AI Voice Generator

The voice browser includes four Chinese voices—冰糖, 茉莉, 苏打, and 白桦—and four English voices—Mia, Chloe, Milo, and Dean—plus an automatic default. These are synthetic presets rather than real people. You can use a preset immediately or switch to an authorized reference recording for a one-time cloned generation.

A preview request uses a short fixed sentence in the selected language instead of your complete script. That makes it faster to compare voices and prevents a preview from consuming the same character credits as a full authenticated generation. Preview calls still use provider capacity and remain subject to IP rate limits, the global daily budget, and the emergency generation switch. Listen for clarity, pacing, pronunciation, and how the voice handles the language you plan to publish.

Start with the audience and listening context rather than an imagined celebrity style. A product walkthrough may need a measured pace, while an energetic tutorial may benefit from a slightly faster delivery. Test a sentence that contains names, numbers, abbreviations, and the punctuation patterns found in your real script. Compare two or three built-in voices at the same speed, then keep the settings stable across related clips so the final series sounds consistent.

Built-in synthetic voices can support tutorial videos, e-learning modules, app onboarding, accessibility prototypes, podcast segments, product explainers, internal training, and multilingual content drafts. The generator is not a consent bypass or identity tool. You should disclose synthetic audio when context requires it and confirm that you own or can use the source text. Commercial rights depend on the provider terms, your plan, applicable law, and any third-party content included in the script.