MiniMax Audio has launched the Speech-02 series of voice models, which can input up to 200,000 characters at a time.

MiniMax Audio has officially released the Speech-02 series of speech models. The model supports over 30 languages, allows input of up to 200,000 characters at a time, and achieves a human-like similarity of up to 99%. The synthesized speech is natural and smooth, effectively resolving issues such as audio playback stuttering and unstable rhythm. MiniMax Audio has also introduced the “Read Anything” feature, enabling users to upload local files or paste web URLs to convert documents, e-books, web articles, and more into their preferred voice for listening.