When the primary mode of communication is speech, the crucial ingredient for successful communication is intelligible speech. The speech of children with disordered phonologies is often unintelligible. Accurate and reliable measurement of speech with compromised intelligibility is essential if appropriate treatment procedures are to be chosen and implemented. The focus of this investigation was the measurement of speech intelligibility in young children. The purpose of this study was to investigate the relationship between the subjective method of trained ear estimation and the objective method of orthographic transcription when measuring the speech intelligibility of young speakers with a wide range of phonological profiency. For this study, the standard measurement of intelligibility was operationally defined as the percentage of words understood in a continuous speech sample derived from orthographic transcription of the sample. The secondary purpose was to investigate the accuracy of the speech-language pathologists' estimates as compared to the standard measure for each of the three groups: (a) the children with the most intelligibility, (b) with average intelligibility, and (c) with the least intelligibility. Data were collected from 47 children, aged 4:0 to 5:6, who comprised three groups with varying levels of intelligibility. Two groups of listeners who were unfamiliar with the speakers, but familiar with the topic, rated the children's percentage of intelligibility from continuous speech samples via orthographic transcription or trained ear estimation. The two methods of measuring speech intelligibility investigated in this study were found to correlate highly (£ = .96). However, there was a significant difference between the percentages derived from orthographic transcription and those derived from trained ear estimation for some speakers. The 1-test analyses revealed significant differences between the two measures for the two most intelligible groups, and no significant difference for the least intelligible group. It appears that the subjective method of estimating speech intelligibility with trained ears correlates with the objective method of orthographic transcription, but yields a different percentage score for some speakers.

Linguistic diversification within our public schools has demanded professional responsibility from speech-language pathologists (SLPs) serving nonstandard English speaking students. Understanding and recognizing normal cultural linguistic differences is the responsibility of the SLP. This study focused on the relationship of three speech features to intelligibility ratings of 10 preschool aged Black English speakers as assigned by 4 licensed standard English speaking SLPs with varying experience. The SLPs also rated the perceived effect of these speech features (i.e., articulation, speaking rate, and resonance) on intelligibility. Using the Pearson product-moment correlation, ratings were correlated and found to demonstrate an association between intelligibility ratings and all three speech features assessed. To determine which speech feature affected intelligibility the most, a linear association using a stepwise regression was applied to all listeners' ratings. For 3 of the 4 listeners, the strongest association between intelligibility and articulation. Ratings of the 4th listener, the listener with the most experience(> 3 years) demonstrated the strong association between intelligibility and resonance. The listener with _the least amount of experience tended to assign higher severity ratings to ratings for intelligibility, rate, and resonance than did the other listeners. Findings from this study demonstrate a need for more studies within the area of Black English as well as further investigative studies to assess listeners' perception of dialectical differences based on the experience within linguistically different or similar communities. Various measures of intelligibility of Black English speakers should also be explored for more accurate assessment tools for this population. Clinical implications focus on the SLP's responsibility to be experienced and knowledgeable of the linguistic community they are serving.

Speech signals radiated in confined spaces are subject to reverberation due to reflections of surrounding walls and obstacles. Reverberation leads to severe degradation of speech intelligibility and can be prohibitive for applications where speech is digitally recorded, such as audio conferencing or hearing aids. Dereverberation of speech is therefore an important field in speech enhancement. Driven by consumer demand, blind speech dereverberation has become a popular field in the research community and has led to many interesting approaches in the literature. However, most existing methods are dictated by their underlying models and hence suffer from assumptions that constrain the approaches to specific subproblems of blind speech dereverberation. For example, many approaches limit the dereverberation to voiced speech sounds, leading to poor results for unvoiced speech. Few approaches tackle single-sensor blind speech dereverberation, and only a very limited subset allows for dereverberation of speech from moving speakers. Therefore, the aim of this dissertation is the development of a flexible and extendible framework for blind speech dereverberation accommodating different speech sound types, single- or multiple sensor as well as stationary and moving speakers. Bayesian methods benefit from – rather than being dictated by – appropriate model choices. Therefore, the problem of blind speech dereverberation is considered from a Bayesian perspective in this thesis. A generic sequential Monte Carlo approach accommodating a multitude of models for the speech production mechanism and room transfer function is consequently derived. In this approach both the anechoic source signal and reverberant channel are estimated using their optimal estimators by means of Rao-Blackwellisation of the state-space of unknown variables. The remaining model parameters are estimated using sequential importance resampling. The proposed approach is implemented for two different speech production models for stationary speakers, demonstrating substantial reduction in reverberation for both unvoiced and voiced speech sounds. Furthermore, the channel model is extended to facilitate blind dereverberation of speech from moving speakers. Due to the structure of measurement model, single- as well as multi-microphone processing is facilitated, accommodating physically constrained scenarios where only a single sensor can be used as well as allowing for the exploitation of spatial diversity in scenarios where the physical size of microphone arrays is of no concern. This dissertation is concluded with a survey of possible directions for future research, including the use of switching Markov source models, joint target tracking and enhancement, as well as an extension to subband processing for improved computational efficiency.

Thesis (M. Sc.)--University of Hong Kong, 1998. / Includes bibliographical references (leaf 32-34).

Thesis (Au.D.)--James Madison University, 2010. / Includes bibliographical references.

Thesis (Ph. D.)--University of Texas at Austin, 2006. / Vita. Includes bibliographical references.

