google-cloud-speech

What it does

Uses Google Cloud Speech as a provider option when the job is fundamentally about speech audio, enabling transcription, caption generation, live caption planning, Text-to-Speech narration, and consented custom voice workflows, among other Google-specific capabilities.

How it works

• It targets tasks where speech is central: transcribing audio/video into scripts, captions (SRT/WebVTT), long-form analysis, live caption planning, and TTS narration.
• It supports Google voice families (Chirp 3, Gemini-TTS, Studio, Neural2, WaveNet, Standard) when governance, region, or infrastructure alignment calls for Google Cloud.
• It instructs to create a consented custom voice only when official access, voice-owner consent, and rights review are present.
• It cautions against using this skill for generic audio editing or non-speech tasks and notes that production finishing remains outside the Speech APIs.
• It presents an API boundary detailing separate Cloud Speech-to-Text, Cloud Text-to-Speech, and Gemini-TTS routes, with guidance to treat transcription and synthesis as one provider package for planning but separate implementations.
• It documents detailed operational facts for STT and TTS (model identifiers, outputs, quotas, pricing) and provides workflow steps for STT and TTS production.
• It outlines safety, consent, and data-governance requirements, including navigation of data logging and explicit voice rights confirmations.

When to use it

• Triggered when the project requires transcription, captions, long-form captions, or TTS voice assets, and when Google governance, regional controls, or existing GCP infrastructure are relevant.
• Appropriate when Chirp 3 / Gemini-TTS voice selection and consented custom voice workflows are part of the brief.
• Use for planning and budgeting (quotas, pricing, region checks) before implementation.

What it can touch

• Tools declared: claude-code, codex, copilot, cursor. The skill guides usage of these tools to implement or reason about Google Cloud Speech workflows within the project.

Caveats

• Do not treat this as a generic audio-editing or non-speech skill; finishing and QC require separate tools.
• Regional availability, quotas, and pricing are subject to change; re-check against official docs before spending.
• Access to Instant Custom Voice is restricted to allow-listed users and requires proper consent and rights.
• The skill emphasizes explicit consent for voice cloning and data governance; avoid uploading sensitive content without client approval.