Vocapia Research (France)

Stevia rebaudiana Bertoni is a perennial cultivated plant that produces steviol glycosides, which are natural acaloric sweet compounds in its leaves. Many pathogens can infect this species. Among these pathogens, Septoria steviae, the causal agent of Septoria leaf spot, is the most widely described on S. rebaudiana. The objective of this study is to validate a leaf disk screening method for S. rebaudiana response to S. steviae infection on 10 S. rebaudiana genotypes with contrasting responses to the fungus. Three independent experiments were carried out in which 40 leaf discs per genotype were infected with an inoculum at a concentration of 10 5 conidia/ml or with water. The response of the 10 genotypes was scored through image analysis. Image analysis was conducted at 9, 12, 15, 18 and 21 days after inoculation (DAI), allowing the percentage of leaf symptomatic area (PLSA) and the area under the disease progress curve (AUDPC) to be quantified. The evaluation of symptom speed development was realized through a linear regression between PLSA and time for each tested genotype. Genotypes exhibited the lowest susceptibility at 21 DAI, with Cult102_SPA and Cult76_GER displaying 3% of the surface symptomatic. Genotypes with the highest susceptibility were Cult33_FRA and Cult75_GER, with 24% severity. These results correlate with visual scoring performed at 21 DAI and with previous inoculation protocols on S. rebaudiana cuttings by maintaining the ranking of the different genotypes regardless of the test. The development of a leaf-disk inoculation method offers new avenues for breeding strategies to develop tolerant or resistant S. rebaudiana cultivars. [Formula: see text] Copyright © 2025 The Author(s). This is an open access article distributed under the CC BY 4.0 International license .

This paper investigates the potential of improving a hybrid automatic speech recognition model trained on 10 hours of transcribed data with 200 hours of untranscribed data in lowresource languages. First, we compare baseline methods of cross-lingual transfer with MFCC features and features extracted with the multilingual self-supervised model XLSR-53. Subsequently, we compare two approaches that can leverage the untranscribed data: semi-supervised training with LF-MMI and continued self-supervised pre-training of XLSR-53. Our results on well-resourced English broadcast data derived from MGB show that both methods achieve 18% and 27% relative improvements compared to the baseline, respectively. On the low-resource South African Soap Opera dataset, the relative improvement with semi-supervised training is only 3% due to the inherently weak language model. However, continued pretraining achieves 8.6% relative improvement because it does not rely on any external information.

Speech recognition for low-resource languages remains challenging and can be addressed with techniques such as multilingual modeling and transfer learning. In this work, we explore several solutions to the multilingual training problem: training monolingual models with multilingual features, adapting a multilingual model with transfer learning and using language embeddings as additional features. To develop practical solutions we focus our work on medium size hybrid ASR models. The multilingual models are trained on 270 hours of iARPA Babel data from 25 languages, and results are reported on 4 Babel languages for the Limited Language Pack (LLP) condition. The results show that adapting a multilingual acoustic model with language embeddings is an effective solution, outperforming the baseline monolingual models, and providing comparable results to models based on state-of-the-art XLSR-53 features but with the advantage of needing 15 times fewer parameters.

The realization of language through vocal sounds involves a complex interplay between the lungs, the vocal cords, and a series of resonant chambers (e.g.mouth and nasal cavities).Due to their connection to the outside world, these body parts are popular spots for viruses and bacteria to enter the human organism.Affected people may suffer from an upper respiratory tract infection (URTIC) and consequently their voice often sounds breathy, raspy or sniffly.In this paper, we investigate the audible effects of a cold on a phonetic level.Results on a German corpus show that the articulation of consonants is more impaired than that of vowels.Surprisingly, nasal sounds do not follow this trend in our experiments.We finally try to predict a speaker's health condition by fusing decisions we derive from single phonemes.The presented work is part of the INTER-SPEECH 2017 Computational Paralinguistics Challenge.

We present work in progress on an intelligent embodied conversation agent in the basic care and healthcare domain. In contrast to most of the existing agents, the presented agent is aimed to have linguistic cultural, social and emotional competence needed to interact with elderly and migrants. It is composed of an ontology-based and reasoning-driven dialogue manager, multimodal communication analysis and generation modules and a search engine for the retrieval of multimedia background content from the web needed for conducting a conversation on a given topic.

