YouTube, one of the most visited platforms worldwide, generates immense data through user content and interactions. This information, especially in the form of comments, presents both an opportunity and a challenge for understanding public sentiment and content acceptance. The problem lies in efficiently processing and analyzing this large amount of textual data to extract meaningful insights about audience reactions and preferences, particularly when dealing with the Spanish language and its slang in social media. To address this challenge, we propose a solution that uses a Bidirectional Encoder Representation from Transformers (BERT) model for sentiment classification in YouTube comments. BERT’s advanced Natural Language Processing (NLP) capabilities, practical implementation, and low computational requirements make it an ideal choice for this task. Additionally, we utilize parallel computing techniques to handle large volumes of data efficiently, improving the speed and accuracy of the analysis. Our approach involves extracting comments from specific YouTube channels, preprocessing them through data cleaning and tokenization, and passing them through the BERT model trained to classify sentiment as positive or negative to understand audience acceptance. The results demonstrate high accuracy in sentiment classification across various types of content, showing potential for application in diverse YouTube channels and genres. This provides a powerful tool for evaluating public acceptance, offering content creators, marketers, and researchers valuable information about user sentiment. This can influence content strategies and audience engagement tactics. The method’s versatility opens up possibilities for broader applications in social media analysis and public opinion research.