MAI-Voice-2.1 pt-BR: six voices, four male, for ad casting

Microsoft lists six pt-BR voices for MAI-Voice-2.1: Caio, Grant, Pedro and Rafael are male, Harper and Luana are female. Only Caio and Luana list both whispering and shouting, and four voices list excited. For a Brazilian ad read, that is a small cast with two clearly expressive picks (read 2026-10-07).

The six voices

The voice table on Microsoft Learn gives each voice its own list of styles. I counted them for pt-BR. Grant and Harper are the two shared names that Microsoft repeats across 24 locales, and here they carry role styles, not emotion.

Reading the table for an ad

For a sale spot, start with Luana or Caio: they have the longest style lists and the only shouting. Rafael and Pedro give two more male voices with excited and happy. Harper is the narrator voice, with no emotion styles, so she suits an explainer more than a promo. Grant fits a support line.

The cast is lopsided: four male and two female. If the brand voice should be female and energetic, Luana is the only pt-BR choice that lists excited. Plan for that and audition her first.

• Four voices list excited: Caio, Luana, Pedro and Rafael.
• Two list whispering and shouting: Caio and Luana.
• Neutral-only voices in pt-BR: none.

Costs, and the same ad on Sume

A 30-second ad read of about 450 characters (an assumption of 15 characters per second) costs 450 x $15 / 1,000,000 = $0.00675 on Flash and 450 x $22 / 1,000,000 = $0.0099 on the standard model, from the prices on Microsoft's model page.

On Sume the same 450 characters is 0.45 x $0.0475 = $0.021, which bills 3 cents. Set language to pt for the request. Sume's voice library tags Portuguese as pt, with no split between Brazil and Portugal, so judge the accent by ear and pick a voice whose sample sounds Brazilian.

Brazil or Portugal

MAI-Voice-2.1 lists pt-PT as a separate locale with three voices, two of them male, and two listed as neutral-only. If the audience is in Brazil, do not take a pt-PT voice because it is the first on the list. Vocabulary and pronunciation differ, and a Brazilian listener hears the difference in the first sentence.

Always have a native speaker check the final take before you spend on media.

A quick audition plan

Render the same 60-character line with each of the four excited voices and the two role voices, six files in all. On Sume that is six jobs of 1 cent each, 6 cents total. On Microsoft's side a 60-character line costs $0.0009 on Flash, so the audition is nearly free there as well.

Listen for stress on numbers and brand names first. Prices, dates and phone numbers are where a voice slips, and a Brazilian audience will notice an odd reading of a currency amount faster than any difference in tone.