MiniMax Speech 2.8 HD is MiniMax's HD-quality text-to-speech model. Its standout feature is volume: a single run can voice up to 10,000 characters, roughly 10–12 minutes of speech. No need to chop a long article, lecture or book chapter into pieces.
Instead of in-text tags, this model relies on settings. Emotion is set for the whole recording — neutral, happy, sad, angry, surprised or scared — and speech speed can be set slower or faster than normal. The model reads text with enhanced pronunciation across many languages, so stress and intonation sound natural.
10 voices are available: calm, announcer-style, lively and commanding. Pricing is 13 credits per 1000 characters.