The ElevenLabs-compatible API: point your client at us

"Drop-in compatible" gets thrown around a lot. Here's what it means at SecretVoice, concretely: the public surface mirrors the ElevenLabs protocol: same xi-api-key auth header, same /v1/text-to-speech/{voice_id} shape, same model aliases accepted (eleven_multilingual_v2, eleven_flash_v2_5, eleven_turbo_v2_5 all map to secret-v1).

So the migration is a base-URL change:

# before: https://api.elevenlabs.io/v1/text-to-speech/velvet
# after:  https://api.secretvoice.dev/v1/text-to-speech/velvet

Same header. Same payload. Same audio back.

▶ Listen to Titan

What you get

• 29 voices, 21 languages: the same voice keeps its character in every language.
• Three delivery modes: synchronous files, HTTP streaming, and realtime WebSocket for live agents.
• Transparent pricing: 1 credit per character, published per-plan rates, no surprise seats.
• Zero-retention option: audio is never stored; transcripts are kept in history unless you pass enable_logging=false, then nothing is retained.

If your project runs on ElevenLabs-compatible clients, and most voice-agent frameworks do, you can benchmark SecretVoice in an afternoon without touching your code. Swap the base URL, run the same prompts, and hear whether the voice fits your product.

▶ Listen to Professor

The API reference is at secretvoice.dev/docs, the machine-readable spec at openapi.json, and agents can read the full surface at llms-full.txt.