Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Multi-talker Automatic Speech Recognition on Libri3Mix Clean (dev)
Loading...
8.6
WER
ID-30.D
8.052
11.751
15.45
19.149
Mar 11, 2026
Mar 13, 2026
Mar 16, 2026
Mar 19, 2026
Mar 22, 2026
Mar 25, 2026
Mar 28, 2026
WER
Updated 19d ago
Evaluation Results
Method
Method
Links
WER
ID-30.D
LLM usage=With LLMs
2026.03
8.6
ID-20.D
LLM usage=With LLMs
2026.03
9.8
w/o decoder: CTC
LLM Usage=Without LLMs...
2026.03
12.7
GEncSep w/o decoder
LLM usage=Without LLMs...
2026.03
12.7
GEncSep
LLM Usage=Without LLMs...
2026.03
13.3
GEncSep
LLM usage=Without LLMs...
2026.03
13.3
ID-8
LLM Usage=With LLMs
2026.03
13.7
Training from Scratch
LLM Usage=Without LLMs...
2026.03
15
Training from Scratch
LLM usage=Without LLMs...
2026.03
15
SOP-Llama-3B
LLM Usage=With LLMs
2026.03
17
SOP-Llama-3B
LLM usage=With LLMs, B...
2026.03
17
SOP-Llama-1B
LLM Usage=With LLMs
2026.03
20.8
SOP-Llama-1B
LLM usage=With LLMs, B...
2026.03
20.8
SOT-Llama-1B
LLM Usage=With LLMs
2026.03
21.5
SOT-Llama-1B
LLM usage=With LLMs, B...
2026.03
21.5
SOT-Llama-3B
LLM Usage=With LLMs
2026.03
22.3
SOT-Llama-3B
LLM usage=With LLMs, B...
2026.03
22.3
Feedback
Search any
task
Search any
task