<![CDATA[Emotion detection from speech is a challenging task, especially when relying solely on acoustic features. This study aims to design and implement a text-based emotion detection system derived from speech transcription using Speech-to-Text (STT) and Natural Language Processing (NLP). The system uses the Whisper model for speech-to-text conversion and a weighted mini-lexicon for emotion classification. The workflow includes audio preprocessing, speech transcription, text normalization, and lexicon-based emotion analysis. The system supports both manual text input and real-time speech input, providing flexibility in use. Performance evaluation using Word Error Rate (WER) for transcription quality and Precision, Recall, and F1-score for emotion detection shows stable and satisfactory results, with an average F1-score of 86.8% and WER of 4.7%, exceeding the common transcription quality standard. The results indicate that a lexicon-based approach supported by Whisper STT technology is an effective, lightweight, and competitive solution for text-based emotion detection in interactive web systems]]>
Copyrights © 2026
