Arabic texts suffer from missing short vowels. Arabic Speech Recognition is not as good as English speech recognition due to the short vowels not being recognized. And the Arabic language is unlike the English languag...Arabic texts suffer from missing short vowels. Arabic Speech Recognition is not as good as English speech recognition due to the short vowels not being recognized. And the Arabic language is unlike the English language in characteristics such as the number of vowels. English has more than 24 vowels that are close to each other in pronunciation. The Arabic language only has three short vowels that are far from each other in utter and measurement, by elongating those short vowels, long vowels arose. Researchers said that the vowels could be recognized using formants. The formants’ measurements of Arabic vowels are far from each other too, so it is possible to recognize them so that Arabic Speech recognition can give more accurate results. The paper applies this idea to the corpus Phonemes of Arabic. It uses the Euclidian distance method to measure the distances between formant values to recognize Arabic from words with a CV3 structure, the Linear Predictive Coding method and MATLAB to develop the programs that will extract the formants and calculate the means of the short vowels by using the corpus to identify the short vowels within words in the corpus. The results showed that if highly qualified readers were chosen to read the Arabic text, then higher rates of recognition of the short vowels involved in words will be achieved. This paper revealed that some of the characteristics of a language can be utilized for vowel recognition or to enhance the existing methods for speech recognition.展开更多
文摘Arabic texts suffer from missing short vowels. Arabic Speech Recognition is not as good as English speech recognition due to the short vowels not being recognized. And the Arabic language is unlike the English language in characteristics such as the number of vowels. English has more than 24 vowels that are close to each other in pronunciation. The Arabic language only has three short vowels that are far from each other in utter and measurement, by elongating those short vowels, long vowels arose. Researchers said that the vowels could be recognized using formants. The formants’ measurements of Arabic vowels are far from each other too, so it is possible to recognize them so that Arabic Speech recognition can give more accurate results. The paper applies this idea to the corpus Phonemes of Arabic. It uses the Euclidian distance method to measure the distances between formant values to recognize Arabic from words with a CV3 structure, the Linear Predictive Coding method and MATLAB to develop the programs that will extract the formants and calculate the means of the short vowels by using the corpus to identify the short vowels within words in the corpus. The results showed that if highly qualified readers were chosen to read the Arabic text, then higher rates of recognition of the short vowels involved in words will be achieved. This paper revealed that some of the characteristics of a language can be utilized for vowel recognition or to enhance the existing methods for speech recognition.
