Yes and no IMO. Some of the improvements have resulted in major changes in performance, the simple transition from electromagnetic magnets to permanent magnets was one such monumental change, others include things like edge wound voice coils, inside/outside wound voice coils, stronger motors with deeper gaps, differential drive motors, cloth surrounds with unique geometries, linen-Kevlar spiders, polyamide insulation systems, Kapton and fiberglass bobbins, epoxy adhesives, polyester filaments within tinsel wire, advancements in cone materials, double spiders, demodulation rings, vented motors etc. All of these improvements have contributed to speakers with 1/10th the unintended distortion and close to 50x the power handling of earlier speakers while maintaining within 3dB of the best sensitivity of the day.
Good question. Based on my perspective, class D will continue to evolve, become more reliable, more efficient, wider bandwidth, greater power, etc. Looking forward 10 years, I don't see any major changes, though if my crystal ball is correct, we may see algorithms that allow digital audio directly to class D under DSP control without going through the discrete PWM conversion that may simplify amplification for the guys working strictly in the digital domain (modelers and profilers).