MiniMax Speech 2.8｜Script Studio
Paste your draft and get a paste-ready sound-tag script plus a settings card. Audio is synthesized on MiniMax official channels.
Turn your draft into a MiniMax Speech 2.8 voiceover script with sound tags and an HD/Turbo settings card. Synthesize audio on official channels. Try free now.
Features
- Paste-ready script——A dry narration draft rewritten as segmented script with 20+ sound tags like (laughs) and (breath). Copy and go.
- HD/Turbo settings card——Voice direction, emotion, speed and pitch in one card. HD for texture, Turbo for speed — never guess the build again.
- De-robot checklist——Short sentences, filler words, spoken-out numbers, light room tone — every script passes a human-feel check before delivery.
- Script plus short video——Optional 5–10s mood short (MiniMax-family video) when you need visuals. Atmosphere B-roll only — no lip-sync promised.
How it works
- Paste your draft Paste the narration draft, attach a cover image (add a 10-second reference recording to fix the timbre direction).
- Pick purpose & build Pick short-video narration, the Turbo build, and tick script-plus-short if you want visuals.
- Take script, synthesize Take the sound-tag script plus settings-card doc and the mood short; paste the script officially to synthesize.
FAQ
- Where is the audio synthesized?
- Not in-station. Take the script to MiniMax official voice page (minimax.io/audio), Hailuo AI, or the open platform / Replicate, pick speech-2.8-hd or turbo, paste and synthesize — mp3 in seconds. Every delivery includes this 3-step guide.
- HD or Turbo: which build?
- HD (speech-2.8-hd) for texture: audiobooks and narrative films at 0.95–1.0x. Turbo (speech-2.8-turbo) for speed: short videos and realtime agents with punchy lines. Pick a build in the tool and the card locks the settings.
- Can 10 seconds clone a voice?
- Official cloning works with 10 seconds of clean audio, but cloning is not done in-station. Upload a reference and we will capture its timbre in the card and recommend the closest built-in voice; do real cloning officially with 15–30s of dry voice.
- Why does it sound so robotic?
- Usually the text is too perfect: long sentences, zero breaths, misplaced stress. Scripts split sentences, add fillers, tag emotional turns and spell out numbers; finish with light room tone per the checklist.
- Can it make talking videos?
- No — Speech 2.8 outputs voice only, no faces or lips, same here. Companion shorts are mood B-roll with no lip-sync promised. For a talking head: take the mp3 plus a frontal photo to a HeyGen/D-ID style lip-sync tool.
- Can it clone someone else?
- No. Only your own or properly licensed voices; celebrity imitation raises portrait and voice rights issues, so we decline and offer an alternative: perform a similar style with a built-in voice.
- Paid? Any cracked version?
- Script and card generation here — and there is no cracked version, ignore phishing links. Official synthesis bills per character, HD above Turbo, with free quotas for new users. Always go through official channels.
- Which languages work?
- Speech 2.8 covers 40+ languages. For global releases we localize sentence by sentence with pronunciation notes; synthesize one language per run — never stuff multiple languages into one task or accents bleed.
Try it free