Voice / Piper API

Method	Path	Purpose
`GET`	`/api/v1/voice/config`	Wake word + engine selection
`PATCH`	`/api/v1/voice/config`	Update safe fields
`POST`	`/api/v1/voice/stt`	`multipart` audio → transcript JSON
`POST`	`/api/v1/voice/tts`	`{ "text": "..." }` → audio stream
`GET`	`/api/v1/voice/intents`	Registered intents (debug)

⚠️ Warning: POST /stt handles raw microphone audio — disable on untrusted networks or require mutual TLS.

Troubleshooting

415 Unsupported Media Type
Wrong Content-Type / codec — send PCM/WAV as documented for your core version.