MAI-Voice-2.1 has 97 voices: en-US has 7, es-ES and en-AU have 1 each

Microsoft's MAI-Voice-2.1 page lists 97 prebuilt voices across 28 locale codes, but they are spread unevenly: en-US has 7 and Spanish (Spain) and Australian English have 1 each. Counting the list ourselves, a buyer casting a Spanish-Spain or en-AU ad gets no choice of speaker.

Voices per locale

We counted the voice names in the Learn page (public preview, no SLA). Totals sum to 97.

Why the count matters

An ad series with a recurring presenter needs one voice per language that still sounds like a person you would hire. With one voice, both the male and female option collapse to a single sound, and a follow-up campaign has nobody else to use. Check that the count in your market covers the roles you need before pricing at $22 per million characters (standard) or $15 (Flash).

Sume's side

Sume TTS 1.0 picks a voice by voice.id or by a ready avatar voice, so choice depends on the voices in your account, not a fixed per-locale table. List them before promising a count to a client; the API reference shows the shape. We do not publish a per-language voice total here because we did not verify one.