Synthesizing speech based on a given text is an urgent task of our time, since synthesized speech is used in various fields of human activity, for example, in voice self-service systems, when automating telephone survey processes, creating voice-over interactive training materials, etc. Therefore, the task of evaluating the quality of synthesized speech and comparing synthesis systems with each other is an urgent task. There is a significant amount of research in this area, but most of them are characterized by subjectivity of assessments. This article provides a brief analysis of existing approaches to obtaining an objective assessment of the quality of speech and assesses the possibility of using the considered approaches to assess the quality of synthesized speech.
Synthesizing speech based on a given text is an urgent task of our time, since synthesized speech is used in various fields of human activity, for example, in voice self-service systems, when automating telephone survey processes, creating voice-over interactive training materials, etc. Therefore, the task of evaluating the quality of synthesized speech and comparing synthesis systems with each other is an urgent task. There is a significant amount of research in this area, but most of them are characterized by subjectivity of assessments. This article provides a brief analysis of existing approaches to obtaining an objective assessment of the quality of speech and assesses the possibility of using the considered approaches to assess the quality of synthesized speech.