ClosedMiniMax/10 trustMetrics-checked

MiniMax Speech-02

Name: MiniMax Speech-02
Rating: 7

Studio-grade zero-shot TTS with voice cloning and emotional delivery, via API

About

Speech-02 HD is MiniMax's flagship text-to-speech model, described in their technical report as intrinsic zero-shot TTS with a learnable speaker encoder: clone a voice from a short sample, then steer delivery across 30+ built-in voices with controls for speed, pitch, volume, and subtitle timing, plus streaming output. It targets the professional voiceover tier — audiobooks, dubbing, video narration — and MiniMax's own comparisons pitch it against Azure TTS and AWS Polly on demo samples. It is a closed, API-only product with no open weights, and pricing lives behind the MiniMax platform, so costs need checking against their current rate card before committing a workload.

Specs

Context window

Text input; up to 10,000 characters per request

Pricing

Paid API — usage-based via the MiniMax platform

Free quota

Check the platform for current trial credits

API access

REST API (api.minimaxi.com v1/t2a_v2); also available through Replicate

Weights

Closed — API only

Released

Not disclosed

Benchmarks

Team-reported: demo comparisons favoring Speech-02 HD over Azure TTS and AWS Polly (tech report, minimax-ai.github.io/tts_tech_report)

Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.

Use cases

Audiobook narrationVideo voiceoversAI voice agentsPodcast production

Trust score — 7/10

/10 trustMetrics-checked

Scored 7/10 (metrics tier, metrics-v1). Evidence: Tech report arXiv 2505.07916 and official demo page minimax-ai.github.io/tts_tech_report verified; Proprietary API — pricing not fully verified, kept generic rather than guessing; Exact release day not verified — releasedAt null.