Correlates of vowel clarity in the spectrotemporal modulation domain: Application to speech impairment evaluation.

This article reports on vowel clarity metrics based on spectrotemporal modulations of speech signals. Motivated by previous findings on the relevance of modulation-based metrics for speech intelligibility assessment and pathology classification, the current study used factor analysis to identify regions within a bi-dimensional modulation space, the magnitude power spectrum, as in Elliott and Theunissen [(2009). PLoS Comput. Biol. 5(3), e1000302] by relating them to a set of conventional acoustic metrics of vowel space area and vowel distinctiveness. Two indices based on the energy ratio between high and low modulation rates across temporal and spectral dimensions of the modulation space emerged from the analyses. These indices served as input for measurements of central tendency and classification analyses that aimed to identify vowel-related speech impairments in French native speakers with head and neck cancer (HNC) and Parkinson dysarthria (PD). Following the analysis, vowel-related speech impairment was identified in HNC speakers, but not in PD. These results were consistent with findings based on subjective evaluations of speech intelligibility. The findings reported are consistent with previous studies indicating that impaired speech is associated with attenuation in energy in higher spectrotemporal modulation bands.

Validation of an Intelligibility Test Based on Acoustic-Phonetic Decoding of Pseudo-Words: Overall Results from Patients with Cancer of the Oral Cavity and the Oropharynx.

OBJECTIVES: Loss of intelligibility is a major complaint for patients with speech disorders, as it affects their everyday communication and thus contributes to a decrease in their quality of life. Several tests are available to measure intelligibility, but these tests do not take into account the evaluators' ability to restore distorted sequences. Due to this ability, the evaluator will tend to recognize words despite phonetic distortions, and speech production deficit can go undetected. The results of these tests therefore overestimate the intelligibility of patients and may mask real functional limitations. We propose a new test which uses a large number of pseudowords in order to neutralize the unwanted perceptual effects that cause this overestimation. The purpose of this test is to measure the speech production deficit. It is not intended to assess the communication deficit. Our objective is to validate this test based on acoustic-phonetic decoding of productions from patients with speech disorders. MATERIALS AND METHODS: We tested this method with a population of 39 healthy participants and 78 post-treatment patients with cancers of the oral cavity and the oropharynx (HNC patients). Each speaker produced 52 pseudowords taken from randomly generated lists from large common dictionary, each list of 52 pseudowords containing the same number of phonemes. Forty everyday listeners then transcribed these productions. The orthographic transcriptions were phonetized and compared to the expected phonetic forms. An algorithm provided a Perceived Phonological Deviation score (PPD) based on the number of features that differed between the expected forms and the transcribed items. The PPD thus provided a score representing the loss of intelligibility. RESULTS: The 39 participants in the control group demonstrated significantly lower PPD scores compared to the 41 patients with a T1T2 tumor size or compared to the 37 patients with a T3T4 tumor size. The differences between the three groups were significant. If we use the PPD as a predictor to identify patients versus control group subjects, the AUC of the ROC curve is equal to 0.94, which corresponds to an outstanding group separability. A PPD threshold at 0.6 features per phoneme is the boundary between normal and dysfunctional speech. The analysis showed a close correlation between the PPD and a clinical judgment of the disorder severity obtained from experts. CONCLUSION: This test appears to be effective in measuring the intelligibility of speakers at a phonological level, in particular in the case of head and neck cancers.

Construction of an automatic score for the evaluation of speech disorders among patients treated for a cancer of the oral cavity or the oropharynx: The Carcinologic Speech Severity Index.

