SemanticBoost: Elevating Motion Generation with Augmented Textual Cues