MiniMax Speech
MiniMax Speech provides text-to-speech, voice cloning, and real-time voice agents through web, API, and self-hosted options.
What MiniMax Speech is
MiniMax Speech provides text-to-speech, voice cloning, and real-time voice agents through web, API, and self-hosted options. It supports 40 languages and exports MP3, PCM, FLAC, and WAV audio. Speech-2.8-HD offers fine-grained emotion and style control and can generate audio up to 30 minutes long. Speech-2.8-Turbo targets latency-sensitive workloads and supports low-latency PCM streaming. Both listed plans allow up to 10,000 characters per synchronous request; HD also allows asynchronous requests up to 1,000,000 characters. Instant Voice Clone uses 10 seconds of reference audio and is described as ready in under 30 seconds, with pay-per-use pricing and no setup fee. Professional Voice Clone uses 30 minutes of curated recordings, takes 3 to 5 business days to train, and is available in cloud or on-premise deployments. Voice agents support interruption, state-preserving resumption, and language switching, with LiveKit, Pipecat, custom WebRTC, and SIP integrations. The service is paid with no free plan. MiniMax says cloud and on-premise deployment are available.
Good
- Supports 40 languages
- Offers web, API, and self-hosted access
- Exports MP3, PCM, FLAC, and WAV
- HD supports up to 1,000,000 asynchronous characters
- Provides instant and professional voice cloning
Watch out
- Paid service with no free plan
- Professional cloning takes 3 to 5 business days
- Maximum synchronous input is 10,000 characters
- Professional cloning requires 30 minutes of recordings
Who it suits
It suits developers and teams needing multilingual speech generation, cloned voices, or real-time agents through an API, web service, or self-hosted deployment. Buyers should select a voice workflow according to latency, cloning, and request-size needs.
Verdict
MiniMax Speech combines multilingual synthesis, cloning, and real-time voice-agent capabilities across several deployment options. Its plan limits, paid access, and cloning requirements should guide selection.
Plans and prices
From MiniMax Speech's own pricing page, checked 8 Oct 2026.
| Plan | Price | What it gives you |
|---|---|---|
| Speech-2.8-Turbo | $60, paid once | Sub-200ms time-to-first-byte · PCM streaming · 40+ languages · IVC custom voices · $60 per 1M characters |
| Speech-2.8-HD | $100, paid once | 32kHz audio · 40+ languages · PVC custom voices · Long-form generation up to 30 minutes · $100 per 1M characters |
The facts
Each one read from MiniMax Speech's own pages, with the date it was checked.
What it does
- API access
- Yeschecked 20 Sep 2026
- Deployment
- MiniMax says cloud and on-premise deployment are available.checked 8 Oct 2026
- Export formats
- mp3,pcm,flac,wavchecked 20 Sep 2026
- HD voice generation
- Speech-2.8-HD supports fine-grained emotion and style control and generates audio up to 30 minutes long.checked 8 Oct 2026
- Instant voice cloning
- Instant Voice Clone uses 10 seconds of reference audio and is described as ready in under 30 seconds, with pay-per-use pricing and no setup fee.checked 8 Oct 2026
- Integrations
- The voice agent supports LiveKit, Pipecat, custom WebRTC pipelines, and SIP trunks.checked 8 Oct 2026
- Languages
- 40checked 20 Sep 2026
- Maximum input
- 10000checked 20 Sep 2026
- Professional voice cloning
- Professional Voice Clone uses 30 minutes of curated recordings, takes 3 to 5 business days to train, and is available in cloud or on-premise deployments.checked 8 Oct 2026
- Security
- The MiniMax API Privacy Policy says data exchanged with its servers is protected using technologies including HTTPS, SSL, and AES.checked 8 Oct 2026
- Support
- The product site offers a 30-minute walkthrough with the solutions team and directs users to API reference documentation.checked 8 Oct 2026
- Turbo voice generation
- Speech-2.8-Turbo is designed for latency-sensitive workloads and supports low-latency PCM streaming.checked 8 Oct 2026
- Voice agent features
- Voice agents support mid-sentence interruption, resuming with state intact, and switching languages mid-call without swapping models.checked 8 Oct 2026
- Voice cloning
- Yeschecked 20 Sep 2026
- What it does
- MiniMax Voice offers text-to-speech, voice cloning, and real-time voice agents in 40+ languages.checked 8 Oct 2026
Money
- Free plan
- Nochecked 20 Sep 2026
The company
- Maker
- MiniMax describes itself as a global AI foundation model company founded in early 2022.checked 8 Oct 2026
Where it runs
- Platforms
- web,api,self_hostedchecked 20 Sep 2026
Alternatives to MiniMax Speech
- 1
Voicebox
Free planOpen sourceAndroidiOSWebAPI9.3From $12 a year - 2
Speechify
Free planAndroidiOSWebAPI9.0From $10 a month - 3
Resemble AI
Free planAndroidiOSWebAPI9.0From $280 a month - 4
iSpeech
Free planOpen sourceAndroidiOSWebAPI9.0Free plan - 5
Fliki
Free planAndroidiOSWebAPI8.8From $28 a month - 6
Voice.ai
Free planAndroidiOSWebAPI8.8From $5 a month