Voices & Languages
Voices & Languages
46 speakers across 23 Indic languages. Any voice can speak any language.
Voice vs Language
In the Shunya TTS system, voice and language are independent concepts:
- Voice controls the speaker character — accent, timbre, pitch, and speaking style. Each voice is trained from a native speaker of a particular language, but it is not restricted to that language.
- Language is determined by the input text you provide. The model automatically detects the script and language of the text and synthesizes speech accordingly.
This means you can pick any voice and feed it text in any of the 23 supported languages. A Hindi-native voice can read Tamil text, a Bengali-native voice can read English, and so on. The voice retains its unique character while speaking the target language fluently.
How it works
| PARAMETER | CONTROLS | EXAMPLE |
|---|---|---|
voice | Speaker identity (accent, timbre) | "Varun" (English-native male) |
input | Spoken language (determined by text) | "नमस्ते" produces Hindi speech |