Voice

Generate speech with supported local and explicitly enabled hosted engines.

Availability and installation

This guide covers @wf/voice version 0.3.0-alpha.2, currently a preview. The catalog does not verify public registry installation for this version. Obtain the matching package and dependencies through the delivery method shown on its product page. Do not infer that an unlisted package can be installed from the public npm registry.

The supported runtime, exported entry points, and compatibility commitments are listed in Compatibility. Check Release notes before upgrading.

Configure and use

Keep SDK settings in a YAML file and pass its path to the package’s public client. Keep credentials in application-owned secret storage; configuration examples contain references, not secret values.

• voice.yaml — configuration example for this preview.

Use the supported entry points listed in Compatibility. The current catalog does not provide a standalone buyer script for this preview.

The example leaves paid and noncommercial-only engines disabled. Paid providers require both host authorization and the corresponding credential reference. Breeze-TTS-2 requires allowNonCommercial: true in YAML and a self-hosted server; that opt-in is only for noncommercial projects under the model terms. Commercial clients keep Breeze disabled and can use Pocket or Kokoro. The SDK owns request limits, cancellation and cleanup.

Integration responsibilities

The application owns user authorization, credential storage, and permission to make external calls. Review the example before running it: a connected provider or an update action can have real effects. Preserve the package’s cancellation and disposal behavior where those APIs are exposed.

Preview compatibility is limited to the combinations actually recorded in Compatibility. An empty verified-combinations list is not a promise that every platform or client has been tested.