Vestavěný řídicí systém pro analýzu základních slov řečového signálu

Abstract

This Thesis examines the issue of word spotting in speech and related tasks such as choosing a mmicrophone type or the selection of AD converter. The components which make word spotting possible are described in this thesis. Within the framework of this thesis, a system of two dedicated devices was designed. One of the devices tries to detect words in speech and another reacts to detected words. The Snowboy library on Raspberry Pi 3 is responsible for detecting words. The System has been implemented and verified. The results of detection quality tests are located in the verification section.