Calling an endpoint when using a FunASR model catalog

After creating an endpoint by using a FunASR model catalog, you can call the endpoint’s OpenAI-compatible API or FunASR native API through the model gateway.

Prerequisites

You have obtained the API URL of the endpoint. For details, see Getting the API URL.
You have created an API key that can call the endpoint. For details, see Creating an API key.

Precaution

All calls access the endpoint through the model gateway. The address format is http://<gateway>/workspace/<workspace>/endpoint/<endpoint>.
The request header must include Authorization: Bearer <api_key>, where <api_key> is the API key.
When creating an endpoint by using a FunASR model catalog, the model registry and model name are still required, but they do not take effect for the FunASR image. You can select any valid values to pass validation. The selected values do not affect the deployment result.
When deploying FunASR on Turing-architecture GPUs such as NVIDIA T4, set dtype=float32 in the engine variables. The corresponding startup parameter is --dtype float32. The default half precision cannot be used on this GPU architecture.

Calling the OpenAI-compatible API

Run the following commands to initiate a speech recognition request through the OpenAI-compatible API:

curl -sL -o en.mp3 https://huggingface.co/FunAudioLLM/Fun-ASR-Nano-2512/blob/main/example/en.mp3
curl -X POST <API_URL>/v1/audio/transcriptions \
  -H "Authorization: Bearer <api_key>" \
  -F file=@en.mp3

Calling the FunASR native API

Run the following commands to initiate a speech recognition request through the FunASR native API:

curl -sL -o en.mp3 https://huggingface.co/FunAudioLLM/Fun-ASR-Nano-2512/blob/main/example/en.mp3
curl -X POST <API_URL>/asr \
  -H "Authorization: Bearer <api_key>" \
  -F file=@en.mp3

After the request succeeds, the API returns the speech recognition result with segments, timestamps, and real-time factor (RTF).