We introduce Text2EQ, a human-in-the-loop semantic audio production interface that bridges the gap between intuitive language descriptors and equalization (EQ) parameters. Text2EQ enables users to describe their desired audio sound in natural language and maps these descriptors to EQ settings. The system offers initial suggestions and supports iterative refinement, allowing users to adjust parameters through direct manipulation of the EQ parameters or additional natural language inputs. We aim to contribute to the current dialogue on the role and incorporation of intelligent audio tools into pre-existing workflows, highlighting the importance of balancing usability with creativity.