BRAINIALL Speech v0.1.0 release candidate

BRAINIALL Speech for Dify — release candidate

This candidate adds two Dify model-provider capabilities:

• pt-BR speech-to-text with speaker diarization;
• pt-BR text-to-speech with three allowlisted voices.

Validation completed

• 41 mocked transport and SDK integration tests passed;
• package validated by the Dify CLI;
• runtime-only archive with 15 allowlisted files;
• fixed HTTPS API origin, redirects disabled, zero automatic retries, explicit timeouts;
• credential, provider-body, file-size, text-size and WAV-container safeguards.

Important boundary

This is a release candidate, not a Dify Marketplace release. A valid-key live smoke test
inside Dify is still pending. Use it for evaluation and report reproducible issues; do not treat
the candidate as a production certification.

The plugin is free and open source. BRAINIALL API usage is metered and may be charged after
included credits. Audio and synthesis text are sent to the BRAINIALL API; review PRIVACY.md
before installation.

SHA-256 of brainiall-0.1.0.difypkg:
22c4d257f1b762846946c9d252de3adf0d00bf3fbfb076bc63b0e63258f7c6b0