ASR.
Sovereign speech-to-text. Audio never leaves the machine.
ASR (Automatic Speech Recognition) converts the operator's voice into text on-device. No audio is ever transmitted to a vendor. The brain is part of the Vinis voice subsystem inside Mickai. Operates in extreme environments where conventional ASR fails: cold-induced vocal-tract changes, pressurised cabin acoustics, and helmet-and-mic distortions of EVA suits (patent 06).
- 01On-device speech recognition
- 02Wake-word detection and disambiguation
- 03Extreme-environment tolerance (cold, pressurised, EVA)
- 04Multi-language support without cloud lookup
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, 06
- 02LibriSpeech and Common Voice corpora
- 03Wav2Vec2 and Whisper papers
- 04ISO 639 language codes
- 05Word Error Rate (WER) academic literature
- 06Interspeech proceedings
- 07Kaldi toolkit documentation
- 08Phoneme inventory IPA reference
- 09Speech-recognition robustness benchmarks (CHiME, REVERB)
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.
- 01Echo-Capture (on-device speech recognition)
- 02Cipher (audio-stay-local enforcement)
- 03Wax-Seal (transcript-signing infrastructure)