notos15/proyecto_diarizacion

This is a local audio-speaker diarization application, available through a console menu and a Flask web interface. The user selects or uploads an audio file; both entry points invoke the same analysis, which prepares audio, finds voice intervals, extracts speaker embeddings, groups them, and builds a report of attributed segments and participation times. The web client displays and can download that report. The README documents no implemented source separation, and the current analysis does not separate overlapping voices. Analysis-stage links follow the README; detailed implementation excerpts were partial.