This paper introduces, to the best of the authors' knowledge, the first fine-grained temporal sparsity-aware keyword spotting (KWS) IC leveraging temporal similarities between neighboring feature vectors extracted from input frames and network hidden states, eliminating unnecessary operations and memory accesses. This KWS IC, featuring a bio-inspired delta-gated recurrent neural network (ΔRNN) classifier, achieves an 11-class Google Speech Command Dataset (GSCD) KWS accuracy of 90.5% and energy consumption of 36nJ/decision. At 87% temporal sparsity, computing latency and energy per inference are reduced by 2.4×/3.4×, respectively. The 65nm design occupies 0.78mm^2 and features two additional blocks, a compact 0.084mm^2 digital infinite-impulse-response (IIR)-based band-pass filter (BPF) audio feature extractor (FEx) and a 24kB 0.6V near-Vth weight SRAM with 6.6× lower read power compared to the standard SRAM.