Wireless Multimedia Sensor Networks (WMSNs) are composed of small devices that are able to capture video or audio information and to transmit it over wireless channels. The development of wireless technologies, such as: WiFi, Bluetooth, Zigbee, and UWB, encourages the emergence of heterogeneous networks. However, only a few existing solutions take into account the constraints of multi-tier and multi-MAC wireless sensor networks. In this paper, we propose a cost function coupled to a new generic (i.e. independent from any MAC protocol) cross-layer routing protocol adapted to multimedia traffic over hierarchical and heterogeneous networks. The goal of our proposed protocol is to dynamically assess the routing process cost and the requirement of the multimedia application in order to ensure a sufficient quality of service (Soft QoS). Furthermore, the cross-layer approach is needed to use the physical and MAC parameters on the routing operation. Simulation results show that our solution is efficient and gives better results than classical protocols.