Audio Transcriptions

Use the GLM-ASR-2512 model to transcribe audio files into text, supporting multiple languages and real-time streaming transcription.

POST /paas/v4/audio/transcriptions

Speech to Text

cURL

curl --request POST \
  --url https://api.z.ai/api/paas/v4/audio/transcriptions \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: multipart/form-data' \
  --form model=glm-asr-2512 \
  --form stream=false \
  --form file='@example-file'

200

{
  "id": "<string>",
  "created": 123,
  "request_id": "<string>",
  "model": "<string>",
  "text": "<string>"
}

Authorizations

Authorization · string · header · required

Use the following format for authentication: Bearer

Body

multipart/form-data

file · file · required

The audio file to be transcribed. Supported audio file formats: .wav / .mp3. Specifications: file size ≤ 25 MB, audio duration ≤ 30 seconds.

model · enum<string> · default: glm-asr-2512 · required

The model ID to invoke.

Available options: glm-asr-2512

file_base64 · string

Base64 encoded audio file. Only one of file_base64 or file needs to be provided (if both are provided, file takes precedence).

prompt · string

In long text scenarios, you can provide previous transcription results as context. Recommended to be less than 8000 characters.

hotwords · string[]

Hotword list to improve recognition accuracy for domain-specific vocabulary. Format example: ["person_name","place_name"]. Recommended not to exceed 100 items.

Maximum array length: 100

stream · boolean · default: false

This parameter should be set to false or omitted when using synchronous calls. It indicates that the model returns all content at once after generating all content. Default is false. If set to true, the model will return generated content in chunks via standard Event Stream. When the Event Stream ends, a data: [DONE] message will be returned.

request_id · string

Passed by the user side, needs to be unique; used to distinguish each request, 6–64 characters. If not provided by the user side, the platform will generate one by default.

Required string length: 6 - 64

user_id · string

Unique ID for the end user, 6–128 characters. Avoid using sensitive information.

Required string length: 6 - 128

Response

200

application/json

Request processed successfully

id · string

Task ID

created · integer<int64>

Request creation time, as a Unix timestamp in seconds.

request_id · string

Passed by the client, must be unique. A unique identifier to distinguish each request. If not provided by the client, the platform will generate one by default.

model · string

Model name

text · string

The complete transcribed content of the audio.