Cognitive Radio Networks (CRNs) are networks used to address the spectrum scarcity issue in wireless communication systems. The recent rise of Machine learning (ML) and Deep learning (DL) has been made possible by utilizing frequencies more effectively. Dynamic access to frequencies is made possible by these methods, which protect other users from damage while making use of unutilized spectrum resources. In this study, an Enhanced Reinforcement Learning Algorithm (ERLA) is presented as a new CR access strategy to enable Dynamic Spectrum Allocation (DSA) using Confidence Human Agent Transfer (CHAT). The two main steps are finding the optimal channel based on bandwidth and arranging agents (users) access to the spectrum in a certain sequence. The ERLA yields better outcome of 99.93% of Spectrum Utilization (SU) compared to Artificial Neuron Network (ANN) and Deep Multi-user RL (DMRL), thus enhancing wireless communication for all CR terminals. In addition to maintaining an ideal channel allocation strategy, it assures great user satisfaction and effective spectrum utilization.