Voice conversion with pre-trained representations for audio anonymization

The recording and processing of voice data raises increasing privacy concerns for users and service providers. One way to address these issues is to move processing on the edge device closer to the recording so that potentially identifiable information is not transmitted over the internet. However, this is often not possible due to hardware limitations. An interesting alternative is the development of voice anonymization techniques that remove individual speakers characteristics while preserving linguistic and acoustic information in the data. In this work, a state-of-the-art approach to sequence-to-sequence speech conversion, ini- tially based on x-vectors and bottleneck features for automatic speech recognition, is explored to disentangle the two acoustic information using different pre-trained speech and speakers representation. Furthermore, different strategies for selecting target speech representations are analyzed. Results on public datasets in terms of equal error rate and word error rate show that good privacy is achieved with limited impact on converted speech quality relative to the original method.

The natural reader : a revolution in the art of reading : designed for old and young, at home and at school, adapted to suit every grade of instruction, especially helpful to public speakers and readers : embodying a most important discovery in the art of expression /

Mode of access: Internet.

Verbal fluency in the detection of mild cognitive impairment and Alzheimer`s disease among Brazilian Portuguese speakers: the influence of education

Background: Verbal fluency (VF) tasks are simple and efficient clinical tools to detect executive dysfunction and lexico-semantic impairment. VF tasks are widely used in patients with suspected dementia, but their accuracy for detection of mild cognitive impairment (MCI) is still under investigation. Schooling in particular may influence the subject`s performance. The aim of this study was to compare the accuracy of two semantic categories (animals and fruits) in discriminating controls, MCI patients and Alzheimer`s disease (AD) patients. Methods: 178 subjects, comprising 70 controls (CG), 70 MCI patients and 38 AD patients, were tested on two semantic VF tasks. The sample was divided into two schooling groups: those with 4-8 years of education and those with 9 or more years. Results: Both VF tasks - animal fluency (VFa) and fruits fluency (VFf) - adequately discriminated CG from AD in the total sample (AUC = 0.88 +/- 0.03, p < 0.0001) and in both education groups, and high educated MCI from AD (VFa: AUC = 0.82 +/- 0.05, p < 0.0001; VFf: AUC = 0.85 +/- 0.05, p < 0.0001). Both tasks were moderately accurate in discriminating CG from MCI (VFa: AUC = 0.68 +/- 0.04, p < 0.0001 - VFf:AUC = 0.73 +/- 0.04, p < 0.0001) regardless of the schooling level, and MCI from AD in the total sample (VFa: AUC = 0.74 +/- 0.05, p < 0.0001; VFf: AUC = 0.76 +/- 0.05, p < 0.0001). Neither of the two tasks differentiated low educated MCI from AD. In the total sample, fruits fluency best discriminated CG from MCI and MCI from AD; a combination of the two improved the discrimination between CG and AD. Conclusions: Both categories were similar in discriminating CG from AD; the combination of both categories improved the accuracy for this distinction. Both tasks were less accurate in discriminating CG from MCI, and MCI from AD.

Testing native speakers of German and Portuguese on the understanding of topological operators-line-region relations in gvSIG

Dissertation submitted in partial fulfilment of the requirements for the Degree of Master of Science in Geospatial Technologies

Training and discovering corpus-based data driven exercices in english teaching (L2/FL) to native speakers of portuguese (L1)

Project submitted as part requirement for the degree of Masters in English teaching,

Gender Differences in Spatial Cognition: Wayfinding Performance and Sketch Mapping Activity of German Speakers

Dissertation submitted in partial fulfillment of the requirements for the Degree of Master of Science in Geospatial Technologies

Risky behavior regarding drug use and HIV infection: an Internet questionnaire coupled with short education texts for Portuguese speakers

Surveys of risky behavior relating to HIV/AIDS are generally made for groups at risk of infection, for which HIV/AIDS prevalence is usually expected to be higher than in the general population. Therefore, an educational homepage in Portuguese was created on the Internet to inform/ask internauts regarding knowledge and behavior. The internauts were classified as adolescents (13 to 25 years) and adults (>25 years). The number of STDs was reported as 1. 8 ± 2. 6 infections (range: 1 to 20 infections); 43% used condoms during sexual intercourse. Alcohol consumption was reported by 63% and illicit drug use by 32% (marijuana 24% and inhalants 15%). Among the adolescents, 31% did not classified alcohol as a drug. The adults more frequently reported homosexuality, anal intercourse and STDs, although the adolescents also presented high rates of risky behavior. These results show the need to reach out to internauts through better control strategies. Different types of strategies must be encouraged, in order to reach people that use this means of communication and entertainment.

Comparing heritage speakers and late L2-learners of European Portuguese: verb movement, VP ellipsis and adverb placement

This study compares the performance of Portuguese-German heritage children and adult L2 speakers of European Portuguese whose L1 is German with respect to two aspects of grammar, adverb placement and VP-ellipsis, which depend on a core syntactic property of the language, verb movement. The results show that both groups have acquired V-to-I and adverb placement, showing no influence of a V2 grammar. Performance in the VP-ellipsis task is more complex: heritage children produce VP-ellipsis at the level of controls, as opposed to L2 speakers; however, both L2 and heritage speakers show that crosslinguistic influence may produce a preference for pronoun substitution over VP-ellipsis in a task asking for redundancy resolution. Nevertheless, given that overall results show that heritage children perform at the level of L1 children, we take our results to support approaches to heritage bilingualism which suggest the development of an intact grammar in childhood.

Personality disorders and the Five-Factor Model among French speakers in Africa and Europe

Objective: Several authors have suggested that Personality Disorders (PDs) might be more accurately described using a dimensional model instead of a categorical one. The aim of this study was to describe the relationship between PDs and the Five-Factor Model (FFM)-a dimensional model describing normal personality traits known for its invariance across cultures-in two different cultural settings. Method: Subjects from nine French-speaking African countries (n = 2,014) and from Switzerland (n = 697) completed both the French-version of the IPDE screening questionnaire, assessing the ten DSM-IV PDs, and the French-version of the NEO-PI-R, assessing the five domains and thirty facets of the FFM. Results: Correlations between PDs and the five domains of the FFM were similar in both samples. For example, Neuroticism was highly correlated with Borderline, Avoidant, and Dependent PDs in both Africa and Switzerland. The total rank-order correlation (rho) between the two correlation matrices was very high (rho = 0.93) and significant (P < 0.001), as were the rhos for all domains of the FFM and all PDs, except Paranoid and Dependent PDs. However, the rhos for PDs across facet-scales were all highly significant (P < 0.001). Moreover, 80% of Widiger and colleagues' predictions and 70 % of Lynam and Widiger's prototypes, concerning the relationship between PDs and the FFM, were confirmed in both samples. Conclusions: The relationship between PDs and the FFM was stable in two samples separated by a great cultural distance. These results suggest that a dimensional approach and in particular the FFM might be useful for describing PDs in a variety of cultural settings.