Search

The American Medical Association recently declared homicides of transgender individuals an epidemic. However, transgender homicide victims are often classified as nontransgender. Our objective was to describe existing data and coding of trans (i.e., transgender) victims and to examine the risk factors for homicides of trans people relative to nontrans people across the United States.A retrospective review of the Centers for Disease Control and Prevention's National Violent Death Reporting System for the years 2003-2018 identified victims defined as transgender either through the "transgender" variable or narrative reports. Fisher's exact tests and logistic regression models were run to compare the demographics of trans victims to those not identified as trans.Of the 147 transgender victims identified, 14.4% were incorrectly coded as nontrans despite clear indication of trans status in the narrative description, and 6% were coded as hate crimes. Relative to nontrans victims, trans victims were more frequently Black (54.4% versus 40.7%, P = 0.001), had a mental health condition (26.5% versus 11.3%, P 0.001), or reported being a sex worker (9.5% versus 0.2%, P 0.001). There were disproportionately few homicides of transgender people in the South (13.6% of trans victims versus 29.1% of nontrans victims, P 0.001). Conversely, the West and Midwest accounted for a higher-than-expected proportion of trans victims relative to nontrans victims (23.1% of trans victims versus 16.2% of nontrans victims, P = 0.03; 24.5% of trans victims versus 16.8% of nontrans victims, P = 0.02, respectively).Though the murder of transgender individuals is a known public health crisis, inconsistencies still exist in the assessment and reporting of transgender status. Further, these individuals were more likely to have multiple distinct vulnerabilities. These findings provide important information for injury and violence prevention researchers to improve reporting of transgender status in the medical record and local trauma registries.

In this paper, we describe the “Spoken Web Search” Task, which is being held as part of the 2012 MediaEval campaign. The purpose of this task is to perform audio search in multiple languages, with very little resources being available for each individual language. The data is being taken from audio content that was created in live and realistic low-resource settings. The organizers would like to thank Martha Larson for organizing this event [3], and the participants for putting in a lot of hard work into submitting their systems. The “African” data [5] has kindly been collected by CSIR and made available by NWU.

Spoken recordings that have been transcribed for human reading (e.g. as captions for audiovisual material, or to provide alternative modes of access to recordings) are widely available in many languages. Such recordings and transcriptions have proven to be a valuable source of ASR data in well-resourced languages, but have not been exploited to a significant extent in under-resourced languages or dialects. Techniques used to harvest such data typically assume the availability of a fairly accurate ASR system, which is generally not available when working with resourcescarce languages. In this work, we define a process whereby an ASR corpus is bootstrapped using unmatched ASR models in conjunction with speech and approximate transcriptions sourced from the Internet. We introduce a new segmentation technique based on the use of a phone-internal garbage model, and demonstrate how this technique (combined with limited filtering) can be used to develop a large, high-quality corpus in an underresourced dialect with minimal effort. We would like to thank Bryan McAlister and his team, who performed the initial data collection and processing; and the South African Centre for High Performance Computing (CHPC), who made their facilities available for these experiments.

Building transcribed speech corpora for under-resourced languages plays a pivotal role in developing speech technologies for such languages. We have developed an open-source tool for devices running the Android operating system to facilitate the efficient collection of speech data for Automatic Speech Recognition system development. The tool was designed for use in typical developing-world conditions; we present the relevant design choices and analyse the effectiveness of this tool by means of a case study. In particular, we introduce a novel semi-real-time quality monitoring system, which increases the efficiency of the data collection process. This project was made possible through the support of the South African National Centre for Human Language Technology, an initiative of the South African Department of Arts and Culture. The authors would also like to thank Pedro Moreno, Thad Hughes and Ravindran Rajakumar of Google Research for valuable inputs at various stages of this work.

