This paper presents an end-to-end text-to-speech system with low latency on a CPU, suitable for real-time applications. The system is composed of an autoregressive attention-based sequence-to-sequence acoustic model and the LPCNet vocoder for waveform generation. An acoustic model architecture that adopts modules from both the Tacotron 1 and 2 models is proposed, while stability is ensured by using a recently proposed purely location-based attention mechanism, suitable for arbitrary sentence length generation. During inference, the decoder is unrolled and acoustic feature generation is performed in a streaming manner, allowing for a nearly constant latency which is independent from the sentence length. Experimental results show that the acoustic model can produce feature sequences with minimal latency about 31 times faster than real-time on a computer CPU and 6.5 times on a mobile CPU, enabling it to meet the conditions required for real-time applications on both devices. The full end-to-end system can generate almost natural quality speech, which is verified by listening tests.
Gene expression dictates fundamental cellular processes and its de-regulation leads to pathological conditions. A key contributor to the fine-tuning of gene expression is Dicer, an RNA-binding protein (RBPs) that forms complexes and affects transcription by acting at the post-transcriptional level via the targeting of mRNAs by Dicer-produced small non-coding RNAs. This review aims to present the contribution of Dicer protein in a wide spectrum of human pathological conditions, including cancer, neurological, autoimmune, reproductive and cardiovascular diseases, as well as viral infections. Germline mutations of Dicer have been linked to Dicer1 syndrome, a rare genetic disorder that predisposes to the development of both benign and malignant tumors, but the exact correlation of Dicer protein expression within the different cancer types is unclear, and there are contradictions in the data. Downregulation of Dicer is related to Geographic atrophy (GA), a severe eye-disease that is a leading cause of blindness in industrialized countries, as well as to psychiatric and neurological diseases such as depression and Parkinson’s disease, respectively. Both loss and upregulation of Dicer protein expression is implicated in severe autoimmune disorders, including psoriasis, ankylosing spondylitis, rheumatoid arthritis, multiple sclerosis and autoimmune thyroid diseases. Loss of Dicer contributes to cardiovascular diseases and causes defective germ cell differentiation and reproductive system abnormalities in both sexes. Dicer can also act as a strong antiviral with a crucial role in RNA-based antiviral immunity. In conclusion, Dicer is an essential enzyme for the maintenance of physiology due to its pivotal role in several cellular processes, and its loss or aberrant expression contributes to the development of severe human diseases. Further exploitation is required for the development of novel, more effective Dicer-based diagnostic and therapeutic strategies, with the goal of new clinical benefits and better quality of life for patients.
This paper presents a method for controlling the prosody at the phoneme level in an autoregressive attention-based text-to-speech system. Instead of learning latent prosodic features with a variational framework as is commonly done, we directly extract phoneme-level F0 and duration features from the speech data in the training set. Each prosodic feature is discretized using unsupervised clustering in order to produce a sequence of prosodic labels for each utterance. This sequence is used in parallel to the phoneme sequence in order to condition the decoder with the utilization of a prosodic encoder and a corresponding attention module. Experimental results show that the proposed method retains the high quality of generated speech, while allowing phoneme-level control of F0 and duration. By replacing the F0 cluster centroids with musical notes, the model can also provide control over the note and octave within the range of the speaker.
Purpose – Institutional repositories (IR) are usually used to archive and manage digital collections including research results, educational material, etc. Learning management systems (LMS) form a popular basis for e-learning and blended learning. This paper aims to study how to integrate IR and LMS to support accessibility of disabled students and students with learning difficulties (dyslexic students) in higher education. Customised ontologies focusing on disabled students can be used to facilitate indexing, and access of items in the repository. Design/methodology/approach – The authors propose a simple methodological approach to establish an integrating system for supporting accessibility. First, the authors review research works related to adaptive learning environments (ALEs) and blended learning, and discuss issues of the interoperability of IR and LMS. Then, based on the review, the authors discuss the use of an integrated ALE for supporting disabled students in the domain of higher technological education. The integrated system is based on IR, LMS and assistive and adaptive technology. The open source software platform DSpace is used to build up the repository applications Use of the web ontology language (OWL) ontologies is also proposed for indexing and accessing the various, heterogeneous items stored in the repository. Various open source LMS (e.g. openeclass) could be used to build up the integrated system. Finally, the authors describe experimentation with a prototype implemented to provide the mentioned capabilities. Findings – The technology is mature enough for building up integrated systems, combining capabilities of IR and LMS, for supporting disabled students. The use of ontologies focused on disabled students could facilitate the use of such integrated systems. Customisation and operation of a platform, for the selection and use of portions of OWL ontologies, could be based on the open source software Protégé. Such a platform forms a basis to create an appropriate ontology suitable for specific domains, e.g. the domain of technological education. Finally, the authors argue that the combined use of the OWL platform and the DSpace repository with open source LMS platforms could support domain experts for creating customised ontologies and facilitating searching. Originality/value – A new perception of the term integrated system for supporting disabled students in the higher education context is presented. This perception tries to combine the IR technology that supports the self-archiving approach of information, open LMS technology and the user-centred approach to support students and manage the “life of information”.
Regulation of gene expression has emerged as a fundamental element of transcript homeostasis. Key effectors in this process are the Argonautes (AGOs), highly specialized RNA-binding proteins (RBPs) that form complexes, such as the RNA-Induced Silencing Complex (RISC). AGOs dictate post-transcriptional gene-silencing by directly loading small RNAs and repressing their mRNA targets through small RNA-sequence complementarity. The four human highly-conserved family-members (AGO1, AGO2, AGO3, and AGO4) demonstrate multi-faceted and versatile roles in transcriptome’s stability, plasticity, and functionality. The post-translational modifications of AGOs in critical amino acid residues, the nucleotide polymorphisms and mutations, and the deregulation of expression and interactions are tightly associated with aberrant activities, which are observed in a wide spectrum of pathologies. Through constantly accumulating information, the AGOs’ fundamental engagement in multiple human diseases has recently emerged. The present review examines new insights into AGO-driven pathology and AGO-deregulation patterns in a variety of diseases such as in viral infections and propagations, autoimmune diseases, cancers, metabolic deficiencies, neuronal disorders, and human infertility. Altogether, AGO seems to be a crucial contributor to pathogenesis and its targeting may serve as a novel and powerful therapeutic tool for the successful management of diverse human diseases in the clinic.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.