<p>With the diversification of music genres and the development of music recommendation systems, it is of great significance to use sound sensing and audio feature extraction technology to classify and recommend music. This study aims to explore methods based on sound sensing and audio feature extraction for precise classification and personalized recommendation of music genres, in order to improve the effectiveness and user experience of music recommendation systems. A sound sensor is a device that converts sound signals into electrical signals and captures audio data by sensing sound wave vibrations. The study collected audio data from different music genres through sound sensors to obtain rich music samples. The audio feature extraction algorithm extracts representative features from the original audio data and accurately captures the unique features of each music genre. The classification and recommendation models based on machine learning algorithms can utilize these music features to achieve automatic classification and personalized recommendation of music genres. These models can learn from a large amount of music data, gradually improve the accuracy of classification and recommendation, and provide users with music content that suits their preferences. The recommendation effect is more accurate and personalized.</p>