| 摘要 |
The present invention relates to a method for measuring a phonation error in a sentence, and a system and a device therefor, capable of: recognizing voice data for a predetermined sentence; analyzing the recognized voice data using multiple voice recognizers; and giving feedback of information on a phonation error of a user by comparing analysis results, thereby giving feedback of pronunciation, fluency, grammar errors, etc., by recognizing the phonation error of a speaker to be compared with a right sentence even though the speaker does not properly phonate the sentence if the sentence to be learned is inputted without reflecting error patterns. Furthermore, the voice data is recognized through a precise search process using the voice recognizers in realization of voice recognition for a sentence including multiple words in a limited operation environment such as an embedded platform, thereby improving a speed of the voice recognition without decreasing a voice recognition rate and simultaneously improving entire performance of a system. |
