Skip to main content
MonoRelay exposes two audio endpoints that mirror the OpenAI Audio API. The transcriptions endpoint converts spoken audio into text in the original language. The translations endpoint transcribes audio and translates the result into English. Both endpoints accept multipart file uploads and route through MonoRelay’s standard provider resolution, so you can target any configured provider that supports audio processing. Supported audio formats and file size limits follow the upstream provider’s capabilities. For OpenAI’s Whisper, the accepted formats are flac, mp3, mp4, mpeg, mpga, m4a, ogg, wav, and webm, with a maximum file size of 25 MB.

​
POST /v1/audio/transcriptions

Transcribe an audio file into text in its original language.

​
Authentication

​
Request parameters

All parameters are submitted as multipart/form-data fields.
​
file
required
The audio file to transcribe. Uploaded as a multipart file field. The accepted formats and size limits depend on the upstream provider.
​
string
required
The transcription model to use (e.g. whisper-1). Accepts aliases and model@provider syntax.
​
string
The language of the audio in ISO-639-1 format (e.g. "en", "zh", "fr"). Providing this improves accuracy and speed. When omitted, the model attempts to auto-detect the language.
​
string
Optional text to guide the model’s style or continue from a previous audio segment. The prompt should match the audio language.
​
string
default:"json"
Format of the transcript output. Accepted values: "json", "text", "srt", "verbose_json", or "vtt". Support for each format depends on the upstream provider.
​
number
default:"0"
Sampling temperature between 0 and 1. Higher values produce more varied output. Set to 0 for deterministic transcription.

​
Example

Response when response_format is "json":

​
POST /v1/audio/translations

Transcribe an audio file and translate the result into English, regardless of the original spoken language.

​
Authentication

​
Request parameters

​
file
required
The audio file to transcribe and translate. Uploaded as a multipart file field.
​
string
required
The model to use for translation (e.g. whisper-1).

​
Example

Response:
The translations endpoint always outputs English text. If you need a transcript in the original language, use /v1/audio/transcriptions instead.