So created the whole project, tried different transcription models that should run offline and be accurate. Created the full pipeline and architecture. Thought of optimization and continuing along with the help of some yk