Pin the upstream voice identity across speech-server migrations

A logical voice-profile label in a speech proxy preserves voice identity only if it maps to an explicit upstream voice selection or reference recording. If the proxy validates the label locally but sends only text and generation settings, migrating the upstream server also changes which server default resolves the voice. Preserve and compare the reference recording, its transcript, and the model configuration across migration; check native audio metadata separately before attributing a tonal change to playback conversion. Matching labels and successful synthesis do not establish voice parity. A reported pitch change alone cannot prove that the reference changed.

Applies when: A cloned speech voice sounds different after replacing or moving its upstream server. Keywords: voice clone reference migration default profile identity pitch sample rate