MiniMax Voice Clone

Overview

This model takes an audio input and returns a voice_id that can be passed to the /v1/audio/speech endpoint with MiniMax HD or Turbo models. The cloning process is billed as a single flat call, after which the generated ID is used for subsequent text-to-speech generation. Requests route through an OpenAI-compatible base URL using a single platform API key. Every call receives automatic failover if a serving path degrades, and the response includes the exact cost in usage.cost alongside an X-Kyma-Model header identifying the executed model. The model does not support reasoning, vision, or structured outputs, and it operates strictly as an audio-to-audio utility. It has a medium speed tier and is classified in the premium cost tier. It is intended solely for voice profile creation, not for direct text-to-speech conversion.

Specs

Field	Value
Model ID	`minimax-voice-clone`
Best for	Brand voice cloning, character voices, custom narrator profiles
Context window	0
Input modalities	Audio
Output modalities	Audio
Tool calling	Yes
Structured outputs	Yes
Prompt caching	Yes
Speed	Medium
Cost band	Premium
Release stage	Stable

Pricing

	Per 1M tokens
Input	$0.20
Output	$0.60

Use this when

Brand Voice Profiles — Generate a consistent audio identity for corporate announcements or marketing content.
Custom Character Voices — Create distinct vocal profiles for interactive fiction or game NPCs.
Narrator Audio Setup — Build reusable voice IDs for long-form audiobook or podcast narration pipelines.
Personalized Audio Agents — Clone a specific speaker’s voice for conversational AI applications.

Not ideal for

Do not use this model for direct text-to-speech generation, as it only produces a reusable voice identifier rather than spoken audio.

Pick something else when

You need direct text-to-speech audio output: use minimax-speech-hd or minimax-speech-turbo.
You require multilingual voice synthesis: use eleven-multilingual-v2 or eleven-v3.
You want fast, low-latency speech generation: use eleven-turbo-v2-5 or eleven-flash-v2-5.
You need to generate background music or sound effects: use elevenlabs-music or elevenlabs-sfx.

Example

from openai import OpenAI

client = OpenAI(base_url="https://kymaapi.com/v1", api_key="ky-...")

response = client.chat.completions.create(
    model="minimax-voice-clone",
    messages=[{"role": "user", "content": "..."}],
)

Agent query example

Ask the API which models fit, instead of hardcoding an id:

curl "https://kymaapi.com/v1/models?recommended_for=generate-speech&tools=true&quality_tier=frontier-open"

FAQ

What happens after I clone a voice? The API returns a voice_id that you can pass to the /v1/audio/speech endpoint with MiniMax HD or Turbo models to generate spoken audio. How is the cloning process billed? Voice cloning is charged as a single flat call per reference recording, separate from the downstream speech model rates. Does Kyma cache the cloning requests? Yes, prompt caching is supported, and repeated prompt prefixes are billed at this model’s cached input rate.

​Overview

​Specs

​Pricing

​Use this when

​Not ideal for

​Pick something else when

​Example

​Agent query example

​FAQ