Efficient Sequence Transduction by Jointly Predicting Tokens and Durations

Benchmark Model Rank Results
intent-classification-on-slurpTDT 0-8#1Accuracy (%): 90.07
slot-filling-on-slurpTDT 0-6#1F1: 0.8061
speech-recognition-on-aminvidia/parakeet-tdt-0.6b-v2#14Word Error Rate (WER): 10.4
speech-recognition-on-aminvidia/parakeet-tdt_ctc-110m#26Word Error Rate (WER): 14.7
speech-recognition-on-aminvidia/parakeet-tdt-1.1b#29Word Error Rate (WER): 14.81
speech-recognition-on-ami-cleanednvidia/parakeet-tdt-0.6b-v2#15Word Error Rate (WER): 9.1
speech-recognition-on-ami-cleanednvidia/parakeet-tdt_ctc-110m#27Word Error Rate (WER): 12.69
speech-recognition-on-ami-cleanednvidia/parakeet-tdt-1.1b#29Word Error Rate (WER): 12.87
speech-recognition-on-earnings-22nvidia/parakeet-tdt-0.6b-v2#20Word Error Rate (WER): 10.78
speech-recognition-on-earnings-22nvidia/parakeet-tdt_ctc-110m#30Word Error Rate (WER): 12.02
speech-recognition-on-earnings-22nvidia/parakeet-tdt-1.1b#42Word Error Rate (WER): 13.25
speech-recognition-on-gigaspeech-cleanednvidia/parakeet-tdt-1.1b#9Word Error Rate (WER): 7.9
speech-recognition-on-gigaspeech-cleanednvidia/parakeet-tdt-0.6b-v2#14Word Error Rate (WER): 8.15
speech-recognition-on-gigaspeech-cleanednvidia/parakeet-tdt_ctc-110m#35Word Error Rate (WER): 8.84
speech-recognition-on-gigaspeech-testnvidia/parakeet-tdt-1.1b#8Word Error Rate (WER): 7.98
speech-recognition-on-gigaspeech-testnvidia/parakeet-tdt-0.6b-v2#12Word Error Rate (WER): 8.23
speech-recognition-on-gigaspeech-testnvidia/parakeet-tdt_ctc-110m#33Word Error Rate (WER): 8.93
speech-recognition-on-librispeech-test-cleannvidia/parakeet-tdt-1.1b#5Word Error Rate (WER): 1.05
speech-recognition-on-librispeech-test-cleannvidia/parakeet-tdt-0.6b-v2#17Word Error Rate (WER): 1.27
speech-recognition-on-librispeech-test-cleannvidia/parakeet-tdt_ctc-110m#72Word Error Rate (WER): 2.01
speech-recognition-on-librispeech-test-othernvidia/parakeet-tdt-1.1b#4Word Error Rate (WER): 2.22
speech-recognition-on-librispeech-test-othernvidia/parakeet-tdt-0.6b-v2#18Word Error Rate (WER): 2.73
speech-recognition-on-librispeech-test-othernvidia/parakeet-tdt_ctc-110m#77Word Error Rate (WER): 4.71
speech-recognition-on-spgispeechnvidia/parakeet-tdt-0.6b-v2#9Word Error Rate (WER): 1.94
speech-recognition-on-spgispeechnvidia/parakeet-tdt_ctc-110m#15Word Error Rate (WER): 2.26
speech-recognition-on-spgispeechnvidia/parakeet-tdt-1.1b#30Word Error Rate (WER): 2.72
speech-recognition-on-voxpopulinvidia/parakeet-tdt-1.1b#2Word Error Rate (WER): 5.22
speech-recognition-on-voxpopulinvidia/parakeet-tdt-0.6b-v2#13Word Error Rate (WER): 5.68
speech-recognition-on-voxpopulinvidia/parakeet-tdt_ctc-110m#29Word Error Rate (WER): 6.63
speech-recognition-on-voxpopuli-cleanednvidia/parakeet-tdt-1.1b#24Word Error Rate (WER): 3.62
speech-recognition-on-voxpopuli-cleanednvidia/parakeet-tdt_ctc-110m#26Word Error Rate (WER): 3.72
speech-recognition-on-voxpopuli-cleanednvidia/parakeet-tdt-0.6b-v2#31Word Error Rate (WER): 3.78