Model positioning
Speech-02-Turbo is a text-to-speech model in the MiniMax Speech-02 series, optimized for real-time interaction and focused on balancing low latency with voice quality. The Speech-02 series was released on April 2, 2025. (minimax.io)
Capabilities and use cases
- Synthesizes input text into speech and supports streaming output; speech rate, volume, and pitch can be adjusted, and pauses and emotion can be controlled. (platform.minimaxi.com)
- Supports system voices and cloned voices. The series release announcement states that it supports more than 30 languages, including Mandarin, Cantonese, English, Japanese, and Korean; actual voice and language availability depends on the API and account configuration. (minimax.io)
- Suitable for voice chat, real-time assistants, game interactions, customer service, content voiceovers, and other scenarios. (minimax.io)
Limitations
- The request parameter documentation for the domestic synchronous speech synthesis API specifies that input text must be fewer than 10,000 characters per request; limits for other APIs may differ. (platform.minimaxi.com)
- The API supports MP3, PCM, FLAC, and WAV; WAV supports non-streaming output only. Generation results, latency, and available voices depend on the text, settings, and deployment conditions. (platform.minimaxi.com)
Pricing
- MiniMax international site API pay-as-you-go: Speech synthesis with
speech-02-turbocosts 60 USD per 1 million billable characters. This is character-based pricing for text-to-speech; the official price list does not separately list charges based on output audio duration or output tokens. Voice cloning, design, and previewing may be billed separately, and package rules differ. Accessed: November 18, 2026. (platform.minimax.io)