Polyvox - Accent Trainer

Record your voice, hear how it would sound in different accents, and mimick yourself.

Prerequisites

• Python 3.10+, Node.js, ffmpeg, uv
• OpenVoice: Clone myshell-ai/OpenVoice into backend/, place converter checkpoints in backend/checkpoints_v2/converter/
• AWS credentials for Amazon Polly TTS

Quick start

make sync && make install
make backend   # terminal 1
make frontend  # terminal 2

Open http://localhost:3000

OpenVoice Base Speakers (optional)

To use OpenVoice base speaker voices (e.g. en-default, en-us) in addition to Polly:

1. Ensure checkpoints_v2/base_speakers/ses/ contains .pth files (from the full OpenVoice V2 checkpoint).
2. Run make sync-melotts.