# loqui-cli
The `loqui` command: serve the OpenAI audio API, speak and transcribe in
process, manage tokens, and check what an exposure would allow.
cargo install loqui-cli # needs cmake + a C++ compiler for Whisper
loqui serve # Unix socket, this user only
loqui serve --listen loopback:8100
loqui doctor --listen all:8100
loqui speak "Hello." -o hello.wav
loqui transcribe hello.wav
`--no-default-features` builds text-to-speech only, with no C++ toolchain.
`--features cuda` runs on an NVIDIA GPU; `--features mp3` adds MP3 output
through LAME (LGPL).
Part of [loqui](https://github.com/Awakened-Labs/loqui). MIT; see `NOTICE`
for third-party attributions.