MiniMax Speech 2.8

MiniMax Speech 2.8 (minimax/speech-2.8-hd and minimax/speech-2.8-turbo) is text-to-speech, Russian included. It differs from the fixed-roster engines: voice accepts any MiniMax voice identifier, which is why the catalog shows an empty voice list for these models — that is a hint to bring your own, not an absence of choice.

Version	Key	When to pick it
HD	minimax/speech-2.8-hd	Voice-over where quality is audible: videos, audiobooks
Turbo	minimax/speech-2.8-turbo	Bulk narration and interactive use, where speed and price win

Parameters

POST /v1/audio/speech:

Field	Description
input	Text, required
model	Model key
voice	MiniMax voice identifier
speed	Speech rate

The response is mp3. Billing is per input character, so the cost of a voice-over follows from the length of the text before you send it.

Example

Python

from openai import OpenAI

client = OpenAI(base_url="https://api.mixen.ai/v1", api_key=MIXEN_API_KEY)

speech = client.audio.speech.create(
    model="minimax/speech-2.8-hd",
    input="Hello there! This is a speech synthesis sample.",
    voice="female-shaonv",
)
speech.write_to_file("hello.mp3")

cURL

curl https://api.mixen.ai/v1/audio/speech \
  -H "Authorization: Bearer $MIXEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "minimax/speech-2.8-hd", "input": "Hello there!", "voice": "female-shaonv"}' \
  --output hello.mp3

Things to keep in mind

• Voices differ per model. grok-voice-tts-1.0, Gemini TTS and Qwen have closed rosters listed in the catalog’s voices field, and an invalid voice there returns 400. MiniMax’s roster is open — check the provider’s documentation.
• Price follows input, not duration. A long text with pauses and short audible output costs the same as a dense one: the tariff counts characters.

The full synthesis line-up with prices is in the catalog; the endpoint itself is covered in Speech & embeddings.