Pemodelan Harmonik untuk Pelafalan Makhraj Huruf Hijaiah

Abstract

Learning to pronounce hijaiah letters needs to be assessed objectively, so it is necessary to form digital audio resulting from the synthesis of Harmonic Plus Residual (HPR) modeling, which conducted with two pronunciation methods, taskin and tasydid. The experiment consists data acquisition, signal cutting, framing and windowing, detection of fundamental and harmonic frequencies, synthesis of HPR, to produce synthetic signals. The results of the synthetic signals then analyzed qualitatively by signal spectrogram analysis and scoring.
From the experimental results, it can be concluded that this study was ultimately unable to determine the best HPR parameters, but concluded that the tasydid method was the best method for learning pronunciation and for the HPR model synthesis. This is because the tasydid method with different parameters but all of them can produce good synthetic signal, both in terms of comparative analysis of similar signal spectrograms and from the results of scoring with an average value of 10. On the other hand, the taskin method harf shows unsatisfactory results, where the synthetic sound is mostly just noise, so the scoring results is under 9, and is reinforced by the results of the spectrogram comparison between the original and synthetic signals which visually different.