Developers — Omeife AI

API reference

Every endpoint, straight from the contract.

GET/voices

List available TTS voices

Returns all voices currently enabled on the platform. Filter and group client-side by `language`.

GET /voices → 200json

{
  "voices": [
    {
      "voice_id": "ig-female-01",
      "language": "igbo",
      "gender": "female",
      "name": "Adaeze"
    },
    {
      "voice_id": "en-ng-male-01",
      "language": "nigerian_english",
      "gender": "male",
      "name": "Chidi"
    }
  ]
}

Errors:unauthorizedrate_limitedinternal_error

POST/tts

Convert text to speech

Generates audio from text. If `voice_id` is omitted, the platform detects the text language and assigns a default voice; the response then includes `detected_language`. Generated audio URLs are valid for 24 hours.

Field	Type	Required
text	string	required
voice_id	string	optional
output_format	mp3 \| wav \| pcm	optional

POST /tts → 200json

{
  "audio_url": "https://cdn.omeife.ai/audio/abc123.mp3",
  "voice_id": "ig-female-01",
  "output_format": "mp3",
  "request_id": "req_01J9X4YB4T"
}

Errors:invalid_inputunauthorizedunsupported_languagerate_limitedinternal_error

POST/asr

Transcribe audio (automatic language detection)

Transcribes an audio clip, automatically detecting the spoken language. Accepts wav, mp3, or m4a up to 60 seconds and 10 MB.

Field	Type	Required
audio_file	string	required

POST /asr → 200json

{
  "transcription": "Ndewo, kedu ka i mere?",
  "detected_language": "igbo",
  "request_id": "req_01J9X512M3"
}

Errors:invalid_inputunauthorizedpayload_too_largerate_limitedinternal_error

POST/asr/{language}

Transcribe audio with a forced language model

Forces transcription with a specific language model — no automatic detection. Use when the spoken language is known; accuracy is higher and latency slightly lower than auto mode.

Field	Type	Required
language	path param	required
audio_file	string	required

POST /asr/{language} → 200json

{
  "transcription": "E kaaro, se daadaa ni?",
  "language": "yoruba",
  "request_id": "req_01J9X52T9Q"
}

Errors:invalid_inputunauthorizedpayload_too_largeunsupported_languagerate_limitedinternal_error

POST/language

Detect the language of spoken audio

Field	Type	Required
audio_file	string	required

POST /language → 200json

{
  "detected_language": "hausa",
  "confidence": 0.94,
  "request_id": "req_01J9X53Y1V"
}

Errors:invalid_inputunauthorizedpayload_too_largerate_limitedinternal_error

POST/translate

Translate text between supported languages

Field	Type	Required
text	string	required
source_language	string	required
target_language	string	required

POST /translate → 200json

{
  "translated_text": "E kaaro",
  "source_language": "english",
  "target_language": "yoruba",
  "request_id": "req_01J9X54Q7D"
}

Errors:invalid_inputunauthorizedunsupported_languagerate_limitedinternal_error

Auth & limits

Keys, headers, and caps.

Authentication

Every request needs Authorization: Bearer omf_live_… (or omf_test_… in the console for evaluation). Create and revoke keys from the customer console — never share a key or embed one in browser code.

Rate limits

Every response carries X-RateLimit-Limit, X-RateLimit-Remaining, and X-RateLimit-Reset. A 429 includes Retry-After. Limits are per key, set by your plan.

Input caps

Text: 500 characters per request. Audio: wav, mp3, or m4a, up to 60 seconds and 10 MB. Requests over either limit return invalid_input or payload_too_large before anything is processed.

One REST call away.