Ultrasound Tongue Imaging Dataset for Vowel Articulation in Tibetan Lhasa Dialect

This dataset is an ultrasound tongue imaging collection established to study vowel articulation features in the Tibetan Lhasa Dialect and support speech engineering research. The study employed dynamic ultrasound tongue imaging technology to synchronously capture articulatory organ (tongue) movement data from native speakers. Data collection was conducted in a standard recording studio environment with two Lhasa-native participants (1 males, 1 females), all university students with Lhasa Dialect as their mother tongue and free of speech disorders. The experiment utilized B-mode ultrasound equipment (probe frequency: 5.0 MHz) to record real-time tongue movement images during articulation, paired with high-fidelity audio recording devices (sampling rate: 44.1 kHz) to capture speech signals. The dataset includes eight monophthongs ([a], [i], [u], [e], [ø], [ɛ], [y], [o]) and two diphthongs ([au], [iu]), yielding 200 valid data units (including ultrasound videos, audio recordings, and corresponding text annotations). All data were collected under standardized protocols: participants maintained an upright sitting posture with head stabilization and ultrasound probe fixation, while speech intensity was controlled within 65-75 dB. To ensure data quality, the collected data were double-checked by two phonetics experts to eliminate unqualified samples. This dataset fills the gap in articulatory research on Lhasa Tibetan pronunciation, providing essential foundational data support for Tibetan language phonetics teaching, speech pathology treatment, speech recognition and synthesis, and multimodal deep learning research. It can also be used for cross-language comparative studies of articulatory features, offers technical support for endangered language preservation, and holds significant value for linguistic research and speech engineering applications.