JARVIS / VOICE SYNTHESIS CORE / KOKORO-82M

Engineer a voice.

Shape, stream and analyse neural speech from your private inference stack—with live telemetry down to first audio.

Unraid API address

Enter your Unraid address, then test the connection.

Compose a test

What should the voice say?

Text to speak87 characters

Voice

8 voices available. Press Test connection to refresh the full list from FastKoko.

Speed1.00×

Language is selected by the voice prefix: af/am US English, bf/bm British English.

PCM sample rate