TTS.
Sovereign text-to-speech. Synthesis is signed for provenance.
TTS produces Mickai's spoken responses on-device. Every synthesis carries the AudioSeal dual-layer watermark (patent 11): a robust spread-spectrum signal in the waveform plus a separate ML-DSA cryptographic seal in the metadata. Survives compression, re-encoding, and offline edits. Any downstream listener can verify whether an audio clip was generated by an authorised Mickai system.
- 01On-device speech synthesis
- 02AudioSeal dual-layer watermark embedding
- 03Voice-style adaptation under operator control
- 04Real-time streaming synthesis
Authoritative external corpora and standards this brain treats as canonical. Every retrieval against these sources is signed into the audit ledger so a regulator can prove which evidence drove which output.
- 01Mickai Patent 02, 11
- 02VITS, Tacotron, FastSpeech academic papers
- 03AudioSeal watermarking research (Meta)
- 04ISO/IEC 14496-3 MPEG-4 Audio
- 05Interspeech TTS challenges
- 06VCTK and LJSpeech datasets
- 07Bark and StyleTTS open-source corpora
- 08MOS Mean Opinion Score evaluation methodology
- 09ITU-T P.800 listening-test standard
Mickai-native tooling primitives this brain implements internally. Codex for sovereign plain-text graph PKM, Lectern for spaced-repetition memory, Stele for citation-provenance, and domain-native primitives layered on top. No external services in the trust path; data stays on operator-personalised hardware.
- 01Reed (on-device speech synthesis)
- 02Wax-Seal (AudioSeal dual-layer watermark)
- 03Cipher (synthesis-key custody)