This paper proposes a semantic-aware resource allocation (SARA) framework with flexible duty cycle (DC) coexistence mechanism (SARADC) for 5G-V2X heterogeneous networks (HetNets) based on deep reinforcement learning (DRL) proximal policy optimization (PPO). The framework addresses the challenges of high-speed vehicular networking in urban environments by designing a semantic communication system and introducing two resource allocation metrics: high-speed semantic transmission rate (HSR) and semantic spectrum efficiency (HSSE). The goal is to maximize HSSE while addressing the coexistence of vehicular and WiFi users in 5G New Radio Unlicensed (NR-U) networks. The proposed approach integrates semantic communication into the communication system, optimizing flexible DC coexistence and resource allocation. Experimental results show that the proposed solution outperforms traditional bit transmission methods in terms of HSSE and semantic throughput (ST) for both vehicular and WiFi users. The system model focuses on V2X communication in a high-speed urban environment with N vehicles and W WiFi APs. Each vehicle moves at a constant speed, resulting in N V2I links. The system includes two tiers of base stations: macro base stations (MaBs) operating in licensed bands and micro base stations (MiBs) in unlicensed bands. Vehicles and WiFi APs use DeepSC models to encode and decode semantic information. The proposed SARADC algorithm uses PPO DRL to optimize flexible DC and resource allocation based on semantic awareness to maximize semantic throughput (ST). The algorithm involves a state representation including channel gain, SINR, HSSE, and interference. Actions include BS and RB allocation, transmission power, DC proportion, and semantic symbols per word. The reward function evaluates the performance of the algorithm based on ST and HSSE. The SARADC algorithm outperforms other baselines in terms of HSSE and ST, demonstrating the effectiveness of the proposed scheme and semantic communication. The results show that the proposed algorithm achieves higher HSSE with meaningful semantic data, provides better performance with flexible DC, and shows advantages when using semantic information transmission for data mapping below 8 bits per word.This paper proposes a semantic-aware resource allocation (SARA) framework with flexible duty cycle (DC) coexistence mechanism (SARADC) for 5G-V2X heterogeneous networks (HetNets) based on deep reinforcement learning (DRL) proximal policy optimization (PPO). The framework addresses the challenges of high-speed vehicular networking in urban environments by designing a semantic communication system and introducing two resource allocation metrics: high-speed semantic transmission rate (HSR) and semantic spectrum efficiency (HSSE). The goal is to maximize HSSE while addressing the coexistence of vehicular and WiFi users in 5G New Radio Unlicensed (NR-U) networks. The proposed approach integrates semantic communication into the communication system, optimizing flexible DC coexistence and resource allocation. Experimental results show that the proposed solution outperforms traditional bit transmission methods in terms of HSSE and semantic throughput (ST) for both vehicular and WiFi users. The system model focuses on V2X communication in a high-speed urban environment with N vehicles and W WiFi APs. Each vehicle moves at a constant speed, resulting in N V2I links. The system includes two tiers of base stations: macro base stations (MaBs) operating in licensed bands and micro base stations (MiBs) in unlicensed bands. Vehicles and WiFi APs use DeepSC models to encode and decode semantic information. The proposed SARADC algorithm uses PPO DRL to optimize flexible DC and resource allocation based on semantic awareness to maximize semantic throughput (ST). The algorithm involves a state representation including channel gain, SINR, HSSE, and interference. Actions include BS and RB allocation, transmission power, DC proportion, and semantic symbols per word. The reward function evaluates the performance of the algorithm based on ST and HSSE. The SARADC algorithm outperforms other baselines in terms of HSSE and ST, demonstrating the effectiveness of the proposed scheme and semantic communication. The results show that the proposed algorithm achieves higher HSSE with meaningful semantic data, provides better performance with flexible DC, and shows advantages when using semantic information transmission for data mapping below 8 bits per word.