No abstract

Buying expensive agricultural produce and fruit such as durians that are unripe can result in a bad experience for a consumer and a loss in profit for a retailer. Therefore, the study of durian striking sounds to create an automatic method of recognizing the ripeness of durians without cutting or damaging them is interesting because it could benefit shoppers and the fruit industry. To solve the problem, the following method of recognizing unripe and ripe striking signals is proposed. First, in the recognition process, the spectral features of the signals are extracted. Then, acoustic models of durian striking sounds are created using syllable-based Hidden Markov Models (HMM). Finally, sequences of syllable-based unripe and ripe durians and defined possible durian ripeness results are applied to recognize the ripeness. Average ripeness recognition rates of more than 90.0% were achieved when using any number of strikes from one to five. When the number of strikes was limited to four and five, higher recognition rates of 95.0% and 92.0% were achieved for the untrained and unknown test sets, respectively. An average total recognition time of less than 80 milliseconds was taken to recognize the unripe and ripe durian striking sounds. The experimental results indicate that the proposed method is a time-efficient, accurate and effective way of recognizing durian ripeness.

No abstract

In this paper, a highly effective system for Thai speech recognition is proposed. The speech recognizer for so-called speaker-independent is created by using Continuous Density Hidden Markov Model (CDHMM). In the acoustic level, the models trained for both speaker genders, and for each separate gender are investigated and tested in terms of accuracy. Experimental evaluation shows that with the acoustic models combination, the accuracy could be improved considerably in the acoustic level. The acoustic combination can support spoken utterances from both genders and still provide the high accuracy simultaneously. Interestingly, when using the acoustic models combination, the syllable accuracy of 89.84% is achieved with 4.53% improvement over using the conventional acoustic models trained for both genders.

No abstract