This book has its unique value as the pioneer study on the regional variants of Mandarin by experimental phonetics methods, sociophonetic methods, and AI assisted big data methods. Dialectal Mandarins are commonly well-recognized variations for standards Mandarin that have regional characteristics. This book is divided into four parts. The first part introduces Chinese, Chinese dialects, and Mandarin tones. In part two, the book makes a comprehensive and systematic comparison among Beijing Mandarin, Shanghai Mandarin, and Guangzhou Mandarin as the three major Mandarin variants by both the experimental phonetic methods and by using deep learning method. In the deep learning chapter, this book explores whether deep learning can recognize regional dialects patterns from the large amounts of data, and whether it can successfully identify the tonal system of each region’s dialects. The experimental results show that deep learning performs perfectly well in regional dialect recognition and tonal system learning. Liu Yuxuan, a student from USST, provided important technical support for the deep learning experiment. The third part further deepens the research perspective, and studies the geographically close subregional Mandarin. This part involves both acoustic research and perceptual research on three geographically close subregional dialects of Mandarin. The last part is a report on two experiments investigating dialectal experience’s role in predicting listeners’ subjective depiction on both non-categorical and categorical tonal variants of the official language, Putonghua. The dialectal resources recorded in this book are also valuable. The book is supposed to contribute to new progress in academic study on language variation, language evolution, experimental phonetics methods, as well as AI programming.