BACKGROUND: Speech disorders impact quality of life for patients treated with oral cavity and oropharynx cancers. However, there is a lack of uniform and applicable methods for measuring the impact on speech production after treatment in this tumor location. OBJECTIVE: The objective of this work is to (1) model an automatic severity index of speech applicable in clinical practice, that is equivalent or superior to a severity score obtained by human listeners, via several acoustics parameters extracted (a) directly from speech signal and (b) resulting from speech processing and (2) derive an automatic speech intelligibility classification (i.e., mild, moderate, severe) to predict speech disability and handicap by combining the listener comprehension score with self-reported quality of life related to speech. METHODS: Eighty-seven patients treated for cancer of the oral cavity or the oropharynx and 35 controls performed different tasks of speech production and completed questionnaires on speech-related quality of life. The audio recordings were then evaluated by human perception and automatic speech processing. Then, a score was developed through a classic logistic regression model allowing description of the severity of patients' speech disorders. RESULTS: Among the group of parameters subject to extraction from automatic processing of the speech signal, six were retained, producing a correlation at 0.87 with the perceptual reference score, 0.77 with the comprehension score, and 0.5 with speech-related quality of life. The parameters that contributed the most are based on automatic speech recognition systems. These are mainly the automatic average normalized likelihood score on a text reading task and the score of cumulative rankings on pseudowords. The reduced automatic YC2SI is modeled in this way: YC2SIp  = 11.48726 + (1.52926 × Xaveraged normalized likelihood reading ) + (-1.94e-06 × Xscore of cumulative ranks pseudowords ). CONCLUSION: Automatic processing of speech makes it possible to arrive at valid, reliable, and reproducible parameters able to serve as references in the framework of follow-up of patients treated for cancer of the oral cavity or the oropharynx.

Stimulation of the pedunculopontine nucleus area in Parkinson's disease: effects on speech and intelligibility.

Improvement of gait disorders following pedunculopontine nucleus area stimulation in patients with Parkinson's disease has previously been reported and led us to propose this surgical treatment to patients who progressively developed severe gait disorders and freezing despite optimal dopaminergic drug treatment and subthalamic nucleus stimulation. The outcome of our prospective study on the first six patients was somewhat mitigated, as freezing of gait and falls related to freezing were improved by low frequency electrical stimulation of the pedunculopontine nucleus area in some, but not all, patients. Here, we report the speech data prospectively collected in these patients with Parkinson's disease. Indeed, because subthalamic nucleus surgery may lead to speech impairment and a worsening of dysarthria in some patients with Parkinson's disease, we felt it was important to precisely examine any possible modulations of speech for a novel target for deep brain stimulation. Our results suggested a trend towards speech degradation related to the pedunculopontine nucleus area surgery (off stimulation) for aero-phonatory control (maximum phonation time), phono-articulatory coordination (oral diadochokinesis) and speech intelligibility. Possibly, the observed speech degradation may also be linked to the clinical characteristics of the group of patients. The influence of pedunculopontine nucleus area stimulation per se was more complex, depending on the nature of the task: it had a deleterious effect on maximum phonation time and oral diadochokinesis, and mixed effects on speech intelligibility. Whereas levodopa intake and subthalamic nucleus stimulation alone had no and positive effects on speech dimensions, respectively, a negative interaction between the two treatments was observed both before and after pedunculopontine nucleus area surgery. This combination effect did not seem to be modulated by pedunculopontine nucleus area stimulation. Although limited in our group of patients, speech impairment following pedunculopontine nucleus area stimulation is a possible outcome that should be considered before undertaking such surgery. Deleterious effects could be dependent on electrode insertion in this brainstem structure, more than on current spread to nearby structures involved in speech control. The effect of deep brain stimulation on speech in patients with Parkinson's disease remains a challenging and exploratory research area.

Top-down mechanisms in dysphonia perception: the need for blind tests.

The purpose of this study was to determine the extent to which the information a therapist or a physician has about a dysphonic speaker, particularly whether he or she is in the pretreatment or posttreatment period, can influence judgments of the patient's voice. The voices of 53 dysphonic speakers were used in the study. For each speaker, we selected a pair of voice samples recorded under different circumstances. Seven listeners who were speech therapists, ear, nose, and throat surgeons, or voice pathologists took blind-listening tests in which they were asked to compare the two voices in each pair (phase 1: blind listening). A few weeks later, the listeners took the very same test again, except that this time, they were given bogus information about whether the speaker had/had not been treated by laryngeal surgery or speech therapy (phase 2: influenced listening). The information given for each voice sample either reinforced the judgment made in phase 1 (eg, the voice judged to be better on the blind test was said to be posttreatment) or countered that judgment (eg, the voice rated as better on the blind test was said to be pretreatment). The influenced-listening results showed that in the reinforced condition, the original ratings were significantly amplified. By contrast, in the countering-influence condition, decision changes were frequent: we found that judgment reversals and the countering-information scores were almost independent of the blind-listening scores. These findings point out the dire need to use a blind protocol in perceptual assessments of dysphonia.

