Audio
Converts text into speech. The endpoint is OpenAI-compatible, and the response is the binary audio body, exactly like OpenAI's /v1/audio/speech.
POST
https://console.aimodelhub.ai/v1/audio/speech
| Field | Type | Notes |
|---|---|---|
model | string | Required. A published audio model id (see below). |
input | string | Required. The literal text to speak. |
voice | string | Optional. One of the model's voices; defaults to alloy. |
response_format | string | mp3 (default), opus, aac, flac, wav. |
Response
The response body is the raw audio (e.g. Content-Type: audio/mpeg). The same file is also saved to the library. Each successful call is metered once at list price, and failed calls are never billed.
Why brand is ignored here
A brand field on the audio call is accepted but ignored, and it does NOT return 404. TTS input is the literal spoken text. Running it through a brand rewriter would change the requested words, which is a correctness error, not a style choice. Brand-driven voice selection is a separate, future mechanism.
Errors
The full status/type table is on the shared errors reference. This endpoint has no in-flight cap and no capability gate. Its most likely error is: 400 invalid_request_error means the input field is missing or empty.
Available audio models
| Model | Reference |
|---|---|
| Models are being published and will appear here in a few minutes. | |