Model reference · open weights

vits_rasa_13

vits_rasa_13 is an open-weight audio or speech model from ai4bharat. vits_rasa_13 (FP32) weighs 80 MB; the smallest configuration that runs it is RTX 3060 12 GB.

• vits_rasa_13 is a 40M parameter text-to-speech model developed by ai4bharat using the VITS architecture.
• It supports multiple Indian languages, including Assamese, Bengali, and Tamil, and offers predefined speaker profiles and various speaking styles.
• The model is released under the cc-by-4.0 license.

Summary of the ai4bharat/vits_rasa_13 model card, 2026-10-01

Hugging Face ↗