当前，机器合成普通话广泛进入教育教学、新闻传媒、公共服务等现实场景，如何将普通话规范要求纳入其评价体系，成为语言文字规范建设与智能语音治理中的重要议题。面向自然人的普通话水平测试早已成熟，其面向机器的拓展，本质上是评价对象改变所引发的相关理论与实施方法的重新建构，考查重点应转为合成语音对普通话语言规范的实现程度及表现的自然程度。教育部语言文字应用研究所研制《机器合成普通话水平测评等级标准及测评大纲》过程中，就测评需求确立、对象扩展适配、制度经验迁移和标准科学验证等关键问题提出“目标导向”“人机对齐”“借鉴创新”“规范科学”4项基本原则，以“继承等级框架、重构分级依据”为基本思路，延续普通话水平测试中的三级六等结构及其失分率设置，重新描述等级特征，并围绕题型与分值、题量及内容、评分标准及测评方法和实验验证机制等环节进行建构，在保持普通话测试制度连续性的同时，形成了能够反映机器合成普通话特性的分级体系。

With the rapid advancement of speech synthesis technology, synthesized Putonghua has been widely applied in education, news media, public services, and other real-world scenarios. Consequently, increasing attention is being paid to its linguistic normativity and speech quality. Existing speech synthesis evaluations primarily focus on technical performance and do not adequately address the needs of language governance, including the national standards for spoken and written Chinese, and Putonghua proficiency assessment. Drawing on the development practice of the Standards and Guidelines of Synthesized Putonghua Proficiency Assessment, this study builds upon the theoretical and institutional foundations of the Putonghua Proficiency Test (PSC) and examines the core approaches and frameworks for synthesized Putonghua proficiency assessment, particularly in light of the shifts from evaluating human speakers to machines. The study proposes four guiding principles: goal orientation, human-machine alignment, inheritance with innovation, and normativity and scientific validity. Key components discussed include level classification, characteristic descriptors, task design, scoring methods, and experimental validation. This study outlines a structured approach to constructing both assessment standards and assessment design. It argues that synthesized Putonghua proficiency assessment is not simply an expansion of the PSC’s scope; rather, it requires a reconstruction of rating logic, assessment content, and implementation mechanisms in response to changes in assessment objects and represents an important extension of language governance in the era of intelligent speech technologies.