Task-Oriented Semantic Communication (TOSC) has been considered as a new communication paradigm to serve various samrt devices that depend on Artificial Intelligence (AI) tasks in future wireless networks. The existing TOSC frameworks rely on the Neural Network (NN) model to extract the semantic feature from the source data. The semantic feature, constituted by analog vectors of a lower dimensionality relative to the original source data, reserves the meaning of the source data. By conveying the semantic feature, TOSCs can significantly reduce the amount of data transmission while ensuring the correct execution of the AI-driven downstream task. However, standardized wireless networks depend on digital signal processing for data transmission, yet they necessitate the conveyance of semantic features that are inherently analog. Although existing TOSC frameworks developed the Deep Learning (DL) based \emph{analog approach} or conventional \emph{digital approach} to transmit the semantic feature, but there are still many challenging problems to urgently be solved in actual deployment. In this article, we first propose several challenging issues associated with the development of the TOSC framework in the standardized wireless network. Then, we develop a Digital-Analog transmission framework based TOSC (DA-TOSC) to resolve these challenging issues. Future research directions are discussed to further improve the DA-TOSC.