MiniMax's speech-2.8-hd text-to-speech model, described as ultra-realistic with sound tags, in 40 languages. It is the quality tier of the Speech 2.8 pair; speech-2.8-turbo is the speed tier.
Pick it if
you want the higher-quality Speech 2.8 voice for narration
you want interjection tags such as laughs and breaths written into the text
you want to pick the voice and speaking speed
Skip it if
you need the lowest latency; speech-2.8-turbo is built for speed
you need whisper emotion; Speech 2.8 does not support it
MiniMax's text-to-speech model from the Speech-02 series, built for voiceovers and audiobooks. It is the high-fidelity tier; Speech-02-Turbo is the low-latency sibling for interactive use.
Pick it if
you narrate voiceovers or audiobooks where sound quality beats latency
you need speech in 30+ languages from 300+ preset voices
you want speed, volume and pitch control over the voice
Skip it if
you need real-time, low-latency speech for live conversation
you need music or sound effects rather than spoken voice
FAQ
Which is cheaper, MiniMax Speech 2.8 HD or MiniMax Speech 02 HD?
MiniMax Speech 2.8 HD is cheaper — $0.003 / second vs $0.015 / second.
How do I use MiniMax Speech 2.8 HD or MiniMax Speech 02 HD?
In Octomind Cloud on a paid plan, ask the agent to generate with either model by name; it runs the job and returns the file.
Try either model
Switch between models inside the same Octomind session.