elevenlabs

1# ElevenLabs
2 
3## Purpose
4Work with the user's ElevenLabs account: subscription tier and character usage (`me`), the voices available on the account (`voices`), and text-to-speech generation (`text-to-speech`). TTS consumes characters from the account quota (real money on paid plans): it requires the exact-match `--confirm` string on every generation, stating the voice and character count.
5 
6## Tooling
7All commands go through `bin/elevenlabs.py`:
8 
9```bash
10bin/elevenlabs.py me        # subscription tier, character_count, character_limit
11bin/elevenlabs.py voices    # available voices (name, category)
12 
13# TTS: every generation requires the exact --confirm string the CLI prints
14bin/elevenlabs.py text-to-speech --text "Hello there" --voice-name Rachel --out /tmp/line.mp3
15bin/elevenlabs.py text-to-speech --text "Hello there" --voice-id VOICE_ID --out /tmp/line.mp3
16bin/elevenlabs.py text-to-speech --text "Hello" --voice-name Rachel --out /tmp/line.mp3 --voice-settings '{"stability":0.5,"similarity_boost":0.75}'
17```
18 
19`--voice-id` wins over `--voice-name`. Audio saves to `--out` (MP3). Default model is `eleven_multilingual_v2`.
20 
21## Auth
22- Provider id: `elevenlabs` (credential is collected as `custom.elevenlabs`)
23- Collection: API key via the secure credential flow (`credentials.request_api_access`): create one at elevenlabs.io/app/settings/api-keys
24- Connect placement: `custom_header:xi-api-key` (ElevenLabs uses `xi-api-key`, not `Authorization: Bearer`)
25…