The present invention provides a computer-aided conversion test system and method for generating intelligible speech. The test system includes an acoustic test module with a nasal-genio-oropharyngeal tract, a transmitting module generates a detecting signal, a first receiving module, a second receiving module, and a central processing module with a plurality of first phonetically oral cavity shape spectra. By adjusting the transmitting module, the first receiving module, or the second receiving module, a second phonetically oral cavity shape spectrum is correctly compared and identified by a central computing unit as one of the corresponding first phonetically oral cavity shape spectra. After testing, training and adjusting through the test method, the detecting signal transmitted by the transmitting module is analyzed and identified by the central processing module to increase its interpretation accuracy and shorten the time of machine learning.
展开▼