Mixture Encoder for Joint Speech Separation and Recognition

被引:2
作者
Berger, Simon [1 ,2 ]
Vieting, Peter [1 ]
Boeddeker, Christoph [3 ]
Schlueter, Ralf [1 ,2 ]
Haeb-Umbach, Reinhold [3 ]
机构
[1] Rhein Westfal TH Aachen, Dept Comp Sci, Machine Learning & Human Language Technol, Aachen, Germany
[2] AppTek GmbH, Berlin, Germany
[3] Paderborn Univ, Paderborn, Germany
来源
INTERSPEECH 2023 | 2023年
关键词
speech separation; speech recognition; meeting transcription;
D O I
10.21437/Interspeech.2023-1815
中图分类号
O42 [声学];
学科分类号
070206 ; 082403 ;
摘要
Multi-speaker automatic speech recognition (ASR) is crucial for many real-world applications, but it requires dedicated modeling techniques. Existing approaches can be divided into modular and end-to-end methods. Modular approaches separate speakers and recognize each of them with a single-speaker ASR system. End-to-end models process overlapped speech directly in a single, powerful neural network. This work proposes a middle-ground approach that leverages explicit speech separation similarly to the modular approach but also incorporates mixture speech information directly into the ASR module in order to mitigate the propagation of errors made by the speech separator. We also explore a way to exchange cross-speaker context information through a layer that combines information of the individual speakers. Our system is optimized through separate and joint training stages and achieves a relative improvement of 7% in word error rate over a purely modular setup on the SMS-WSJ task.
引用
收藏
页码:3527 / 3531
页数:5
相关论文
共 33 条
[1]   IMAGE METHOD FOR EFFICIENTLY SIMULATING SMALL-ROOM ACOUSTICS [J].
ALLEN, JB ;
BERKLEY, DA .
JOURNAL OF THE ACOUSTICAL SOCIETY OF AMERICA, 1979, 65 (04) :943-950
[2]  
Chang X., 2018, P ICASSP, P6256
[3]   MONAURAL SOURCE SEPARATION: FROM ANECHOIC TO REVERBERANT ENVIRONMENTS [J].
Cord-Landwehr, Tobias ;
Boeddeker, Christoph ;
Von Neumann, Thilo ;
Zorila, Catalin ;
Doddipatla, Rama ;
Haeb-Umbach, Reinhold .
2022 INTERNATIONAL WORKSHOP ON ACOUSTIC SIGNAL ENHANCEMENT (IWAENC 2022), 2022,
[4]  
Drude L., 2019, ARXIV191013934
[5]  
Hadian H, 2018, INTERSPEECH, P12
[6]  
Heitkaemper J, 2020, INT CONF ACOUST SPEE, P6359, DOI [10.1109/icassp40776.2020.9052981, 10.1109/ICASSP40776.2020.9052981]
[7]   Single-Channel Multi-Speaker Separation using Deep Clustering [J].
Isik, Yusuf ;
Le Roux, Jonathan ;
Chen, Zhuo ;
Watanabe, Shinji ;
Hershey, John R. .
17TH ANNUAL CONFERENCE OF THE INTERNATIONAL SPEECH COMMUNICATION ASSOCIATION (INTERSPEECH 2016), VOLS 1-5: UNDERSTANDING SPEECH PROCESSING IN HUMANS AND MACHINES, 2016, :545-549
[8]   Serialized Output Training for End-to-End Overlapped Speech Recognition [J].
Kanda, Naoyuki ;
Gaur, Yashesh ;
Wang, Xiaofei ;
Meng, Zhong ;
Yoshioka, Takuya .
INTERSPEECH 2020, 2020, :2797-2801
[9]   Multitalker Speech Separation With Utterance-Level Permutation Invariant Training of Deep Recurrent Neural Networks [J].
Kolbaek, Morten ;
Yu, Dong ;
Tan, Zheng-Hua ;
Jensen, Jesper .
IEEE-ACM TRANSACTIONS ON AUDIO SPEECH AND LANGUAGE PROCESSING, 2017, 25 (10) :1901-1913
[10]   Streaming Multi-talker Speech Recognition with Joint Speaker Identification [J].
Lu, Liang ;
Kanda, Naoyuki ;
Li, Jinyu ;
Gong, Yifan .
INTERSPEECH 2021, 2021, :1782-1786