ICMC-ASR: THE ICASSP 2024 IN-CAR MULTI-CHANNEL AUTOMATIC SPEECH RECOGNITION CHALLENGE

被引：1

作者：

Wang, He ^{[1
]}

Guo, Pengcheng ^{[1
]}

Li, Yue ^{[1
]}

Zhang, Ao ^{[1
]}

Sun, Jiayao ^{[1
]}

Xie, Lei ^{[1
]}

Chen, Wei ^{[2
]}

Zhou, Pan ^{[2
]}

Bu, Hui ^{[3
]}

Xu, Xin ^{[3
]}

Zhang, Binbin ^{[4
]}

Chen, Zhuo ^{[5
]}

Wu, Jian ^{[6
]}

Wang, Longbiao ^{[7
]}

Chng, Eng Siong ^{[8
]}

Li, Sun ^{[9
]}

机构：

[1] Northwestern Polytech Univ, Xian, Peoples R China

[2] Space AI, LI Auto, Chengdu, Peoples R China

[3] Beijing AI Shell Technol Co Ltd, Beijing, Peoples R China

[4] WeNet Open Source Community, Shanghai, Peoples R China

[5] ByteDance, Beijing, Peoples R China

[6] Microsoft Corp, Redmond, WA USA

[7] Tianjin Univ, Tianjin, Peoples R China

[8] Nanyang Technol Univ, Singapore, Singapore

[9] China Acad Informat & Commun Technol, Beijing, Peoples R China

来源：

2024 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH, AND SIGNAL PROCESSING WORKSHOPS, ICASSPW 2024 | 2024年

关键词：

Multi-channel; Automatic Speech Recognition;

D O I：

10.1109/ICASSPW62465.2024.10627712

中图分类号：

O42 [声学];

学科分类号：

070206 ; 082403 ;

摘要：

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

引用

页码：63 / 64

页数：2

共 13 条

[11]

Yao Zhuoyuan, 2021, Wenet: Production oriented streaming and non-streaming end-to-end speech recognition toolkit

[12] The ISCSLP 2022 Intelligent Cockpit Speech Recognition Challenge (ICSRC): Dataset, Tracks, Baseline and Results [J].

Zhang, Ao ;

Yu, Fan ;

Huang, Kaixun ;

Xie, Lei ;

Wang, Longbiao ;

Chng, Eng Siong ;

Bu, Hui ;

Zhang, Binbin ;

Chen, Wei ;

Xu, Xin .

2022 13TH INTERNATIONAL SYMPOSIUM ON CHINESE SPOKEN LANGUAGE PROCESSING (ISCSLP), 2022, :507-511

[13]

Zhang SL, 2018, 2018 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP), P5869, DOI 10.1109/ICASSP.2018.8461404

← 1 2 →