InspirationAbout the Project – SpeechAnalyticsAI
SpeechAnalyticsAI is an AI-powered speech analysis system developed to compare audio recordings and identify potential speech differences. The system analyzes audio features such as duration, energy, pitch, and silence to understand speech characteristics.
It uses Faster-Whisper for speech recognition and Librosa for audio feature extraction. The application generates word-level timestamps, compares recognized speech with a reference transcript, and highlights potential differences for further review.
An interactive Streamlit dashboard presents the analysis results through tables, charts, and downloadable CSV reports. It also supports manual audio evidence review, helping users record observations about possible speech issues.
The main goal of SpeechAnalyticsAI is to make speech analysis more efficient, understandable, and evidence-based. The system supports human verification rather than automatically treating every detected difference as a confirmed speech flaw.
