Automatic Pronunciation Calibration Method of Language Resource Base Based on Dynamic Time Rounding Algorithm

OpenAlex · English

The pronunciation accuracy of language resource library is the key to improve the quality of language resource library. An automatic pronunciation calibration model construction method based on dynamic time normalization algorithm is proposed. By analyzing the dynamic characteristics of the pronunciation of the language resource library, the acquisition model of the pronunciation of the language resource library is constructed, and the pronunciation of the language resource library is obtained. The ambiguity detection method is used to suppress the noise of the pronunciation signal of the language resource library. According to the processing results, the speech interaction method of the language resource library is used to analyze the matching domain of the voice signal obtained by interval uniform sampling, and extract the voice characteristics of the language resource library, Based on the extracted features, the dynamic time normalization algorithm is used to recognize the speech similarity, and the voice signal detection model of the optimal language resource library is established under the given false alarm probability, so as to improve the automatic voice calibration capability of the language resource library within the prior Doppler frequency range. The simulation results show that this method has a high accuracy probability and a low false alarm probability for speech detection in the language resource library, which improves the ability of speech interaction and dynamic feature analysis of the language resource library.