Vidu Audio Clone

Problem It Solves

Voice generation workflows often need to preserve a speaker’s timbre while reading new text. vidu audio clone targets this use case: it clones a voice through the dLazy Vidu interface and generates new text-to-speech audio. It is not a local inference tool; it is a thin client over a hosted API, making it suitable for wiring voice samples, prompts, and generation tasks into CLI or agent pipelines.

How It Works

The skill submits requests through the dlazy CLI. When a user provides an audio file or relevant path, the local file is uploaded to dLazy media storage so the backend model can read it. Prompts and parameters are sent to api.dlazy.com for inference, and generated audio URLs are returned from files.dlazy.com. If an asynchronous task is used, the returned outputs may be empty, with a generateId and status field included, requiring follow-up polling for task state.

Boundaries

The skill requires a valid dLazy API key and routes requests through a cloud SaaS path. The provided material does not describe local private deployment, batch concurrency limits, or server-side data retention, so sensitive voice samples should be checked for authorization and compliance first. Error handling matters: unauthorized, insufficient_balance, missing parameters, local file read failures, and safety-policy rejections should each be handled separately rather than retried as generic generation failures.