ICMC-ASR: The ICASSP 2024 In-Car Multi-Channel Automatic Speech Recognition Challenge

• DOI: 10.48550/arxiv.2401.03473
• Publication Date: 2024-01-01
• Authors: He Wang; Pengcheng Guo; Yue Li; Ao Zhang; Jiayao Sun; Lei Xie; Wei Chen; Pan Zhou; Hui Bu; Xin Xu; Binbin Zhang; Zhuo Chen; Jian Wu; Longbiao Wang; Eng Siong Chng; Li Sun
• Keywords: Word error rate; Speaker diarisation

To promote speech processing and recognition research in driving scenarios, we build on the success of Intelligent Cockpit Speech Recognition Challenge (ICSRC) held at ISCSLP 2022 launch ICASSP 2024 In-Car Multi-Channel Automatic (ICMC-ASR) Challenge. This challenge collects over 100 hours multi-channel data recorded inside a new energy vehicle 40 noise for augmentation. Two tracks, including automatic (ASR) diarization (ASDR) are set up, using character error rate (CER) concatenated minimum permutation (cpCER) as evaluation metrics, respectively. Overall, ICMC-ASR attracts 98 participating teams receives 53 valid results both tracks. In end, first-place team USTCiflytek achieves CER 13.16% ASR track cpCER 21.48% ASDR track, showing an absolute improvement 13.08% 51.4% compared to our baseline,