In the Andean region of Patagonia, Argentina, in the provinces of Neuquén, Río Negro and Chubut, there are numerous Douglas Fir [Pseudotsuga menziesii (Mirb) Franco] plantations, which constitute one of the timber species most widely used in the region for commercial plantations. In this work, we studied Douglas Fir wood basic density and its variation within trees and among trees. Also, the relationship between wood density, ring-with and the proportion of latewood was examined by regression analysis, determining at the some time wood density of juvenile and mature wood. We found that mean wood density of Douglas Fir was 0,366 g/cm3, which is one of the lowest wood density values reported for this species elsewhere. Results show that rings-width and latewood proportion were significantly related to specific gravity. We also found that Douglas fir trees begun to produce mature wood at around 17 years of age., En la región andina de las provincias de Neuquén, Río Negro y Chubut de Argentina existen numerosas plantaciones de pino Oregón [Pseudotsuga menziesii (Mirb.) Franco], constituyendo una de las especies más utilizadas en plantaciones comerciales. En el presente trabajo se estudió la densidad básica de la madera de esta especie y su variación entre árboles y dentro de un mismo individuo, determinándose su relación con el ancho de anillos y la proporción leño temprano por medio de análisis de regresión. Paralelamente se determinó la densidad correspondiente a madera juvenil y madura. La densidad básica media para la población analizada fue de 0,366 g/cm3, valor que se encuentra entre los mínimos citados para diferentes lugares del mundo en los cuales se cultiva la especie. La densidad básica se correlacionó significativamente con el ancho de anillos y la proporción de madera tardía. Se encontró que la edad a partir de la cual se forma madera madura es 17 años.

The objective of this study was to evaluate different growth zones of Douglas-fir [Pseudotzuga menziesii (Mirb.) Franco] in the Argentinean Patagonian Andes and to characterize them based on their potential productivity. Three preliminary zones were defined based on their precipitation regimes, since this is the most important climatic factor in the region. These zones lied within the isohyets that goes from 600 to 1,000 mm of annual precipitation, other from 1,000 to 1,400 mm, and the third one from 1,400 mm and up. For each zone and for the whole region, site and top height functions were adjusted by using stochastic differential equations. The database was obtained from stem analysis of 126 trees. These trees were selected from 71 temporal plots established on 64 stands. Height growth and site index functions were fit for the whole region and for each growth zone. The height growth was modelled by a stochastic differential equation in wich the deterministic part is the Richards model. All parameters are estimated by a maximum likelihood procedure. The polymorphyc model with the free origin was the best. The evaluation of the models defined for each zone and their comparison with the use of a regional model was made through graphic and statistical analyses (likelihood ratio test). The results showed that the regional model was not satisfactory to predict productivity of Douglas fir for the whole Patagonian region. Therefore, the growth zones and the functions of site and dominant height growth which have been presented are recommended as the better predictors of the potential productivity of Douglas-fir plantations growing in Patagonia., El objetivo de este estudio fue evaluar la existencia de diferentes zonas de crecimiento para pino Oregón [Pseudotzuga menziesii (Mirb.) Franco] en la Patagonia Andina Argentina y caracterizar las mismas en base a su productividad. Tres zonas preliminares fueron definidas en base a rangos de precipitación media anual, quedando delimitadas por las isoyetas que van de 600 a 1.000 mm, de 1.000 a 1.400 mm y de 1.400 mm a más. La base de datos utilizada fue obtenida del análisis de tronco de 126 árboles seleccionados de 71 parcelas temporales establecidas en 64 rodales de toda la región. Se ajustaron funciones de índice de sitio y crecimiento en altura para la región completa y para cada una de las zonas de crecimiento predefinidas. El crecimiento en altura fue modelado por una ecuación diferencial estocástica donde el componente determinístico es el modelo de Chapman-Richards. Todos los parámetros fueron estimados por máxima verosimilitud. El modelo polimórfico con el origen libre fue el de mejor comportamiento. La evaluación de los modelos definidos para cada zona y su comparación con el uso de un modelo regional fue hecha a través de análisis gráfico y estadístico, empleando en particular el test de razón de verosimilitud. Los resultados mostraron que el modelo regional no fue satisfactorio, por ello se presentan las zonas de crecimiento definidas y las funciones de índice de sitio y altura dominante obtenidas por zona.

Different taper equations for Pseudotsuga menziesii (Mirb.) Franco plantations were evaluated in the Argentina Patagonian Andes. Functions fitted by Bruce et al. (1968), Real and Moore (1986), Kozak (1988) and Pérez et al. (1990) were tested. Data were collected from 268 trees scattered throughout 68 stands. The models were compared based on their bias and error when predicting volume, diameter and height. The validation was performed on a separate set of data. The best models were those developed by Kozak (1988) and Bruce et al. (1968)., Se evalúan distintos modelos fustales para ser utilizados en Pseudotsuga menziesii (Mirb.) Franco en la Patagonia Andina Argentina. Los modelos evaluados corresponden a los propuestos por Bruce et al. (1968), Real y Moore (1986), Kozak (1988) y Pérez et al. (1990), utilizando para el ajuste información de 268 árboles provenientes de 68 rodales distribuidos en el área geográfica donde se foresta con esta especie en el país. La validación de la estimación de volúmenes, diámetros y alturas, se realizó utilizando medidas de error y sesgo, empleando una base de datos independiente a la utilizada en el ajuste de cada uno de los modelos. Se analizó el comportamiento de cada uno de ellos para el total del fuste y para diferentes porciones del mismo. Finalmente, la evaluación realizada indica que los mejores modelos correspondieron a los de Kozak (1988) y Bruce et al. (1968).

