Write lines, give them voices and listen. The page runs here; the speech is made by the vox engine on your computer.
Start the engine and open the link it prints, or paste its address and secret above:
git clone https://github.com/dariumi/Vox && cd Vox && pip install . vox web --models ./models --voices ./voices
From another computer, start it with --host 0.0.0.0 --tls: an https page can only reach plain http on the same computer.
Takes appear here, newest at the bottom. Click a waveform to play from that spot.
No voices for this model yet: record one in Voices.
A profile keeps several voices in one file. A voice has styles (emotions): a recording of how it should sound, an instruction, or both. One profile works with every model.
A model opens on first use. "Fit to this computer" tries the CPU and every graphics card and keeps the fastest; after that "auto" uses it.
Everything spoken through this engine, newest first. The clips stay on the engine's computer.