Can I use Audexum as an ElevenLabs or OpenAI TTS drop-in?

How does the OpenAI-compatible API work?

• Speech: POST /v1/audio/speech with input, voice and response_format. Speed accepts 0.25 to 4.0, and instructions (up to 300 characters) becomes the style direction.
• Transcription: POST /v1/audio/transcriptions with model=whisper-1, returning json, text, verbose_json, srt or vtt.
• Models: model is accepted and ignored. OpenAI voice names such as alloy and nova map to Audexum V3 voices. Output formats are mp3, opus and wav.

Send the key as Authorization: Bearer sk_live_....

How does the ElevenLabs-compatible API work?

• POST /v1/text-to-speech/{voice_id} and /stream, plus GET /v1/voices.
• Send the key in the xi-api-key header. A Bearer header works too.
• Common ElevenLabs default voice ids map to an Audexum voice. Output format comes from the output_format query and defaults to mp3.

A step-by-step example is on Move from ElevenLabs.

What does not carry over?

Voices sound different, because Audexum does not host other providers' voices. Mapped voices are billed at the V3 rate of 2 credits per character.

Not supported: OpenAI translations and the Realtime API, and ElevenLabs cloning, speech-to-speech, sound effects and dubbing endpoints. The /stream path returns the whole file at once. The V4 engine is only available through the native API and Studio. See How do I get an API key and make my first call?