Two mechanisms to deploy real-time voice transmission over internet of things (VoIoTs)‎

Abstract EN

Internet of things (IoTs) is an example of the last advances in information and communication technologies.

in particular, with the development of embedded low-cost, low-power WSN wireless sensor network (WSN) technologies, researchers began to take benefits of integrating this technology into a wide range of IoTs applications.

an important application is real-time voice transmission over IoTs (VoIoTs).

this paper presents two mechanisms to deploy real-time VoIoTs: voice streaming and automatic speech recognition (asr) systems.

the two mechanisms are implemented and evaluated by using a framework composed of a raspberry pi3 (rpi3), open source ffmpeg technology for streaming voice and open source pocketsphinx package for ASR system.

the performance of proposed voice streaming system is evaluated by studying its behavior utilizing three audio encoding algorithms: ac3, mp3 and opus with different sampling rates and evaluation metrics: one-way delay, jitter, bandwidth (B.W), CPU usage and packet losses.

while the performance of proposed ASR system is evaluated by a set of metrics such as : decoding time, speed, word error rate (WER) and accuracy with full and optimized language model.