VibeVoice-ASR Technical Report

Benchmark Model Rank Results
speech-recognition-on-amimicrosoft/VibeVoice-ASR-HF#37Word Error Rate (WER): 15.48
speech-recognition-on-ami-cleanedmicrosoft/VibeVoice-ASR-HF#23Word Error Rate (WER): 12.03
speech-recognition-on-earnings-22microsoft/VibeVoice-ASR-HF#34Word Error Rate (WER): 12.15
speech-recognition-on-gigaspeech-cleanedmicrosoft/VibeVoice-ASR-HF#8Word Error Rate (WER): 7.87
speech-recognition-on-gigaspeech-testmicrosoft/VibeVoice-ASR-HF#9Word Error Rate (WER): 8.05
speech-recognition-on-librispeech-test-cleanmicrosoft/VibeVoice-ASR-HF#40Word Error Rate (WER): 1.55
speech-recognition-on-librispeech-test-othermicrosoft/VibeVoice-ASR-HF#81Word Error Rate (WER): 4.93
speech-recognition-on-spgispeechmicrosoft/VibeVoice-ASR-HF#41Word Error Rate (WER): 3.47
speech-recognition-on-voxpopulimicrosoft/VibeVoice-ASR-HF#44Word Error Rate (WER): 7.9
speech-recognition-on-voxpopuli-cleanedmicrosoft/VibeVoice-ASR-HF#2Word Error Rate (WER): 2.02