Azure Speech at Build 2026: Powering Voice Agents with Real-Time and Life-like Experiences

Benchmark Model Rank Results
speech-recognition-on-amimicrosoft/azure-speech-06-2026Word Error Rate (WER): 7.93
speech-recognition-on-ami-cleanedmicrosoft/azure-speech-06-2026Word Error Rate (WER): 6.95
speech-recognition-on-earnings-22microsoft/azure-speech-06-2026Word Error Rate (WER): 7.52
speech-recognition-on-gigaspeech-cleanedmicrosoft/azure-speech-06-2026Word Error Rate (WER): 7.26
speech-recognition-on-gigaspeech-testmicrosoft/azure-speech-06-2026Word Error Rate (WER): 7.34
speech-recognition-on-librispeech-test-cleanmicrosoft/azure-speech-06-2026Word Error Rate (WER): 1.19
speech-recognition-on-librispeech-test-othermicrosoft/azure-speech-06-2026Word Error Rate (WER): 2.49
speech-recognition-on-spgispeechmicrosoft/azure-speech-06-2026Word Error Rate (WER): 2.31
speech-recognition-on-voxpopulimicrosoft/azure-speech-06-2026Word Error Rate (WER): 5.07
speech-recognition-on-voxpopuli-cleanedmicrosoft/azure-speech-06-2026Word Error Rate (WER): 3.87