Automatic Speech Recognition on AfriSpeech-200 (target domains) benchmark leaderboard · SOTA2 Research

Automatic Speech Recognition on AfriSpeech-200 (target domains)

46.5Igbo Error RatePseudo2Real

Evaluation Results

Method	Links
Pseudo2Real2025.10		46.5	42.3	78	38.5	35	44.1	47.1	32.4	48.9	37.1	45
topline2025.10		49.6	35.7	52.5	36.8	35	41.3	45.4	29.5	47.7	36.2	41
EC2025.10		50.6	40.9	63.1	44.3	43.8	50.8	137	30.6	55.5	39.8	55.6
conf.2025.10		52.1	43	78.3	42.5	38.2	50.2	49.9	36.2	55.8	41.9	48.8
theta real s2025.10		52.9	39.7	73.8	40.7	37.5	56.9	52.3	36	55.6	40.3	48.6
theta pseudo t2025.10		53.1	39.9	58.2	41.1	37.5	57.2	52.4	36.4	55.8	40.7	47.2
theta pre2025.10		56.4	49.6	60.6	49.7	49.7	62.8	63.2	41.3	66	49.7	54.9
theta real s2025.10		60.1	67.3	135.3	70.8	53.8	105	155.8	51.8	128.2	54.4	88.2
Pseudo2Real2025.10		60.3	51.9	78.8	67	45.5	61.3	61.3	45.4	60.7	44.8	57.7
theta pseudo t2025.10		61.8	70.5	119.6	75.2	59.4	112.3	157.7	52.5	129	55	89.3
EMA2025.10		62.9	47.3	66.8	49.4	52.9	62	61.8	41	63.7	48.6	55.6
topline2025.10		65.7	45.8	54.1	59.7	88	56.1	58.1	41.4	112.4	44.8	62.6
conf.2025.10		73	58.7	147.7	56	57.7	97.8	194.6	52	107.2	41.9	88.7
EC2025.10		81.7	64.5	93	79.4	75.6	122.2	137.2	47.8	83.7	53.1	83.8
EMA2025.10		88.2	62.4	131.8	79.4	74.7	122	167.1	57.4	140.1	63.8	98.7
theta pre2025.10		93.2	77.7	142.8	79	73.3	94.8	191.2	54.4	182.9	75.6	106.5