MiniMax Voice Design

MiniMax Voice Design is the prompt-to-voice companion to Voice Cloning: instead of needing a reference recording, you describe the voice you want in natural language ('young female English speaker, slightly raspy, calm pacing') and the model synthesises a voice profile matching that description. Use when you need a custom-sounding voice but don't have rights-cleared reference audio. Pair with the standard TTS endpoint to drive the synthesised voice with any text.

Constraints & limits

The gateway auto-clamps requests to the model's limits where possible.

Quick start

This sample hits the text-to-speech endpoint. Change base_url and your key, then run it.

from openai import OpenAI

client = OpenAI(
    api_key="tl-xxxxxxxxxxxxxxxxxxxxxxxx",
    base_url="https://api.thalam.ai/v1",
)

response = client.audio.speech.create(
    model="minimax/voice-design",
    input="Welcome to Thalam, your unified gateway to AI models.",
)

with open("out.mp3", "wb") as f:
    f.write(response.content)