Three-stage training and orthogonality regularization for spoken language recognition

EURASIP Journal on Audio, Speech, and Music Processing

Table 1 Previous proposed methods

	Year	encoder	ASR-loss	Strategy
Ren [25]	2019	ResNet	CTC	2-stage-F
Zhao [23]	2019	TDNN	Cross-entropy	Multi-task
Wang [26]	2021	Conformer	Hybrid CTC/attention	2-stage
Li [24]	2021	TDNN	Cross-entropy	Multi-task
Duroselle [27]	2021	Conformer	CTC	2-stage-F
Duroselle [16]	2021	Conformer	CTC	2-stage
Alum [28]	2022	Conformer	Unknown	2-stage