Nbility logoNbility Docs

Search documentation

Search guides and API reference content

Nbility provides OpenAI-compatible speech-to-text, audio translation, and text-to-speech endpoints. Available models, languages, voices, and audio formats depend on the live catalog and selected channel.

Transcription

POST /v1/audio/transcriptions

This endpoint uses multipart/form-data:

curl https://api.nbility.ai/v1/audio/transcriptions \
  -H "Authorization: Bearer $NBILITY_API_KEY" \
  -F "model=YOUR_TRANSCRIPTION_MODEL" \
  -F "file=@meeting.mp3" \
  -F "response_format=json"

Common fields include file, model, and—when supported—language, prompt, temperature, and response_format. A JSON response normally exposes the transcript in text.

Translation

POST /v1/audio/translations

The request is also multipart:

curl https://api.nbility.ai/v1/audio/translations \
  -H "Authorization: Bearer $NBILITY_API_KEY" \
  -F "model=YOUR_TRANSCRIPTION_MODEL" \
  -F "file=@speech.m4a"

The target language and optional fields depend on the selected model.

Text to speech

POST /v1/audio/speech
curl https://api.nbility.ai/v1/audio/speech \
  -H "Authorization: Bearer $NBILITY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "YOUR_TTS_MODEL",
    "voice": "YOUR_VOICE",
    "input": "Welcome to Nbility.",
    "response_format": "mp3"
  }' \
  --output speech.mp3

Common TTS fields:

FieldRequiredDescription
modelyesText-to-speech model ID
inputyesText to synthesize
voiceyesVoice ID from the model's supported set
instructionsnoStyle or delivery instruction, only for supported models
response_formatnoSuch as mp3 or wav; supported values are model-specific
speednoSpeaking speed; allowed range is model-specific

Production guidance

  • Enforce file size, duration, and MIME-type limits before forwarding user uploads.
  • Audio processing may take longer than text generation, so configure an appropriate overall timeout.
  • When TTS returns binary audio, do not parse it as JSON; inspect Content-Type as well.
  • Speech can contain personal data. Align storage, logs, and retention with your privacy policy.