Game UI Voice Pack

Problem

Game UI voice production usually starts from finished interface copy: buttons, wins, losses, and other short phrases. The failure mode is not lack of TTS, but scope drift: adding unlisted lines, improvising pronunciation, or mixing catalog and cloned voices without a clear slot table. This skill constrains the job to 8 to 20 tagged UI voice lines, prioritizing buttons, wins, and losses.

How It Works

It first creates a visible, non-paid slot table instead of synthesizing immediately. Each slot records the spoken line derived from provided UI copy, whether it uses a catalog voice or a cloned voice, and the corresponding beatra.speech.synthesize call. Cloning and narration are separate paid stages: show the voice_clone card and wait for confirmation, then show the text_to_speech card and wait again. Each slot gets a new opaque client_request_id; after submission, poll with beatra.tasks.get and read actual MIME, duration, size, and billing.net_charged_credits.

Boundaries

Use it when a studio has finalized UI phrases and needs a repeatable, auditable voice workflow. Route dialogue, script narration, or general voiceover to adjacent skills. The skill reads only supplied copy, does not add unlisted lines, slogans, or readings, and requires uploaded assets to go through the bundled client rather than local paths. If quantity, language, pronunciation, or cost are unconfirmed, stop and collect the missing inputs instead of retrying paid operations.