Speech Recognition for Mobile Linux Distrubitions in the Case of Aurora OS

Abstract

Aurora OS is a POSIX-compatible mobile Linux distribution designed for enterprise and business purposes. The main application area of devices based on Aurora OS is corporations in Russia. For example, such devices are used by field workers to receive tasks and send reports. The reports are generated by filling out complex forms with a large number of input fields and options to choose from. Speech recognition allows to significantly speed up this process. Aurora OS has no built-in tools to implement such functions, so developers of software for field workers need to use a third-party solution. There are several voice recognition engines for POSIX-compatible systems, but only those that support the Russian language, can work locally on devices, and have a free license are suitable for the task. The only solution for the task that meets the criteria is the Kaldi engine. However, it depends on libraries implemented using Fortran, and Aurora OS does not support this programming language. Therefore, it was necessary to develop a way to use software implemented in Fortran in an environment that does not have a native support for it. This paper proposes an approach for solving such a problem, which can be applied to all similar cases.

Keywords

• asr
• aurora os
• f2c
• fortran
• kaldi
• linux
• posix
• sailfish os
• speach recognition
• tts