API Documentation: Whisper Audio Transcription 🎙️

This document specifies the API for transcribing audio files using the hosted Whisper large-v3-turbo model, following the OpenAI-compatible schema. This "Turbo" variant is a weakly supervised pre-trained model featuring a 4-layer decoder, offering 8.8x faster inference than the standard large-v3 while maintaining near-identical accuracy.

---

Endpoint

Method	URL	Summary
POST	/v1/audio/transcriptions	Transcribe audio/video using the Whisper model.

---

Authentication

The API utilizes Bearer Token authentication. A valid API_KEY must be included in the header for all requests.

Header	Example	Description
Authorization	Bearer YOUR_API_KEY	Your server-provided API Key.
x-request-id	UUID_string	Optional. Unique identifier for tracking. Generated by server if omitted.

---

Request Parameters

Requests must be sent as multipart/form-data.

Parameter	Type	Required	Description
file	File / URL	Yes	The audio file object or a direct URL to an audio/video file (mp3, mp4, mpeg, mpga, m4a, wav, webm, aac).
model	string	Yes	Use "openai/whisper-large-v3-turbo" for high-speed ASR.
response_format	string	No	Formats: json, text, srt, vtt, verbose_json. Default: json.
temperature	number	No	0.0 to 1.0. Controls randomness. Higher values increase variability. Default: 0.0.
language	string	No	ISO-639-1 code (e.g., en, zh, ja) to improve transcription accuracy.
prompt	string	No	Optional text to guide the model's style or vocabulary.
timestamp_granularities	array	No	Only for verbose_json. Can include word or segment.