2.4.1 Component Modules

• Acoustic Analysis : It visualizes speech signal and its acoustic features, and compare them with the native model speech.
• Acoustic Analysis + Automatic Segmentation : By automatic segmentation, phoneme boundaries are displayed on the acoustic windows to make learners be more easily able to compare their pronunciation with the model's.
• Automatic Pronunciation Scoring : The quality of pronunciation is automatically assessed based on HMM-likelihood score.
• Pronunciation Error Detection : Pronunciation score is examined on the basis of the thresholds that were calculated from native speakers.
• Articulation Instruction : For pronunciation problems, the feedback is given on how to accurately produce the sounds using articulatory configuration and vowel diagrams.
• Speech Rhythm Measurement : Each mora duration is measured based on HMM segmentation and is expressed in the ratio of its length to the word length, not in millisecond.
• Rhythm Error Detection : Rhythmical errors of each mora - longer or shorter - are provided based on the standard rhythm pattern that is measured from native speakers.