Chinese-Mongolian、Uyghur-Chinese Speech Translation Database

The development of end-to-end speech translation is limited by resources.Referring to the research idea of international speech translation data sets, this paper uses the public speech recognition data sets (AISHELL and THUYG-20) to convert them into speech translation data sets through machine translation. After data processing, they are reviewed and verified by experts, so as to obtain high-quality speech translation data sets.The data set includes Sino Mongolian speech translation data set and Uygur Chinese speech translation data set, and the audio sampling rate is 16kHz.

The Sino Mongolian speech translation dataset contains 1919 samples with a size of 238MB. The Uygur Chinese speech translation data set contains 3692 samples with a size of 652MB.This data set can be used for the research of end-to-end speech translation, and provides some data support for exploring the speech translation of Chinese and minority languages.Because the data set has been reviewed and verified by experts, it can also be combined with speech recognition data set to study machine translation.