The increasing exploration of smart space voice assistants and audio monitoring IoT devices leads to significant trust and privacy concerns. Among them, some of the data are sensitive and a user may not aware of when and where audio recordings are sent for processing. This provides burning questions- how do users gain knowledge of, and control the amount of data captured in the physical world. Moreover, the sending and processing of these sensitive data to the far cloud do not provide an efficient solution for sensitive data which causes serious privacy and security concerns. In this paper, a new architecture for providing audio transparency and privacy in the IoT-rich environment is proposed. The proposed privacy enforcement module is used to operate within a nearby fog node, which situated close to the data sources and enforces privacy according to the data owner's desire. To show the effectiveness of the proposed architecture, a well-known privacy threat framework is investigated.