O objetivo principal deste estudo foi avaliar a influência da adaptação da prótese rebaixadora de palato na inteligibilidade e na ressonância de fala, e nas características acústicas espectrográficas dos três primeiros formantes das vogais orais do português brasileiro de pacientes submetidos à glossectomia. Participaram 36 pacientes, 33 do sexo masculino e 3 do sexo feminino, com idades entre 30 e 80 anos (Média=53,91±10,53 anos), sendo 14 submetidos à glossectomia total, 12 à glossectomia total e mandibulectomia parcial, 6 à hemiglossectomia e, 4 à glossectomia subtotal. Amostras de fala (conversa espontânea e repetição de 18 sílabas) de pacientes com e sem prótese foram randomizadas e a ressonância e inteligibilidade de fala foram julgadas por 2 fonoaudiólogos. Também foi realizada a análise espectrográfica dos formantes das 7 vogais do português brasileiro com e sem prótese, extraindo-se as médias dos três primeiros formantes. Houve melhora significante na inteligibilidade de fala e de sílabas após adaptação da prótese rebaixadora de palato (p < 0,001). Não houve diferença estatística nos julgamentos da ressonância com e sem prótese. Existiu diferença significante na situação com e sem prótese para o primeiro formante nas vogais /a/, /e/, /u/ (p < 0,001) e tendência estatística na vogal /o/ (p = 0,09); para segundo formante nas vogais /o/, / /, /u/ (p < 0,001), e tendência estatística nas vogais /e/ (p = 0,058) e /i/ (p = 0,080) e para o terceiro formante nas vogais /a/ e / / (p < 0,001). A prótese rebaixadora de palato melhorou a inteligibilidade da fala espontânea e das sílabas dos pacientes glossectomizados. Também aumentou os valores de F2 e F3 para todas as vogais e, de F1 para as vogais /o/, / /, /u/. Isto gerou uma aproximação dos valores dos formantes da maioria das vogais junto aos valores de normalidade. / The main objective of this study was to assess the influence of the palatal augmentation prostheses on speech intelligibility, ressonance and acoustic spectrographic characteristics of the first three formants of oral vowels in Brazilian Portuguese, in patients submitted to glossectomy. Thirty-six (33 men and 3 women), aged 30 to 80 years (mean = 53.9 ±10.5 years) underwent glossectomy (14 = total glossectomy; 12 = total glossectomy associated partial mandibulectomy; 6=hemiglossectomy, and 4 = subtotal glossectomy) constituted the study sample. Speech samples (spontaneous conversation and repetition of 18 syllables) from the patients with and without prosthesis were randomly distributed and the resonance and speech intelligibility were judged by 2 speech language pathologist. A spectrographic assessment of formants of the 7 vowels of Brazilian Portuguese, with and without prosthesis, was performed. The mean values of the first three formants were extracted. There was a statistically significant improvement of the spontaneous speech intelligibility and of the average number of correctly identified syllables with the use of the prosthesis (p<0.05). The speech resonance results have not demonstrated a significant reduction in hypernasality with the palatal augmentation prostheses in place. Statistically significant differences occurred in F1 values for the vowels /a/, /e/, /u/, for F2, there was a significant difference in the vowels /o/, / /, /u/; and for F3, there was a significant difference in the vowels /a/, / / (p < 0.001). Palatal augmentation prosthesis improved intelligibility of spontaneous speech and syllables in glossectomized patients. It also increased F2 and F3 values for all vowels and F1 values for the vowels /o/, / /, /u/. This effect has brought the values of many vowel formants closer to normality.

Intelligible speech is a primary component for successful communication. However, the speech of children with disordered phonologies is often unintelligible. Therefore, when assessing the speech intelligibility of children in order to determine whether they qualify for intervention services, speech-language pathologists need reliable evaluation tools. The focus of this investigation was the measurement of speech intelligibility. The purpose of this study was to examine the relationship between two methods for measuring speech intelligibility. The first, identification method, involves the listener transcribing a speech sample from which the percentage of words understood is calculated. The second, scaling procedure, involves the listener estimating the percentage of words understood from a continuous speech sample. The secondary purpose of this study was to examine the accuracy of the scaling method as measured by ear estimation compared to the identification method as measured through orthographic transcription for each of three groups of children with: (a) the most intelligibility, (b) average intelligibility, and ( c) least intelligibility. Four unsophisticated listeners rated the speech intelligibility of 48 speakers aged 4:0 to 5:6 who comprised three groups with varying levels of phonological proficiency. The listeners who were unfamiliar with the speakers, but familiar with the topic, rated the children's continuous speech samples using ear estimation. The data collected were then compared with intelligibility ratings as measured in a previous study (Gordon-Brannan, 1994) via orthographic transcription. The two methods of measuring speech intelligibility examined in this study were found to be positively correlated (r = .86). However, the t-test analysis revealed significant differences between the two measures for the most and least intelligible groups, indicating discrepancies between the two methods when measuring the speech intelligibility of some children. Additional statistical analysis revealed poor intrajudge reliability which should be considered when interpreting the results presented. It does appear, however, that when measuring speech intelligibility, using ear estimation, is reflective of the orthographic transcription measure, although the actual estimated percentages of intelligibility appear to differ from the percentages derived from orthographic transcription.

The purpose of this study was to investigate the effects of a 500 Hz and 3150 Hz one-third octave band augmentation on the speech discrimination ability of normal hearing listeners and whether such effects vary with signal presentation level. The augmented portion of monosyllabic words was systematically varied from 5-55dB above the intensity level of the unfiltered version of the words and presented simultaneously to one ear.

No description available.