Reading evaluation method based on face and voice combined recognition

Abstract

The invention relates to the technical field, and discloses a face and voice combined recognition reading evaluation method, which comprises an acquisition module, a recognition module, a statement, face and voice combined module and an evaluation module, and is characterized in that the acquisition module comprises a video recognition module, the video recognition module comprises a camera, a microphone, a DSP module and a CPU, the microphone is connected with the DSP module, and the CPU is connected with the recognition module; the system has the beneficial effects that in the error correction process, a reader is guided to enter articles or poems or other reading works to deeply understand the emotions of the works, so that the reading artistic emotion expression is increased, and the reading artistic emotion expression is improved under the condition that only voice recognition evaluation is carried out originally, the same reading emotion effect is achieved, and no voice standard error exists. Evaluation scores have large errors, and accuracy is low. And after face emotion recognition is added, the error of evaluation scores can be maintained within 10 scores, and the relative error is reduced by 30%.

Description

technical field [0001] The invention relates to the technical field of reading aloud evaluation, in particular to a method for reading aloud evaluation with combined recognition of human face and voice. Background technique [0002] Reading aloud is not just about conveying text through sound, but also an expression of emotion. How to score this emotion, how to score it more objectively, and to be more in line with human perception require more dimensions of analysis. [0003] The current mainstream reading evaluation method is to use speech recognition and then analyze the matching degree of speech and text, and score after analyzing the tone and emotion. The evaluation technology of speech recognition can give more objective scores in terms of standardization and accuracy, but in There are relatively large limitations in understanding feelings and artistic emotional expression. The disadvantage of this method is that sometimes more emotional expressions of the reader will ...

AI Technical Summary

Problems solved by technology

[0003] The current mainstream reading evaluation method is to use speech recognition and then analyze the matching degree of speech and text, and score after analyzing the tone and emotion. The evaluation technology of speech recognition can give more objective scores in terms of standardization and accuracy, but in There are relatively large limitations in understanding feelings and artistic emotional expression. The disadvantage of this method is that sometimes more emotional expressions of the reader will be revealed from the face, and it is not enough to judge the emotional expression from speech recognition alone. It is necessary to recognize the expression of the reader through facial vision to provide more dimensional scoring factors. Therefore, the combination of voice and facial emotion recognition can score more full-bodied emotional and emotional expressions in reading, improve the accuracy and standardization of evaluation, and give more Objective scores, for which a reading evaluation method combining face and voice recognition is proposed

Images