Análise das medidas aerodinâmicas no português brasileiro por meio do método multiparamétrico de avaliação vocal objetiva assistida (EVA) / Analysis of aerodynamic measures in Brazilian portuguese through multiparameter assessment method vocal objective assist (EVA)

OBJETIVO: definir medidas aerodinâmicas em falantes do português brasileiro, sem queixas vocais, obtidas pelo programa EVA. MÉTODO: participaram do estudo 20 homens e 20 mulheres, que tiveram suas vozes analisadas. RESULTADOS: os valores médios referentes à pressão subglótica para vozes femininas e masculinas foram, respectivamente: Pressão subglótica= 5,84 hPa e 6,7 hPa, média da intensidade= 79,21 dB e 81,7 dB, média do fluxo oral= 0,09 dm³/s e 0,16 dm³/s, eficiência glótica= 13,87 dB/hPa e 12,78 dB/hPa, eficiência laríngea= 188,08 dB/(hPa.dm³/s) e 97,11 dB/(hPa.dm³/s), resistência laríngea= 96,26 hPa/(dm³/s) e 52,64 hPa/(dm³/s), média da frequência fundamental (F0)= 208,28 Hz e 136,56 Hz e pico da pressão do fluxo oral= 0,093 e 0,098. Houve diferença estatisticamente significante em relação ao gênero para as medidas de média de intensidade, eficiência laríngea, resistência laríngea e média da frequência fundamental. Os valores médios referentes ao fluxo oral para vozes femininas e masculinas foram respectivamente: Pressão subglótica= 6,05 hPa e 6,6 hPa, média da intensidade= 65,50 dB e 66,3 dB, média do fluxo oral= 0,10 dm³/s e 0,13 dm³/s, eficiência glótica= 11,12 dB/hPa e 11,77 dB/hPa, eficiência laríngea= 144,83 dB/(hPa.dm³/s) e 97,89 dB/(hPa.dm³/s), resistência laríngea= 78,98 hPa/(dm³/s) e 61,81 hPa/(dm³/s), média da F0= 222,52 Hz e 139,20 Hz e pico da pressão do fluxo oral= 0,047 e 0,053. CONCLUSÃO: o EVA é um programa ainda novo no Brasil, e a análise de medidas aerodinâmicas, em falantes do português brasileiro, permite a obtenção de valores de referência, possibilitando assim comparações com estudos futuros.

PURPOSE: to define aerodynamic measures in Brazilian Portuguese speakers without voice complaints, obtained by the EVA program. METHOD: the study included 20 men and 20 women who had their voices analyzed. RESULTS: the mean values with subglottic pressure for female and male voices were: Subglottic pressure = 5.84 hPa and 6.7 hPa; average intensity = 79.21dB and 81.7dB; oral mean flow = 0.09 dm3/s and 0.16 dm3/s; glottic efficiency = 13.87dB/hPa and 12.78 dB/hPa; laryngeal efficiency = 188.08 dB/(hPa.dm3/s) and 97.11dB/(hPa.dm3/s); laryngeal resistance = 96.26 hPa/(dm3/s) and 52.64 hPa/(dm3/s); mean fundamental frequency (F0) = 208.28 Hz and 136.56 Hz and peak mouth pressure flow = 0.093 and 0.098. There was a statistically significant difference in relation to gender for the measures of average intensity, efficiency, larynx, laryngeal resistance and mean fundamental frequency. The average flow for the oral female and male voices were: subglottic pressure = 6.05 hPa and 6.6 hPa; average intensity = 65.50 dB and 66.3 dB; oral mean flow = 0, 10 dm3/s and 0.13 dm3/s; glottic efficiency = 11.12 dB/hPa and 11.77 dB/hPa; laryngeal efficiency = 144.83 dB/( hPa.dm3/s) and 97.89 dB/( hPa.dm3/s); laryngeal resistance = 78.98 hPa/(dm3/s) and 61.81 hPa/( dm3/s); average F0 = 222.52 Hz and 139.20 Hz and peak pressure oral flow = 0.047 and 0.053. CONCLUSION: EVA is a new program in Brazil, and analysing the aerodynamic measures in Brazilian Portuguese speakers, allows us to obtain reference values, thus allowing comparisons with future studies.

