Speech to text (Audio Transcriptions)

POST /v1/audio/transcriptions takes an audio file and returns the recognized text, either plain or as timed SRT or VTT subtitles. Common uses are meeting notes, call-center quality checks and video subtitles. The file is uploaded as multipart/form-data in the same format as the OpenAI transcription API, and the request is relayed to the model service as-is.

POSThttps://open.tutujin.com/v1/audio/transcriptions

Headers

Header	Required	Description
`Authorization`	Yes	`Bearer sk-...`, an API key from the console
`Content-Type`	Yes	`multipart/form-data`. SDKs and `curl -F` set it for you; do not write it by hand

Request body

Parameter	Type	Required	Description
`file`	file	Yes	Audio file, usually mp3, mp4, mpeg, mpga, m4a, wav or webm. Each file has a size limit, so split long recordings
`model`	string	Yes	Model name, e.g. `whisper-1`
`language`	string	No	Audio language as an ISO-639-1 code such as `zh` or `en`. Setting it usually improves accuracy and speed
`prompt`	string	No	Hint text: product names, people's names or the end of the previous segment, to keep wording consistent
`response_format`	string	No	`json` (default), `text`, `srt`, `vtt`, `verbose_json`
`temperature`	number	No	0 to 1, default 0. Usually leave it at the default
`timestamp_granularities[]`	array	No	`segment` or `word`, used with `verbose_json` to return segment or word timestamps

Request example

curl https://open.tutujin.com/v1/audio/transcriptions \
  -H "Authorization: Bearer $TUTUJIN_API_KEY" \
  -F file=@meeting.mp3 \
  -F model="whisper-1" \
  -F language=en \
  -F prompt="Tutujin, quarterly review, customer success"

import os
from openai import OpenAI

client = OpenAI(base_url="https://open.tutujin.com/v1", api_key=os.environ["TUTUJIN_API_KEY"], timeout=300)

with open("meeting.mp3", "rb") as f:
    result = client.audio.transcriptions.create(
        model="whisper-1",
        file=f,
        language="en",
        prompt="Tutujin, quarterly review, customer success",
    )
print(result.text)

import fs from 'node:fs'
import OpenAI from 'openai'

const client = new OpenAI({ baseURL: 'https://open.tutujin.com/v1', apiKey: process.env.TUTUJIN_API_KEY, timeout: 300_000 })

const result = await client.audio.transcriptions.create({
  model: 'whisper-1',
  file: fs.createReadStream('meeting.mp3'),
  language: 'en',
  prompt: 'Tutujin, quarterly review, customer success',
})
console.log(result.text)

Response example

With response_format set to json:

JSON

{
  "text": "Let's start with this quarter's customer success metrics..."
}

With srt you get subtitle text you can save straight to an .srt file:

Text

1
00:00:00,000 --> 00:00:03,200
Let's start with this quarter's customer success metrics

2
00:00:03,200 --> 00:00:06,900
Overall renewals are up on last quarter

Field	Description
`text`	The full recognized text
`segments`	With `verbose_json`: each segment with `start`, `end` (seconds) and text
`language`	With `verbose_json`: the language the model detected

Notes

This endpoint is relayed as-is and has not been verified separately on this site; test it with a short clip first.
File extensions must match the real format. Oversized files may be rejected (413); split long recordings into 5 to 10 minute pieces and join the results in order.
Pass language when you know it; for mixed-language recordings, leave it out and let the model decide.
If product or people's names come out wrong, put them in prompt.
Transcription time grows with audio length, so give the client a generous timeout. Billing follows the model's price list; the Usage page in the console is the record.

Models that support this endpoint

Model	Vendor	Cheapest group	Input ¥/1M	Output ¥/1M	Per request ¥
whisper-1	OpenAI	Standard	¥60	¥60	—

Prices in CNY, cheapest group for each model; updated Nov 12, 2026, 8:17 PM.

Still stuck? Add our WeChat support JsnonoChat and send the request id from the error message, and we will look into it.