Every voice, grouped by the language it speaks.

Match the locale, not just the language. Anen-GB voice reading American copy gets the stress patterns right but the vowels wrong, and the mismatch is more audible than most people expect.

Multilingual voices switch languages mid-sentence.Voices with Multilingual in the identifier handle quoted foreign phrases without a locale change. For single-language copy a plain locale voice is usually cleaner.

Subtitle timing does not vary by voice. Word boundaries come back from the synthesis itself, so every voice here produces subtitles that line up with its own audio exactly.