|
|
--- |
|
|
title: Svara TTS |
|
|
emoji: 🗣️ |
|
|
colorFrom: blue |
|
|
colorTo: indigo |
|
|
sdk: gradio |
|
|
sdk_version: 5.49.1 |
|
|
app_file: app.py |
|
|
pinned: false |
|
|
license: apache-2.0 |
|
|
--- |
|
|
|
|
|
# 🗣️ Svara-TTS |
|
|
|
|
|
**An open multilingual text-to-speech (TTS) model bringing expressive, human-like speech to India’s languages.** |
|
|
|
|
|
Svara-TTS is trained on **1,900+ hours** of diverse Indian speech across **17 languages**, with balanced male and female voices. It captures **emotion**, **tone**, and **rhythm**, producing natural, human-like speech for real-world Indian use cases. |
|
|
|
|
|
> 💡 Supports `<happy>`, `<sad>`, `<fear>`, `<relief>`, and other emotion tags — plus multilingual and zero-shot voice cloning. |
|
|
|
|
|
--- |
|
|
|
|
|
## 🌍 Supported Languages |
|
|
|
|
|
Hindi, Marathi, Tamil, Telugu, Bengali, Kannada, Gujarati, Malayalam, Punjabi, Assamese, Odia, Bhojpuri, Chhattisgarhi, Maithili, Magahi, Nepali, Sanskrit, Bodo and English |
|
|
|
|
|
--- |
|
|
|
|
|
### 🧩 About |
|
|
|
|
|
Built by [**Kenpath Technologies**](https://kenpath.ai) |
|
|
Trained on open datasets including SYSPIN, SPICOR, RASA, and IndicTTS |
|
|
Released under **Apache-2.0 License** |
|
|
|
|
|
> “svara-tts is how India sounds — open, expressive, and human.” |
