The FCC's ruling for white-space (WS) spectrum utilization has raised the popularity of database-assisted dynamic spectrum access (DSA). However, how to dynamically share the WS spectrum overcoming the spectrum property of time-frequency variance while considering the quality of service (QoS) requirements of secondary users (SUs) still remains challenging. To this end, we propose a dynamic spectrum allocation scheme for database-assisted DSA. Firstly, we model the spectrum requirements for diverse services considering the minimum required service data-rate and spectrum access duration. More importantly, the spectrum demand evaluation and bidding policy are formulated based on the service classes of SUs. Secondly, we design a double-phase (DP) spectrum allocation scheme, including the initial resource allocation phase and resource allocation adjustment phase. Simulation results are given to demonstrate that the proposed DP scheme can increase the spectrum trading revenue and adapt to varying service requirements.