ASCEND: A Spontaneous Chinese-English Dataset for Code-switching in Multi-turn Conversation
This provides a dataset for researchers working on code-switching in conversational AI, though it is incremental as it focuses on a specific language pair and data collection method.
The paper tackles the lack of spontaneous code-switching data by introducing ASCEND, a high-quality Mandarin Chinese-English corpus built from multi-turn conversations, resulting in 10.62 hours of speech and baseline experiments achieving 22.69% character error rate and 27.05% mixed error rate.
Code-switching is a speech phenomenon occurring when a speaker switches language during a conversation. Despite the spontaneous nature of code-switching in conversational spoken language, most existing works collect code-switching data from read speech instead of spontaneous speech. ASCEND (A Spontaneous Chinese-English Dataset) is a high-quality Mandarin Chinese-English code-switching corpus built on spontaneous multi-turn conversational dialogue sources collected in Hong Kong. We report ASCEND's design and procedure for collecting the speech data, including annotations. ASCEND consists of 10.62 hours of clean speech, collected from 23 bilingual speakers of Chinese and English. Furthermore, we conduct baseline experiments using pre-trained wav2vec 2.0 models, achieving a best performance of 22.69\% character error rate and 27.05% mixed error rate.