A Review of the Mandarin-English Code-switching Corpus: SEAME

In this paper, we report the development of the South East Asia Mandarin-English (SEAME) corpus, including 63 hours of transcribed spontaneous Mandarin English code-switching speech in its first release, and an update of additional 129 transcribed hours of speech. The corpus was developed for code-switching speech recognition research, such as LVCSR, language recognition, and language segmentation. It was made publicly available through LDC since 2015. The corpus was recorded under unscripted interview and conversation settings, therefore, consisting of spontaneous speech. This paper seeks to present a comprehensive statistics and analysis of the corpus after the update in term of its composition, speaker profile and code-switch characteristics. This paper will also review its suitability for various code-switch related researches and possible further developments.