Kokoro (local)
Kokoro (local) is one of the components that give a voice agent its voice: it turns the agent’s answers into audio for the caller to hear. It runs on the NappAI server: no credential, the text is not sent to any outside provider and there is no per-word cost. It has Spanish and English voices.
It plugs into the TTS connector of the Voice Agent node. It holds the voice settings the agent uses on every call.
When to use it
- You want a local voice with a warmer tone than Piper (local).
- You do not want the agent’s answers to leave the NappAI server.
- Your calls are in Spanish or English.
Configuration
| Field | What it does |
|---|---|
| Voice | The voice. Those starting with ef_ and em_ are Spanish (female and male) and those starting with af_ and am_ are English. The list offers ef_dora (default), em_alex, em_santa and af_bella. |
| Speed | Speaking pace. 1.0 is the voice’s natural pace; higher is faster and lower is slower. Default: 1.0. |
Output
The TTS output holds the voice settings. It connects to the TTS connector of the Voice Agent node, or to Fallback TTS when you turn on Fallback Engines and want it as a backup. The editor will not let you plug it into an STT connector by mistake.
Example
A neighbourhood shop wants an agent that takes phone orders without hiring any voice service.
- Add Kokoro (local) and connect it to the TTS connector of the Voice Agent.
- Leave Voice as
ef_dora. - Use a local STT such as Parakeet (local, multilingual).
- Test the agent in the playground and listen to how it sounds.
Good practices
- If the agent’s first sentence takes long to sound, try Pocket TTS (local), which starts speaking sooner.
- The differences between local and cloud voices are explained in Quality, noise and latency.
Related components
- Voice Agent: the node this component connects to.
- Supertonic (local): the recommended local voice, no credential needed.
- Pocket TTS (local): the local voice that starts speaking soonest, Spanish only.