Engines

`GET /v1/engines`

Lists every engine voxd knows about, including ones that can't run on this machine.

Response 200: an array of:

Field	Type	Description
`id`	string	Pass this as `engine` elsewhere, e.g. `"system"`
`name`	string	Display name
`capabilities`	string[]	What it can do: `asr` (transcribes speech), `tts` (speaks text), `clone` (speaks in custom voices), `emotion` (honours `emotion` in speech requests)
`license`	string	License of the underlying model or tool
`available`	boolean	`true` if usable right now
`unavailable_reason`	string \| null	Why not, e.g. `"Install espeak-ng to enable system voices"`

curl -s "$VOXD/v1/engines" -H "Authorization: Bearer $VOXD_TOKEN"
# [{"id":"system","name":"System Voices","capabilities":["tts"],"license":"Built into your OS","available":true,"unavailable_reason":null},
#  {"id":"kokoro","name":"Kokoro","capabilities":["tts"],"license":"Apache-2.0","available":false,"unavailable_reason":"Download Kokoro from Models to use it"}]

Available engines

id	Platforms	Notes
`system`	macOS (built-in voices), Linux (`espeak-ng`)	Instant, no download. Windows support is planned.
`kokoro`	macOS, Windows, Linux (CPU)	54 natural voices in 9 languages. Needs the `kokoro-v1` model (354 MB). Until it's installed, `unavailable_reason` says so.
`chatterbox`	macOS, Windows, Linux (Apple GPU, NVIDIA or CPU)	Voice cloning with emotion control (English). Needs the `chatterbox-v1` model (3.2 GB) plus its engine runtime (about 1.3 GB), which is installed in its own isolated environment and runs as a separate process.
`whisper`	macOS, Windows, Linux (CPU; NVIDIA GPU via CUDA)	Speech recognition in ~99 languages. Needs a Whisper model and its runtime (about 260 MB). See Transcription.