Overview and testing of speech recognition systems

This article describes and compares the characteristics of such most common speech recognition systems as CMU Sphinx, Google Speech, HTK, Julius, iAtros, RWTH ASR. The comparison is based on the following important characteristics: supported recognition languages, supported operating systems, availability of documentation, open source code, presence or absence of offline mode, WER, WRR, SF indicators. It also describes the process and results of testing the best systems for various indicators.