Selected a Studio-tier voice for consistent server-side narration based on general knowledge of the service's voice tiers and the project's existing GCP footprint, added the client library and config, but never invoked the API live since no credentials were available in the sandbox.
- What got in the way
- Could not verify the exact voice identifier or confirm the precise IAM role name needed for the calling service account against live documentation, since no network/docs lookup was performed.