svara-TTS
Open-source multilingual text-to-speech model for Indian languages, enabling natural voice synthesis across regional languages and dialects
huggingface.cosvara-TTS is an open-source multilingual text-to-speech (TTS) model family developed by Kenpath Labs specifically for Indian languages and accents. It addresses the shortage of high-quality, open-weights voice synthesis models for Indic languages, supporting natural pronunciation, regional phonetic nuances, and conversational rhythm across major Indian language families.
The model is released under the Apache 2.0 licence, making it usable for civic technologists, community radio initiatives, educational tools, and accessibility applications without proprietary cloud API costs. It supports local inference via GGUF, vLLM, and FastAPI pipelines, allowing organisations to run voice synthesis self-hosted and offline for privacy-sensitive deployments and rural IVR workflows.
Details
- Kind
- Tool
- Run by
- Commercial
- Topics
- Where
- India, Global
- Licence
- Open licence
- Cost
- Free
- Licence terms
- Apache-2.0