himahande45's picture
Switch Space to VM-backed frontend
0e9e909 verified

A newer version of the Gradio SDK is available: 6.13.0

Upgrade
metadata
title: 'IndicVox: Hindi & Tamil Code-Switching TTS'
emoji: 🎙️
colorFrom: indigo
colorTo: green
sdk: gradio
sdk_version: 6.12.0
app_file: frontend_app.py
pinned: false
python_version: 3.10.16
suggested_hardware: cpu-basic

IndicVox

IndicVox is a research demo for multilingual text-to-speech across Hindi, Tamil, and code-switched prompts. The Space hosts the frontend UI, while inference runs on an external GPU VM backend.

What it includes

  • Hindi Focus for Hindi and Hindi-English prompts
  • Tamil Focus for Tamil and Tamil-English prompts
  • Research Baseline for direct comparison against the untuned multilingual model
  • Built-in research voice presets for fast demo playback
  • Zero-shot Text Only mode if you want to skip reference conditioning
  • VM-backed inference on a dedicated GPU server

Usage

  1. Pick a model profile.
  2. Type a Hindi, Tamil, or code-switched prompt.
  3. Pick a built-in voice preset or Text Only.
  4. Click Generate Speech.

Notes

  • The frontend expects INDICVOX_API_URL to point at the VM backend.
  • If the backend is token-protected, set INDICVOX_BACKEND_TOKEN in Space secrets too.
  • The Space is meant for inference/demo usage, not batch evaluation.