Speech recognition technology is transformation skill using sound sensor such as microphone to transfer the acoustical signal to words or sentence. Speech conversation system using this technology and artificial intelligence is receiving attention as next generation of interface, and it is used in variable areas like smartphone, smart TV, car and so on. Recently, Samsung released ‘Bixby’ which is speech conversation program with artificial intelligence, and a lot of company such as Google and Naver are providing speech recognition open API. In this paper, we select three typical APIs and do comparison analysis of APIs’ features. In addition to that, we do three experiment in mobile for analysis of APIs’ speech recognition accuracy. First, we test number recognition. In second test, we test Korean word recognition. Lastly, we test sentence recognition with mobile instruction sentence. With result, we expect developers can select appropriate speech recognition open API in each situation.