Speech Recognition On Facebook Multilingual

WER

Results

Performance results of various models on this benchmark

Comparison Table

Model Name	WER
efficient-sequence-transduction-by-jointly	3.93