Less is More: Accurate Speech Recognition & Translation without Web-Scale Data

Benchmark Model Rank Results
speech-recognition-on-aminvidia/canary-1b#20Word Error Rate (WER): 12.42
speech-recognition-on-ami-cleanednvidia/canary-1b#21Word Error Rate (WER): 10.87
speech-recognition-on-earnings-22nvidia/canary-1b#29Word Error Rate (WER): 11.63
speech-recognition-on-gigaspeech-cleanednvidia/canary-1b#24Word Error Rate (WER): 8.47
speech-recognition-on-gigaspeech-testnvidia/canary-1b#25Word Error Rate (WER): 8.55
speech-recognition-on-librispeech-test-cleannvidia/canary-1b#10Word Error Rate (WER): 1.15
speech-recognition-on-librispeech-test-othernvidia/canary-1b#13Word Error Rate (WER): 2.53
speech-recognition-on-spgispeechnvidia/canary-1b#4Word Error Rate (WER): 1.86
speech-recognition-on-voxpopulinvidia/canary-1b#9Word Error Rate (WER): 5.52
speech-recognition-on-voxpopuli-cleanednvidia/canary-1b#32Word Error Rate (WER): 4.01