In the rapidly growing digital era, user reviews on distribution platforms such as the Google Play Store are a key indicator in assessing the popularity, quality, and user satisfaction of applications. This study aims to compare the performance of SVM, Naive Bayes, and Logistic Regression classification algorithms in analyzing user reviews of the Noice app, an audio content platform. The research involves steps such as data collection, data pre-processing, word embedding, modeling, model evaluation, and sentiment analysis. Testing was conducted using 1877 data. The data from the reviews were divided into scenarios, with training and testing data divided in ratios of 90:10, 80:20, and 70:30. The results showed that the SVM algorithm achieved the highest accuracy rate (80%) in the 90:10 data split scenario. However, Naive Bayes also showed competitive results with 78% accuracy in the same scenario. Meanwhile, Logistic Regression achieved 78% accuracy when the data was split in an 80:20 ratio. Evaluation was done using metrics such as accuracy, precision, recall, and F1-score. Sentiment analysis showed a positive trend with 1194 positive data compared to 683 negative data. From the comparison of data sharing scenarios and algorithms, SVM at 90:10 data sharing gave the best results.