most commonly used were collected from the words in 21st Century Sejong project. The 21st Century Sejong words were weighted according to the frequency of usage. The 21st Century Sejong words and words from the Ham's Korean PB word list were analyzed using an acoustic analysis program that has the MATLAB function. RESULTS: Syllables with frequencies above 2,000 Hz comprise 34% of the 21st Century Sejong words and 16% of the Ham's Korean PB word list. The Ham's Korean PB word list does not represent the high frequency characteristics of the contemporary spoken Korean language. The 21st Century Sejong words have more 'eu', 'i' and open syllables than the Ham's Korean PB word list. CONCLUSION: The Ham's Korean PB word list does not reflect the high frequency characteristics of the contemporary spoken Korean language included in the 21st Century Sejong words. Therefore, a new PB word list that incorporates a greater number of syllables with high frequency characteristics is required.

Frequencies of Korean Syllables and the Distribution of Syllables of PB Word List / 대한이비인후과학회지

BACKGROUND AND OBJECTIVES: There can be 3, 192 syllables in Korean language but the frequencies of each syllables are variable. This study was aimed to investigate the frequencies of each Korean syllable in the everyday dialogue and to compare the result with the existing Korean phonetically balanced (PB) word list, which can provide the basic data for the standardization of Korean PB word list. MATERIALS AND METHOD: Using 33 materials (total 537, 245 syllables) collected from TV drama (MBC Best Theater), the author calculated the frequencies of each Korean syllables composed of initial consonant and vowel and compared with existing PB word list by computer program. RESULTS: Each syllable appeared in the different frequency pattern. The 138 of 399 syllables represented the 95% of everyday dialogue. CONCLUSION: This study showed that established PB word list was made relatively well but needs somewhat modification to reflect the actual frequencies of syllable in colloquialism.