SpeechSynthesiser distinguishes and correctly selects macOS voice variants

On macOS, SpeechSynthesiser can list several voices with exactly the same name. These entries represent different language or quality variants, but selecting any of the identically named entries currently uses the same variant.

Give each available variant a distinguishable name and ensure that assigning an entry from Voices to Voice selects that exact variant for both spoken audio and WAV export. Preserve the existing string-based API, default voice behaviour and compatibility with existing voice names.