Automatic Speech Recognition on LibriSpeech (WER, length) benchmark leaderboard · SOTA2 Research

Automatic Speech Recognition on LibriSpeech (WER, length)

2.99WERclean

Evaluation Results

Method	Links
clean2026.01		2.99	21.84
clean2026.01		3.01	21.84
clean2026.01		3.64	21.84
clean2026.01		4.9	21.77
clean2026.01		6.66	21.84
VMI-FGSM2026.01		29.77	20.79
SAGO2026.01		30.26	21.12
SlothSpeech2026.01		31.09	81.75
PGD2026.01		33.33	21.8
SlothSpeech2026.01		34.21	79.78
MI-FGSM2026.01		34.27	21.4
SlothSpeech2026.01		38.25	110.4
SlothSpeech2026.01		46.8	119.38
MORE2026.01		53.72	301.47
SlothSpeech2026.01		54.63	156.07
VMI-FGSM2026.01		60.74	25.38
MI-FGSM2026.01		61.16	25.7
SAGO2026.01		62.46	25.97
MORE2026.01		64.04	234.25
PGD2026.01		64.77	26.94
VMI-FGSM2026.01		73.84	27.34
MI-FGSM2026.01		74.18	27.07
MORE2026.01		74.28	213.94
SAGO2026.01		74.67	27.45
PGD2026.01		75.23	27.93
MI-FGSM2026.01		87.57	34.44
VMI-FGSM2026.01		87.91	32.39
SAGO2026.01		88.23	33.85
MORE2026.01		88.42	300.13
PGD2026.01		88.73	31.65
MORE2026.01		91.01	296.28
PGD2026.01		93.17	35.02
SAGO2026.01		93.19	31.93
MI-FGSM2026.01		93.32	34.38
VMI-FGSM2026.01		93.47	33.9