Suseli (सुसेली)

Natural Nepali text-to-speech by Fiste AI. In development.

Suseli is a text-to-speech model for the Nepali language, being built for:

• Natural Nepali (स्वाभाविक नेपाली): clear pronunciation and the rhythm of spoken Nepali, including the rise at the end of a question.
• Expressive speech (भावसहितको बोली): calm, cheerful, serious or sad on request.
• Many voices (धेरै आवाज): built-in voices, and a new voice from a short recording, only with the speaker's consent.
• Small and fast (सानो र छिटो): a compact version that runs on a laptop processor or a phone.

The weights are not released yet. Follow Fiste AI on Hugging Face or on GitHub to hear about the first release.

The name

सुसेली (suseli) is a tune you whistle. फिस्टो, also called फिस्टे (fiste), is the name for the small warbler birds of Nepal; the सुसेली फिस्टो is the Whistler's warbler.

Responsible use

A voice belongs to a person. Suseli will create a new voice only from recordings the user has the right to use, and its license will forbid impersonation and fraud.