A Real-time Application Framework for Web-Based Speech Recognition Using HTTP/2 and SSE

Kalamullah Ramli (Departement of Electrical Engineering, Faculty of Engineering, Universitas Indonesia, Depok 16424, INDONESIA)
Asril Jarin (Center for Information and Communication Technology, BPPT Puspiptek Serpong, Tangerang Selatan 15314, INDONESIA)
Suryadi Suryadi (Department of Mathematics, Faculty of Mathematics and Natural Sciences, Universitas Indonesia Depok 16424, INDONESIA)

Indonesian Journal of Electrical Engineering and Computer Science
Vol 12, No 3: December 2018

The performance of network-based speech recognition application is mainly determined by the availability of all speech data received on the server and also the realtimeness in delivering the recognition results from the server. On the basis of full-duplex speech recognition application this paper proposes a real-time application framework for web speech recognition using HTTP/2 protocol and Sender-Sent Events (SSE). A number of experiments were performed to compare the latency of both the application using HTTP/2 plus SSE and the full-duplex application using WebSocket. The results showed that the proposed framework offers better alternative for a web-based speech recognition than the framework using WebSocket.