20 lines
979 B
JSON
20 lines
979 B
JSON
{
|
|
"id": "tts_local",
|
|
"name": "Speak (local, Kokoro)",
|
|
"category": "audio",
|
|
"description": "Text → speech WAV entirely on this Mac via Kokoro-82M on MLX (Apache 2.0, ungated, ~350MB first download). ~50 built-in voices spanning male/female US/UK = instant character voice templates. Faster than realtime, no API cost, no cloud.",
|
|
"accepts": [],
|
|
"produces": ["audio"],
|
|
"resources": "gpu",
|
|
"entry": "run.py",
|
|
"python": "venvs/kokoro/bin/python",
|
|
"params_schema": {
|
|
"type": "object",
|
|
"properties": {
|
|
"text": {"type": "string", "default": "", "description": "The line to speak"},
|
|
"voice": {"type": "string", "default": "af_heart", "description": "Kokoro voice id (af_/am_ = US female/male, bf_/bm_ = UK): af_heart, af_bella, af_nicole, am_adam, am_michael, am_onyx, bf_emma, bf_isabella, bm_george, bm_lewis, ..."},
|
|
"speed": {"type": "number", "default": 1.0, "minimum": 0.5, "maximum": 2.0, "description": "Speaking rate"}
|
|
}
|
|
}
|
|
}
|