Multilingual TTS for Sámi languages (+ Finnish and Estonian)

Welcome! This is a demonstration of a multi-lingual and multi-speaker Text-to-Speech (TTS) model. The demo is related to research on TTS for low-resource languages, and the effect of augmenting the training data with areally close languages.

Disclaimers: For convenience, the demo uses pretrained HiFi-GAN vocoder which doesn't work well with male voices. English does not well due to small dataset and orthographic transcriptions. Use the demo just for testing, not for frequent or commercial use.