Title: Interactive administration service based on HMM speech recognition system

Authors: Mohamed Hamidi; Hassan Satori; Ouissam Zealouk; Khalid Satori; Naouar Laaidi

Addresses: Department of Arts and Media, Faculty of Languages, Letters and Arts, Ibn Tofail University, Kénitra, Morocco; LISAC, Department of Mathematics and Computer Science, FSDM, USMBA, Fez, Morocco ' LISAC, Department of Mathematics and Computer Science, FSDM, USMBA, Fez, Morocco ' LISAC, Department of Mathematics and Computer Science, FSDM, USMBA, Fez, Morocco ' LISAC, Department of Mathematics and Computer Science, FSDM, USMBA, Fez, Morocco ' LISAC, Department of Mathematics and Computer Science, FSDM, USMBA, Fez, Morocco

Abstract: The objective of this study is designing a secure interactive Amazigh speech system over the VoIP network. Our work explores the use of interactive voice response (IVR) system to manage the firewall and backup tasks where a voice platform was created based on HMM automatic speech recognition. To increase the security level the biological voiceprint is adopted for identification and management tasks. The speech system is trained to recognise the five words, 33 alphabets and ten first digits of Amazigh language. We performed experiences by varying IVR and ASR system parameters to manage a distant voice control administration system of the network. Our finding shows that the best performance for admin test is above 80% when the system was trained by using three HMMs and eight GMMs whereas the non-admin recognition rate is less than 5% which demonstrates the security aspect of our system.

Keywords: automatic speech recognition; interactive voice response; IVR; Asterisk server; Amazigh language; speech security; voice control; acoustic model; GMMs; HMMs; speech technology; administration service.

DOI: 10.1504/IJCAET.2022.120819

International Journal of Computer Aided Engineering and Technology, 2022 Vol.16 No.2, pp.266 - 282

Received: 02 Apr 2019
Accepted: 13 Jun 2019
Published online: 11 Feb 2022 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article