Supported languages
The ten languages Sprag generates speech in, and how voice accent relates to them.
Overview
Chorus generates speech in ten languages, the same ten on every Chorus model.
| English | Portuguese |
| Chinese | Russian |
| Japanese | German |
| Korean | French |
| Spanish | Italian |
Coverage by direction
Transcription reaches further than synthesis. Rhythm transcribes thirty languages plus twenty-two Chinese dialects, and every language above is in that set.
| Direction | Model | Coverage |
|---|---|---|
| Speech out | Chorus | The ten above |
| Speech in | Rhythm | Thirty, plus twenty-two Chinese dialects |
| Speech in and out | Symphony | The eighteen it converses in |
Symphony's model card states nineteen understood languages and lists eighteen. Those sets overlap rather than divide the work: seventeen of Symphony's eighteen are also in Rhythm's thirty. Rhythm's is the wider one, reaching Thai and Hindi among others, and Urdu is the language Symphony has that Rhythm does not list. Speech-to-text languages has the transcription side in full.
Check both directions before committing to a language list. The columns differ, so a product can transcribe a user speaking Thai or Hindi and have no voice to answer them in.
Voices and languages
A preset voice has a native accent, and that is not a restriction on what it can read. A voice listed as Chinese reads English; it sounds like a Chinese speaker reading English.
Use a preset when the accent does not have to match the output language, as in a global assistant. Match the accent to the language for a localized experience.
GET /v1/audio/voices returns ids and a category, nothing about how a voice
sounds. Accent and description are published in Sprag's voice library, covered
under voice presets.
To get a specific accent in a specific language, use voice design: describe the accent you want and the model synthesizes a voice to match.
Setting the language
language selects the output language. Most requests do not need it, since
the model infers the language from input.
Set it when the text is ambiguous: a short string, a proper noun that reads differently in two languages, or a number and date format that could belong to either.