Seed Audio Models Documentation

Seed Audio AI documents the model capabilities that matter for browser-based voice generation: what each workflow accepts, what it can produce, where public information stops, and what rights are required before cloning or imitating a voice.

This documentation is written for the Seed Audio AI product. It is not official ByteDance, Seed, Volcengine, or Doubao documentation.

Available Docs

Document	Use it for
Seed Audio 1.0 Model Documentation	Capabilities, generation modes, inputs, outputs, prompt examples, limits, and safety rules for Seed Audio 1.0 workflows.

Current Seed Audio AI Workflows

Seed Audio AI currently exposes three browser workflows:

• Text to Speech: paste a script, choose a voice, adjust delivery controls, generate, preview, and download.
• Voice Clone: upload reference audio only when you have rights or clear permission to use that voice.
• Voice Design: describe a reusable private voice and use it for future generations.

Future model pages should stay narrow: add only documented capabilities, avoid official-affiliation claims, and mark uncertain provider behavior as version- or account-dependent.