Full catalogue

This is a database of moving images of the midsagittal section of the vocal tract during the production of Japanese utterances, recorded at a rate of 14 or 27 frames per second by using a medical MRI system with special operating settings. This data has realized the dream of articulatory phoneticians to visualize articulatory movements and may be widely used for critical review of the existing articulatory phonetic descriptions, education of Japanese pronunciation, or speech synthesis from vocal tract shape. The data has been collected by the author's research group since 2017. In April 2021 the trial version of rtMRIDB containing data of about 13,000 utterances by 10 speakers was released. In April 2022, Version 1 of the database (abb. rtMRIDB_v1) containing more than 26,000 utterances by 22 speakers was released. In April 2024, Version 2 of the database (abb. rtMRIDB_v2) became publicly available. This version included the following enhancements: (1) Data from three new speakers was added, providing a total of approximately 30,000 utterances from 25 speakers. (2) New movies with superposed dentition image were provided for all utterances of all speakers. (3) New audio noise reduction technique was applied for all audio data. (4) Some improvements in the search system