voice — Speech in a voice — openbash
Text spoken in a cloned or stock voice, as an MP3 filed under a URL; and voices cloned from the account's own clips.
voice
voice is a command on the openbash shell, reachable over HTTP, SSH and MCP with one key. It composes with the other programs through pipes; there is no per-program endpoint and no schema for the join.
voice say b067832732f2439fae364829ff6b6b58 "Then it all sped up." /tmp/line.mp3 --publish spoke 20 bytes of text to /tmp/line.mp3 (22 KB) url: https://…/audio/3f1c…-line.mp3?…
help
voice: speech in a voice. USAGE voice say VOICE_ID TEXT OUT.mp3 [--speed X] [--temperature T] [--publish] voice say VOICE_ID --file TEXT.txt OUT.mp3 [...] voice list voice clone TITLE CLIP [CLIP…] [--description D] [--enhance] SAY Speaks TEXT in the voice and writes the audio to OUT; the format follows OUT's extension, .mp3, .wav or .opus. VOICE_ID `-` is the stock voice. --speed is 0.5 to 2, --temperature 0 to 1 (lower is steadier). --publish files the take in the account's store and prints `url: …`: that is how it leaves the session. A published URL is presigned and expires in seven days. Write numbers the way they should be SAID: "nineteen twenty-eight", not "1928", if the year must not come out as a count. LIST The account's voices: voice_id, title and state. A voice is usable once its state is `trained`, usually within a minute of cloning. CLONE Makes a private voice from 1 to 20 clips the session can read — one person speaking, 10 to 30 seconds each, no music under it — and prints its voice_id. --enhance asks for the clips to be cleaned first. Clone only a voice whose owner has agreed to it. COSTS `say` is billed as fish_tts, by the UTF-8 bytes of the text. `list` and `clone` are free.