DL for sound/audio recognition in IoT

It is important to understand the working principle of an Automatic Speech Recognition (ASR) system before discussing the useful DL models.

ASR system model

An Automatic Speech Recognition (ASR) system needs three main sources of knowledge. These sources are known as an acoustic model, a phonetic lexicon, and a language model [4]. Generally, an acoustic model deals with the sounds of language, including the phonemes and extra sounds (such as pauses, breathing, background noise, and so on). On the other hand, a phonetic lexicon model or dictionary includes the words that can be understood by the system, with their possible pronunciations. Finally, a language model includes knowledge about...