This paper proposes a semantic-aware spectrum sharing algorithm (SSS) based on deep reinforcement learning (DRL) for high-speed mobile Internet of Vehicles (IoV) environments. The goal is to address spectrum scarcity and improve communication performance by leveraging semantic information in vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communications. The SSS algorithm introduces new metrics, high-speed semantic spectrum efficiency (HSSE) and semantic transmission rate (HSR), to evaluate the effectiveness of semantic communication. It employs the soft actor-critic (SAC) DRL approach to optimize spectrum sharing strategies, transmission power, and semantic symbol length, aiming to maximize HSSE in V2I and enhance the success rate of semantic information transmission (SRS) in V2V. The algorithm is designed to handle the dynamic and heterogeneous nature of IoV environments, where vehicles may enter or leave communication ranges, leading to potential conflicts and reduced performance. Experimental results show that the SSS algorithm outperforms traditional spectrum sharing methods and other reinforcement learning approaches, achieving a 15% increase in HSSE and approximately a 7% increase in SRS. The SSS algorithm is implemented in a system model where vehicles communicate with a base station (BS) and other vehicles, with semantic information extracted and transmitted using a deep semantic communication (DeepSC) model. The algorithm optimizes spectrum sharing by considering channel conditions, interference, and semantic similarity, using a Markov decision process (MDP) and SAC for decision-making. The SSS algorithm is evaluated through simulations, demonstrating its effectiveness in improving communication performance in high-speed mobile IoV environments.This paper proposes a semantic-aware spectrum sharing algorithm (SSS) based on deep reinforcement learning (DRL) for high-speed mobile Internet of Vehicles (IoV) environments. The goal is to address spectrum scarcity and improve communication performance by leveraging semantic information in vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communications. The SSS algorithm introduces new metrics, high-speed semantic spectrum efficiency (HSSE) and semantic transmission rate (HSR), to evaluate the effectiveness of semantic communication. It employs the soft actor-critic (SAC) DRL approach to optimize spectrum sharing strategies, transmission power, and semantic symbol length, aiming to maximize HSSE in V2I and enhance the success rate of semantic information transmission (SRS) in V2V. The algorithm is designed to handle the dynamic and heterogeneous nature of IoV environments, where vehicles may enter or leave communication ranges, leading to potential conflicts and reduced performance. Experimental results show that the SSS algorithm outperforms traditional spectrum sharing methods and other reinforcement learning approaches, achieving a 15% increase in HSSE and approximately a 7% increase in SRS. The SSS algorithm is implemented in a system model where vehicles communicate with a base station (BS) and other vehicles, with semantic information extracted and transmitted using a deep semantic communication (DeepSC) model. The algorithm optimizes spectrum sharing by considering channel conditions, interference, and semantic similarity, using a Markov decision process (MDP) and SAC for decision-making. The SSS algorithm is evaluated through simulations, demonstrating its effectiveness in improving communication performance in high-speed mobile IoV environments.