<p>Automatic Speaker Verification (ASV) systems have reached a level of sophistication that makes them highly appealing for real-world security applications. These systems still rely on authentication, as they are vulnerable to various direct and indirect attacks despite their advancements. Strengthening ASV technology depends on constant investigation of spoofing and anti-spoofing methods. This work aims to assess and review significant field development proposals suggested by various academics. Using classical, autoregressive, Cepstral, and current deep learning-based techniques employed in creating ASV front-end systems, this investigation examines various feature extraction methods. With an emphasis on the latter in this work, the back-end processing consists of feature classification using either conventional machine learning or deep learning models. Since the inception of ASV research, producing increasingly robust systems has relied on continually evolving datasets and evaluation criteria. This paper examines key fake speech datasets and field-tested evaluation techniques. Speech synthesis (SS), voice conversion (VC), replay attacks, imitation, and twin voice spoofing are some possible hazards to ASV systems). This work aims to enhance ASV defense measures by understanding the methods used to generate these attacks. Moreover, this poll marks a new stage in ASV evolution and introduces fourth-generation SS attack strategies. This study provides a comprehensive introduction to ASV for novices as deep learning advances while also investigating potential future spoofing risks that could compromise next-generation ASV implementations.</p>