Can I use MAI-Voice-2.1 audio commercially? What Microsoft says

The Microsoft Learn MAI voices page, dated 2026-10-01 and read on 2026-10-04, says the voices are available to third-party developers and that Microsoft holds full licensing rights for commercial use. That sentence reads as Microsoft's licensing position on the voices; it does not state who owns a given output file, so confirm that point with Microsoft before shipping a paid product.

The same page marks the service as public preview, with no SLA and not recommended for production.

What the page says, line by line

What the page does not settle

Nothing on the page tells you who owns a generated clip, whether a customer can resell it as a standalone asset, or how the preview terms change at general availability. Those are contract questions, not documentation ones.

The launch coverage on Unite.ai lists MAI-Voice-2.1 at $22 per 1M characters and the Flash model at $15 per 1M characters. Treat these as reported numbers until you check Azure's own page.

If the answer needs to be firm today

Sume's TTS endpoint is documented in the API reference. It takes a plain transcript and a voice, returns audio, and is priced at $0.0475 per 1,000 characters. Its catalog lists Sonic models, not MAI voices, so it is an alternative, not a MAI wrapper.

• Use MAI voices in preview for prototypes where an SLA is not needed.
• Ask Microsoft in writing about output ownership for paid products.
• Keep consented voices only; cloning is gated for a reason.
• For a production voiceover today, a generally available endpoint avoids the preview caveat.

Pre-launch review list

• Which MAI voice name and region you call.
• Whether you rely on cloning, which needs approval.
• Your own contract wording on output rights.
• A fallback voice if the preview changes.