Examples
Browse voices, generate audio and retrieve history.
Inspect your account
elevenlabs me subscription --format pretty
elevenlabs model list
elevenlabs voice list --format jsonl
elevenlabs voice search narrator --language en --page-size 10Generate speech
Generation consumes account credits. Start with a short text file and a stock voice.
printf 'Hello.' > narration.txt
elevenlabs speech generate --text-file narration.txt \
--voice Xb7hH8MSUJpSbSDYk0k2 --output narration.mp3Generate dialogue
Save this as dialogue.json:
[
{"text":"Hello.","voice_id":"Xb7hH8MSUJpSbSDYk0k2"},
{"text":"Hi there.","voice_id":"onwK4e9ZLuTAKqWW03F9"}
]elevenlabs dialogue generate --input dialogue.json --output dialogue.mp3Speech and dialogue default to the captured eleven_v4_exp model. An unsupported model or plan can still be rejected by the service.
Find and download audio
elevenlabs history list --source TTS --search Hello
elevenlabs history show HISTORY_ID
elevenlabs history download HISTORY_ID --output saved.mp3Sound effects
Disable Explore sharing in the website first. The CLI refuses generation when sharing is enabled and never changes that setting for you.
printf 'A short wooden tap.' > prompt.txt
elevenlabs sound generate --text-file prompt.txt --duration 1 --variants 1
elevenlabs sound download SOUND_ID --output effect.opusAudio commands require an output destination. Existing files are never overwritten. Use --output - for binary stdout. A failed file download removes its incomplete output; bytes already written to stdout cannot be recalled.