 This research demonstrated that brain-computer interfaces, BCIs, based on functional near-infrared spectroscopy, FNIRS, can accurately decode speech imagery of vowel sounds. The study also showed that synchronization information in the motor cortex can be used as features to distinguish different vowels. Furthermore, the study found that the decoding performance of speech imagery BCI based on FNIRS can be improved by reducing the time window size and extracting connectivity features. This article was authored by Zinjigwa and Fei Chen.