Transparency mode is an important feature of modern hearable devices. It lets you hear environmental sounds in real time while keeping you comfortable and aware of what's going on around you. Conventional transparency pipelines usually use static filtering and fixed loudness control systems, which don't work well in very non-stationary acoustic environments and often cause perceptual harshness, clipping, and loudness behaviour that isn't stable.  This paper proposes a low-complexity hybrid adaptive transparency mode architecture that combines statistically driven loudness control, content-aware filtering, and lightweight acoustic feature analysis. Short-time energy and spectral centroid are used together to detect perceptually harsh high-frequency events and classify incoming acoustic frames. According to this classification, harsh events are selectively attenuated through targeted band-stop filtering, while benign ambient sounds are preserved through gentle low-pass filtering. An adaptive RMS-based limiter that uses a combination of exponential moving average and percentile-based ambient estimators to dynamically update its threshold is used to achieve loudness stability.  A real-time MATLAB framework that reflects the constraints of hearable-class digital signal processing is used to implement the suggested system. In comparison to traditional static transparency pipelines, experimental evaluation on real-world environmental recordings shows a lower peak-to-average ratio, better transient handling, quicker recovery, and smoother loudness trajectories while maintaining low latency and computational viability appropriate for embedded hearable devices.