| 摘要 |
PURPOSE:To improve sound quality and to easily perform pitch extraction by the analysis of a sound source by obtaining a sound path parameter after applying a adaptive inverse filter to a sound waveform, inputting it to a sound path inverse filter with the sound waveform, obtaining a sound source by performing the integration of the output of the filter, integreting the sound source, and inputting it to the sound path parameter with the sound path parameter. CONSTITUTION:The adaptive inverse filter 1 calculates a PARCOR coefficient after multiplying an inputted sound waveform by a window function, and inputs the result to a sound path parameter extraction part 2, then, the sound path parameter can be obtained, and it is inputted to the sound path inverse filter 3 with a sound waveform signal, then, a sound source waveform signal can be obtained, and a differential sound source waveform signal can be obtained with a differentiator 4. The differential sound source waveform signal, after being integrated with an integrator 5, is inputted to the sound path filter 6 with the sound path parameter, then, synthetic speech can be obtained. Pitch analysis is performed after the sound source is processed easy to analyze after applying differentiation, and pitch fluctuation, etc., is performed at need, and a differentiated sound source is returned to an original state after integration, and is synthesized by using the sound path parameter calculated with high accuracy, therefore, the sound quality can be improved. |