Medidas eletroglotográficas em falantes do português brasileiro por meio do Método Multiparamétrico de Avaliação Vocal Objetiva Assistida (EVA) / Electroglottography of speakers of Brazilian Portuguese through Objective Multiparameter Vocal Assessment (EVA)

O Método Multiparamétrico de Avaliação Vocal Objetiva Assistida (EVA) foi projetado para o estudo da maioria dos parâmetros de produção da fala. OBJETIVO: Definir as medidas médias dos parâmetros eletroglotográficos em falantes do português brasileiro para o EVA. MATERIAL E MÉTODO: Foram analisadas 40 vozes, 20 homens e 20 mulheres sem queixa vocal, extraindo-se as medidas eletroglotográficas, a fim de obter valores de referência de normalidade. Estudo de caso: estudo descritivo com corte transversal. RESULTADOS: Os valores médios de normalidade encontrados nas vozes masculinas foram: F0 = 127,77 Hz, coeficiente de variação de F0 = 2,51%, jitter absoluto = 1,707 Hz, perturbação média relativa = 0,0083, jitter factor = 1,34%, jitter ratio = 13,45%, e QF = 0,447. Para vozes femininas, foram: F0 = 204,87 Hz, coeficiente de variação de F0 = 1,58%, jitterabsoluto = 3,30 Hz, perturbação média relativa = 0,0102, jitter factor = 1,60%, jitter ratio = 16,23%, e QF = 0,443. O tipo de onda foi em 100% da amostra classificada como pulso inclinado em ambos os gêneros. CONCLUSÃO: Houve diferença estatisticamente significante em relação ao gênero para os parâmetros de média F0 e jitterabsoluto. Ao utilizar um programa de análise acústica, os usuários devem basear-se em parâmetros inerentes ao próprio programa para realizar a análise dos dados coletados.

EVA was designed to study various speech production parameters. OBJECTIVE: This paper aims to define the mean values for electroglottography tests of Brazilian Portuguese speakers on EVA. MATERIALS AND METHOD: The voices of 20 men and 20 women without voice-related complaints were analyzed through electroglottography so as to obtain reference values for normality. Case study: this is a descriptive cross-sectional study. RESULTS: The mean values for normal male voices were: F0 = 127.77 Hz; F0 coefficient of variation = 2.51%; absolute jitter = 1.707 Hz; relative average perturbation = 0.0083; jitter factor = 1.34%; jitter ratio = 13.45%; QF = 0.447. The values for female voices were: F0 = 204.87 Hz; F0 coefficient of variation = 1.58%; absolute jitter = 3.30Hz; relative average perturbation = 0.0102; jitter factor = 1.60%; jitter ratio = 16.23%; QF= 0.443. Wave type for the entire sample was categorized as tilted pulse. CONCLUSION: Statistically significant differences were found for gender on parameters average FO and absolute jitter. While using acoustic analysis software, users must be based on parameters inherent to the software program when analyzing the collected data.

Botulinum toxin injection and airflow stability in spasmodic dysphonia.

OBJECTIVE: The aim of this study was to analyze the effects of botulinum toxin (BT) injection on airflow stability, by measuring mean phonatory oral airflow and its coefficient of variation (CV), in subjects with adductor spasmodic dysphonia (SD). STUDY DESIGN AND SETTING: Twenty-four subjects with SD (aged 31-78 years) and 23 controls (aged 29-63 years) were evaluated for mean airflow and its CV during sustained phonation. Fifteen of the subjects with SD were also evaluated within 3 weeks after BT injection. RESULTS: BT increased airflow in subjects (P = 0.0130) but neither the preinjection nor postinjection values differed significantly from those of controls. Conversely, airflow CV was invariably higher in subjects than in controls (P < 0.0001). In 13 subjects in whom phonation perceptually improved, including 3 in whom airflow did not increase, airflow CV decreased significantly after BT treatment (P = 0.0232). CONCLUSIONS: Subjects with SD have highly unstable phonatory airflow; its CV is a valid measure for assessing the outcome of a BT injection. A reduced airflow CV probably does not depend solely on increased airflow due to thyroarytenoid muscle paresis, and may indicate a change in laryngeal motoneuronal activity. EBM RATING: B-3b.