Oral retelling marking method and system

Abstract

The invention discloses an oral retelling marking method comprising the following steps: voice to be marked of standard text content for user retelling is obtained, a BP nerve network-based continuousspeech word cutting algorithm is used for dividing the voice to be marked into word voice, a fluency mark is obtained via calculation according to average duration of the word voice, the word voice is identified as corresponding words via the BP nerve network-based continuous speech word cutting algorithm, and all words are compared with key words in a standard text; a key word score and a text relevancy score are obtained; according to a fluency score, the key word score and the text relevancy score, a comprehensive score of the voice to be marked can be obtained via calculation. Correspondingly, the invention also discloses an oral retelling marking system. Via the oral retelling marking method and system disclosed in an embodiment of the invention, oral retelling can be evaluated in acomprehensive, objective and reasonable manner.

Description

technical field [0001] The invention relates to the field of speech signal processing, in particular to a method and system for scoring oral retelling. Background technique [0002] Oral retelling questions are a common question type to examine students' oral and listening abilities in language learning. They are generally oral English retelling questions. [0003] In recent years, some computer-based oral English test systems based on computer and network technology have emerged, such as the oral computer-based test system of Shanghai Foreign Languages ​​Education Press and the Blue Pigeon system. organization of oral exams. However, in terms of grading, only the grading of objective questions is supported, and the grading of subjective questions still requires a lot of manpower and material resources, such as the repetition questions in the subjective questions of the English test. [0004] At present, the vast majority of oral retelling questions are still manually corr...

AI Technical Summary

Problems solved by technology

[0004] At the present stage, the vast majority of oral retelling questions are still manually corrected, but because the manual correction results are greatly affected by the subjectivity of the testers, the results of the test are affected by many factors such as the physiological conditions, emotions, and time of the testers. factors, and the efficiency of manual marking is low

In addition, there are still some objective evaluation systems for oral repetition questions that only evaluate key words, and most of them only give a score based on keywords, and do not give corresponding scores based on semantics, sentence similarity, etc.

Therefore, it is impossible to give a comprehensive, objective and reasonable evaluation of open questions such as oral retelling questions.

Images