Production ASR for noisy multilingual audio
MAI-Transcribe-1 is Microsoft’s new multilingual speech-to-text model built for real-world audio. It delivers best-in-class accuracy across 25 languages, strong robustness in noisy environments, faster batch transcription, and pricing aimed at production speech workflows.