Local Voice Studio / SvaraCraft Voice

The need

Academic narration and repeated speech production otherwise require recording every revision or buying hosted speech generation.

Why it was developed

Provide a reusable local narration workflow, local preset voices, optional consent-based cloning, classroom projects and the integrated NavaFrame video backend.

How to use it

1. Install Python 3.10–3.14 and FFmpeg/FFprobe. Create a virtual environment with python -m venv .venv and activate it.
2. Run python -m pip install . from this repository. For optional Chatterbox cloning, install the voice extra only after checking the model requirements.
3. Run local-voice-studio --host 127.0.0.1 --port 7860 --data ./data --open-browser.
4. Use preset narration for a first text passage; listen to the result before exporting. Configure local models using LOCAL-MODELS.md. Voice cloning needs suitable hardware and consented reference audio.

How it can save time or money

Reuses one script and voice configuration across revisions; local inference avoids mandatory per-character cloud inference fees. NavaFrame can assemble locally generated narration into lesson media without buying a separate rendering service.

These are potential benefits, not measured savings. Include setup, maintenance, review and local operating costs when comparing your own workflow.

Current scope & limits

Model downloads, hardware, electricity and review time remain necessary. Hosted signup, mail, subscriptions and GPU services are separate integrations. Original application source is proprietary; third-party model licences remain applicable. NavaFrame is part of this repository, not a duplicate standalone upload.