Speech-Forensics: Towards Comprehensive Synthetic Speech Dataset Establishment and Analysis