<p>Currently, soundbars feature keyword detection systems for its operations. Keywords such as “Hey Google," “Alexa," and “Siri" etc. are spoken to wake up the soundbar and to give it instructions. However, at high playback volumes, soundbars struggle to identify keywords due to self generated musical noise and external distortions. The computational constraints of onboard processor of soundbar, limits the feasibility of employing adaptive filters and machine learning-based solutions for mitigating high-volume distortion. In this paper, an improved stand alone algorithm for noise suppression and keyword improvement has been proposed to address distortions in soundbars functioning at high volumes. Thus, improving overall keyword spotting accuracy. The proposed framework employs a two-step noise reduction method improved with a biased prior signal-to-noise ratio estimator. Additionally, the approach utilizes a biased non-linear function and harmonic regeneration noise reduction technique to restore harmonics lost during noise reduction. The performance of the proposed method is compared with traditional approaches in terms of keyword detection quality, speech distortion and musical noise suppression. The results demonstrate that the proposed approach effectively reduces musical noise, leading to improved keyword spotting performance. Simulation results indicate significant improvements in false rejection rates compared to the counterparts.</p>