Text to speech (Audio Speech)

POST /v1/audio/speech turns text into speech and returns the audio file directly (MP3 by default) — handy for order updates, voice notifications, narrated content and spoken support replies. The format matches the OpenAI speech API. Requests are relayed to the model service in their original format; available voices and formats vary by model and follow that model's official documentation.

Headers

Request body

Request example

curl https://open.tutujin.com/v1/audio/speech \
  -H "Authorization: Bearer $TUTUJIN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "tts-1",
    "input": "Your order has shipped and should arrive tomorrow.",
    "voice": "alloy"
  }' \
  --output notice.mp3

import os
from openai import OpenAI

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

with client.audio.speech.with_streaming_response.create(
    model="tts-1",
    voice="alloy",
    input="Your order has shipped and should arrive tomorrow.",
) as resp:
    resp.stream_to_file("notice.mp3")

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 })

const resp = await client.audio.speech.create({
  model: 'tts-1',
  voice: 'alloy',
  input: 'Your order has shipped and should arrive tomorrow.',
})
fs.writeFileSync('notice.mp3', Buffer.from(await resp.arrayBuffer()))

Response format

A successful call returns raw audio bytes, not JSON. The Content-Type response header gives the audio type, e.g. audio/mpeg for MP3. With cURL, add --output to save it to a file.

Errors come back as a JSON error body with an HTTP error status, so check the status code before treating the body as audio. See Errors and retries.

Notes

• This endpoint is relayed as-is and has not been verified separately on this site. When you switch to another vendor's speech model, confirm its voice values and output formats first.
• Split long text into sentences or paragraphs, synthesize each and join the audio in order; overly long text in one request returns an error.
• Synthesize fixed phrases such as "Your order has shipped" once and reuse the cached audio instead of requesting it every time.
• Billing follows the model's price list; the Usage page in the console is the record of what you were charged.

Models that support this endpoint