Create an IVR voice prompt
Type your script, choose a voice, and download a phone-ready recording. No studio, no voice actor.
Your script
Choose a voice
Generate
Voice delivery settings
Leave untouched to use the voice's defaults.
This format can't be played in a browser; download it to use on the phone system.
For developers
The service is a small JSON-over-HTTPS API. Every request except GET /health needs an x-api-key header. In this demo, the password you entered is your API key; production systems use their own key.
Endpoints
| Endpoint | What it does |
|---|---|
GET /voices | Voice catalog: id, name, gender, language(s), accent, and a preview_url MP3 sample you can play directly. |
GET /formats | Every supported output_format with file extension, content type and plan requirements. |
POST /tts | Generate audio. Returns a time-limited audio_url to download. Identical requests are served from cache. |
GET /health | Liveness check, no auth. |
Generate a prompt (curl)
Generate a prompt (PHP)
Request options
| Field | Notes |
|---|---|
text | Required. The script to speak, up to 1000 characters. |
voice_id | Required. Any id from GET /voices. |
model_id | Optional. eleven_multilingual_v2 (default, best quality), eleven_flash_v2_5 (about half the credit cost), eleven_turbo_v2_5, eleven_v3 (most languages). |
output_format | Optional. Default wav_8000 plays natively on the phone system. See GET /formats for MP3, Opus, higher-rate WAV and raw PCM options. |
voice_settings | Optional. stability, similarity_boost, style (each 0 to 1), speed (0.7 to 1.2), use_speaker_boost (boolean). Omit to use the voice's defaults. |
Errors
Errors come back as { "error": { "code", "message" } } with 400 (validation), 401 (bad key), 404 (unknown voice), 429 (rate limited upstream, retried once internally first) or 502 (upstream failure).
Requests run live against this environment with the key you signed in with.
The full API contract is published as an OpenAPI 3.1 document — the industry-standard format for handing over an API. Import it into Postman, Insomnia, Swagger UI or a code generator to get typed clients and try-it-out tooling for free.
The schema documents every endpoint, request field, response shape and error, including the voice-settings ranges and all output formats.