Redirigiendo al acceso original de articulo en 19 segundos...
Inicio  /  Future Internet  /  Vol: 15 Par: 5 (2023)  /  Artículo
ARTÍCULO
TITULO

Domain Adaptation Speech-to-Text for Low-Resource European Portuguese Using Deep Learning

Eduardo Medeiros    
Leonel Corado    
Luís Rato    
Paulo Quaresma and Pedro Salgueiro    

Resumen

Automatic speech recognition (ASR), commonly known as speech-to-text, is the process of transcribing audio recordings into text, i.e., transforming speech into the respective sequence of words. This paper presents a deep learning ASR system optimization and evaluation for the European Portuguese language. We present a pipeline composed of several stages for data acquisition, analysis, pre-processing, model creation, and evaluation. A transfer learning approach is proposed considering an English language-optimized model as starting point; a target composed of European Portuguese; and the contribution to the transfer process by a source from a different domain consisting of a multiple-variant Portuguese language dataset, essentially composed of Brazilian Portuguese. A domain adaptation was investigated between European Portuguese and mixed (mostly Brazilian) Portuguese. The proposed optimization evaluation used the NVIDIA NeMo framework implementing the QuartzNet15×5 architecture based on 1D time-channel separable convolutions. Following this transfer learning data-centric approach, the model was optimized, achieving a state-of-the-art word error rate (WER) of 0.0503.

 Artículos similares

       
 
Mohammad Fikry Abdullah, Zurina Zainol, Siaw Yin Thian, Noor Hisham Ab Ghani, Azman Mat Jusoh, Mohd Zaki Mat Amin and Nur Aiza Mohamad    
The impact of Big Data (BD) creates challenges in selecting relevant and significant data to be used as criteria to facilitate flood management plans. Studies on macro domain criteria expand the criteria selection, which is important for assessment in al... ver más

 
Nicholus Mboga, Stefano D?Aronco, Tais Grippa, Charlotte Pelletier, Stefanos Georganos, Sabine Vanhuysse, Eléonore Wolff, Benoît Smets, Olivier Dewitte, Moritz Lennert and Jan Dirk Wegner    
Multitemporal environmental and urban studies are essential to guide policy making to ultimately improve human wellbeing in the Global South. Land-cover products derived from historical aerial orthomosaics acquired decades ago can provide important evide... ver más

 
Michele Bonanni, Francesco Chiti, Romano Fantacci and Laura Pierucci    
Software Defined Networking (SDN) provides a new perspective for the Internet of Things (IoT), since, with the separation of the control from the data planes, it is viable to optimise the traditional networks operation management. In particular, the SDN ... ver más
Revista: Future Internet

 
Giuseppe Pulighe, Flavio Lupia, Huajin Chen and Hailong Yin    
The consequences of climate change on food security in arid and semi-arid regions can be serious. Understanding climate change impacts on water balance is critical to assess future crop performance and develop sustainable adaptation strategies. This pape... ver más
Revista: Hydrology

 
Christos Makris and Michael Angelos Simos    
Semantic representation of unstructured text is crucial in modern artificial intelligence and information retrieval applications. The semantic information extraction process from an unstructured text fragment to a corresponding representation from a conc... ver más