API docs by Redocly

さくらの AI Engine Audio Transcriptions API (1.0.0)

Download OpenAPI specification:Download

さくらの AI Engine 音声書き起こし

Create chat completion

チャット補完のリクエストの代表例です、記載のパラメータ以外にも指定可能です。

Authorizations:
BearerAuth
Request Body schema: application/json
model
required
string
Enum: "gpt-oss-120b" "Qwen3-Coder-480B-A35B-Instruct-FP8" "llm-jp-3.1-8x13b-instruct4"

利用するチャットモデル名。利用可能なモデルはコントロールパネル等をご確認ください。

messages
required
Array of arrays
stream
boolean
Default: false

Request samples

Content type
application/json
{
  • "model": "gpt-oss-120b",
  • "messages": [ ],
  • "stream": false
}

Create a transcription

Authorizations:
BearerAuth
Request Body schema: multipart/form-data
file
required
string <binary>

Audio file to transcribe. Common formats: mp3, mp4, m4a, wav, webm, etc. .

model
string
Value: "whisper-large-v3-turbo"

Transcription model identifier served by vLLM.

language
string
Default: "ja"

Source language hint (BCP-47, e.g. "ja", "en-US").

prompt
string

Optional decoding/prompt bias (proper nouns, style hints).

temperature
number [ 0 .. 1 ]
Default: 0

Decoding temperature.

stream
boolean
Default: false

Responses

Response samples

Content type
application/json
{
  • "model": "whisper-large-v3-turbo",
  • "text": "本日はご利用いただきありがとうございます。"
}