September 2026
Microsoft's new speech model joins Whisper Memos. It is our fastest option yet, with 60 languages, custom words, and Meeting mode.
Whisper Memos now supports Microsoft MAI Transcribe 2, the new speech-to-text model from Microsoft AI. It is exceptionally fast. In our tests, it transcribed 90 minutes of audio in under 10 seconds, so even long recordings come back almost instantly.
It supports 60 languages and detects the spoken language on its own. It is available on every plan.
MAI Transcribe 2 uses your custom words, so names and jargon come out spelled the way you expect. It also separates speakers in Meeting mode, and unlike Gemini, it can do both at the same time. Very long meetings may use another model instead.
Open Settings → AI Model and pick Microsoft MAI Transcribe 2. Every new recording will use it, and existing memos stay as they are. Automatic mode doesn't pick MAI Transcribe 2, so you need to select it yourself.
Compare all AI models to find the one that fits your recordings best, or update Whisper Memos to try MAI Transcribe 2 on your next memo.
