Project Description

NEWTON Technologies wants to make video/audio content accessible to deaf people. The app is involved in Automatic Speech Recognition not only in the Czech language, but also in other Slavic languages. The aim of the project is to create a functional prototype of an online application in which audio/video content can be subtitled; from a recording, as well as in real time. Moreover, the solution will allow easy and quick editing of automatically transcribed text and making these corrected texts available to other members of the community. For our hackathon they will be focusing on using this technology within Retail industry both online and offline. For the Automatic Speech Recognition they use a special tool developed by the Technical University of Liberec. For training the acoustic and language models, machine learning algorithms are used (Neural Networks). In future, they plan to use AI (in the area of NLP) in the so-called postprocessing, where they automatically edit and format recognized text according to specific needs of the target application. At the hackathon, they want to consult technological details, such as integration into Azure and use of Microsoft AI tools.