Pictures, transcription and voice with your ORYKSA key

What this is for

The flows that asked OpenAI for a picture, a transcription or a voice can ask ORYKSA instead, with the same ORYKSA key and the interactions of your plan. In n8n, Make or Zapier you only change the address to https://api.oryksa.com/v1 and the key.

What it costs

• A picture: 10 interactions.
• Transcription: 1 interaction per minute of audio.
• Voice: 1 interaction per 500 characters of text.

Nothing is charged when the request fails. If your plan does not have enough interactions for the whole request, it is refused before anything is created.

Pictures

POST /v1/images/generations with prompt and size: 1024x1024, 1024x1536, 1536x1024, 1024x1792, 1792x1024, 512x512 or 256x256. The picture comes in the exact size you asked for, as an address; ask for response_format: b64_json to receive the file itself. One picture per request.

Transcription

POST /v1/audio/transcriptions with the audio in file (mp3, mp4, m4a, wav, webm or ogg; up to 25 MB and 30 minutes). response_format: verbose_json also gives the time of every word, for subtitles.

Voice

POST /v1/audio/speech with input (up to 4000 characters). It speaks with the voice of your account, whatever voice name the flow sends. The reply is an mp3. With timestamps: true you also get the time of every word.

Good to know

• The usage statement shows these requests like the others, with the name of the workflow when the flow says it.
• Video is not available.