In this paper we investigate various techniques in order to build effective speech to text (STT) and keyword search (KWS) systems for low resource conversational speech. Subword decoding and graphemic mappings were assessed in order to detect out-of-vocabulary keywords. To deal with the limited amount of transcribed data, semi-supervised training and data selection methods were investigated. Robust acoustic features produced via data augmentation were evaluated for acoustic modeling. For language modeling, automatically retrieved conversational-like Webdata was used, as well as neural network based models. We report STT improvements with all the techniques, but interestingly only some improve KWS performance. Results are reported for the Swahili language in the context of the 2015 OpenKWS Evaluation.

Adverse events can be associated with treating critically ill patients with immunoglobulin (Ig)G. Some adverse events are due to contaminants like IgA and activated Factor (F)XI. Therefore, new purification strategies are needed for dedicated removal of these contaminants without impairing IgG recovery. An immunoglobulin fraction containing IgG, IgM, and IgA was prepared by caprylic acid precipitation of cryoprecipitate-poor plasma. The capacities of the cation exchangers (S HyperCel and CM Ceramic HyperD F) and anion exchangers (HyperCel STAR AX and Q HyperCel) to remove IgA, IgM, and spiked FXI were tested following a design of experiment approach using microplates and chromatographic column scale-up. FXI removal was also evaluated using Mustang S chromatographic membranes. IgG/IgG subclasses, IgA, IgM, and FXI were assessed by enzyme-linked immunosorbent assay, and caprylic acid, by gas chromatography. Extensive removal of IgA and IgM, but not FXI, was achieved by a two-step chromatographic process combining S HyperCel used in the IgG binding and elution mode and HyperCel STAR AX used in the IgG flow-through mode, providing high IgG and IgG subclass recovery (>85%), high purity (>99.5%), and efficient removal of IgA (<0.5%) and IgM (undetectable). Twenty-six-fold FXI removal was achieved by processing the resulting purified IgG fraction through Mustang S cation-exchanger membranes at pH 6.0 and 12.7 mS/cm. Caprylic acid was removed by S HyperCel. Combining S HyperCel and HyperCel STAR AX extensively removed IgA and IgM, with good IgG recovery. Mustang® S membranes can be used for dedicated removal of FXI.

Abstract Purpose Recombinant Adeno‐Associated virus (rAVV) encoding meganucleases specifically designed to address Herpes simplex virus type 1 (HSV1) reduce HSV1 replication both in vitro and ex‐vivo. The main issue of this potential antiviral treatment is the delivery system. To evaluate a vector‐free method, we tested an electroporation process to transfer HSV1‐specific meganucleases (HSV1‐SM) to the murine corneaMethods Corneal gene transfer was achieved by subconjunctival injection (SCI) of recombinant plasmid encoding either GFP (group 1; n=30) or HSV1‐SM (group 2; n=12), followed by electroporation. Mice were sacrified at day (D)1, 3, 7, 13, 35 (group 1) and at D13 and 35 (group 2). Control group consisted in SCI w/o electroporation. Corneal samples were analyzed for the expression of transcripts encoding GFP and HSV1‐SM using real time PCR. The results were compared to those obtained following SCI of rAAV encoding either GFP or HSV1‐SMResults For HSV1‐SM, SCI plus electroporation induced a stronger expression than SCI alone at D13. In group 1, the expression of GFP transcripts decreased from D1 to D35. However, until D13, their levels of expression were higher than those obtained with the rAAV method (p&lt;0.002). Similarly, the expression of HSV1‐SM transcripts at D13 was higher than those obtained with the rAAV method (p&lt;0.002), suggesting that therapeutic concentration could be optimised with this method. At D35, results were not statistically different between the 2 methodsConclusion In our hands, with the method we used, we can conclude that SCI of plasmid encoding HSV1‐SM with immediate electroporation is an efficient method to improve delivery in short‐term but not in long‐term when compared to rAAV