Different taper equations for Pseudotsuga menziesii (Mirb.) Franco plantations were evaluated in the Argentina Patagonian Andes. Functions fitted by Bruce et al. (1968), Real and Moore (1986), Kozak (1988) and Pérez et al. (1990) were tested. Data were collected from 268 trees scattered throughout 68 stands. The models were compared based on their bias and error when predicting volume, diameter and height. The validation was performed on a separate set of data. The best models were those developed by Kozak (1988) and Bruce et al. (1968). Se evalúan distintos modelos fustales para ser utilizados en Pseudotsuga menziesii (Mirb.) Franco en la Patagonia Andina Argentina. Los modelos evaluados corresponden a los propuestos por Bruce et al. (1968), Real y Moore (1986), Kozak (1988) y Pérez et al. (1990), utilizando para el ajuste información de 268 árboles provenientes de 68 rodales distribuidos en el área geográfica donde se foresta con esta especie en el país. La validación de la estimación de volúmenes, diámetros y alturas, se realizó utilizando medidas de error y sesgo, empleando una base de datos independiente a la utilizada en el ajuste de cada uno de los modelos. Se analizó el comportamiento de cada uno de ellos para el total del fuste y para diferentes porciones del mismo. Finalmente, la evaluación realizada indica que los mejores modelos correspondieron a los de Kozak (1988) y Bruce et al. (1968).

Aiming at both speaker independence and robustness with respect to recognition errors in the spoken queries, we have implemented a two-pass system for spoken web search. In the first pass, unconstrained phone recognition of both the query terms and the content audio is employed to represent these recordings as phone strings. A dynamic-programming approach then finds regions in the content phone strings that correspond closely to one or more query strings. In the sec- ond pass, each of these regions is again processed with a phone recognizer, but now a lattice is extracted; this lattice is com- pared against similar lattices extracted for each of the queries. We find our approach to be somewhat successful in identify- ing the query terms in both the development and evaluation sets, but not to generalize well between these sets. Multilingual Speech Technologies, North-West University, Vanderbijlpark, South Africa HLT Research Group,CSIR Meraka Institute, Pretoria, South Africa Microsoft Research Lab India, Bangalore, India

Text-based language identification (T-LID) of isolated words has been shown to be useful for various speech processing tasks, including pronunciation modelling and data categorisation. When the words to be categorised are proper names, the task becomes more difficult: not only do proper names often have idiosyncratic spellings, they are also often considered to be multilingual. We, therefore, investigate how an existing T-LID technique can be adapted to perform multilingual word classification. That is, given a proper name, which may be either mono- or multilingual, we aim to determine how accurately we can predict how many possible source languages the word has, and what they are. Using a Joint Sequence Model-based approach to T-LID and the SADE corpus - a newly developed proper names corpus of South African names - we experiment with different approaches to multilingual T-LID. We compare posterior-based and likelihood-based methods and obtain promising results on a challenging task.

Pronunciation lexicons can range from fully graphemic (modeling each word using the orthography directly) to fully phonemic (first mapping each word to a phoneme string). Between these two options lies a continuum of modeling options. We analyze techniques that can improve the accuracy of a graphemic system without requiring significant effort to design or implement. The analysis is performed in the context of the IARPA Babel project, which aims to develop spoken term detection systems for previously unseen languages rapidly, and with minimal human effort. We consider techniques related to letter-to-sound mapping and language-independent syllabification of primarily graphemic systems, and discuss results obtained for six languages: Cebuano, Kazakh, Kurmanji Kurdish, Lithuanian, Telugu and Tok Pisin. This work was supported by the Intelligence Advanced Research Projects Activity (IARPA) via Department of Defense U.S. Army Research Laboratory contract number W911NF-12- C-0013. The U.S. Government is authorized to reproduce and distribute reprints for Governmental purposes notwithstanding any copyright annotation thereon. Disclaimer: The views and conclusions contained herein are those of the authors and should not be interpreted as necessarily representing the official policies or endorsements, either express or implied, of IARPA, DoD/ARL, or the U.S. Government.