polyglot-me / README.md
praxelhq's picture
Add X social post link to README
4974f13 verified
|
Raw
History Blame Contribute Delete
1.83 kB

A newer version of the Gradio SDK is available: 6.20.0

Upgrade
metadata
title: Polyglot Me
emoji: 🎙️
colorFrom: indigo
colorTo: pink
sdk: gradio
app_file: app.py
pinned: true
license: mit
short_description: Hear your own voice in English, Hindi, Telugu & Tamil.
tags:
  - gradio
  - build-small-hackathon
  - voice-cloning
  - tts
  - multilingual
  - indic
  - track:wood
  - sponsor:openbmb
  - sponsor:modal
  - achievement:offbrand
  - achievement:fieldnotes

🎙️ Polyglot Me

Record ten seconds of your voice. Hear yourself say one line in English, Hindi, Telugu, and Tamil.

Plenty of tools can speak Tamil. Almost none can speak Tamil in your voice — and then Hindi, and Telugu, and English, still recognisably you. Holding one speaker's identity across four scripts is the hard part, and it's the point. ElevenLabs and Cartesia don't do it.

Open Polyglot Me · 🎬 Watch the 1-min demo · 📣 Post on X · first run ~2–4 min on Modal.

What it does

Record or upload ~10 seconds, type one English line, and get four clips back in your own voice — plus a share card and a caption you can post.

How it works

Sarvam-Translate turns the line into Hindi, Telugu, and Tamil; VoxCPM2 clones your voice for each. The Space is CPU-only; the GPU work is a serverless Modal app.

flowchart LR
  V[~10s of you] --> X[VoxCPM2 clone]
  L[one English line] --> S[Sarvam: hi · te · ta] --> X
  X --> O[🔊 English · Hindi · Telugu · Tamil + share card]

Built by @praxelhq for the HuggingFace Build Small hackathon — Thousand Token Wood. VoxCPM2 (OpenBMB) and Sarvam, on Modal. Build notes: FIELD_NOTES.md.