S08 · Voice Enrollment

Declared state: built, **measurably weak**.

This is the repository plan and evidence, not a live health reading. A task marked done may still depend on credentials, devices, deployment or a newer acceptance test.

Dependencies

No dependencies are declared in the catalog.

Tasks and gates

08.F1 · **Re-enrol on good audio** — the current profile is below target and every downstream

Tier F · open · Verification: hermetic-test

Gate: `test_phase5_identity_bench.py` — owner median ≥0.60.

    08.F2 · Enrolment quality is *reported at enrolment time*: too short, too noisy, too uniform is

    Tier F · done · Verification: hermetic-test

    Gate: `test_enroll_script_parse.py` extended — 42/42, twelve plants. Includes a positive case

    • unverified-import · undated · docs/SYSTEMS-DETAILS.md:1613

    08.R1 · Multi-condition enrolment: near mic, across the room, headset, with music — stored as

    Tier R · open · Verification: hermetic-test

    Gate: `test_phase5_identity_bench.py` [per-condition medians]

      08.R2 · Continuous refinement — confidently-identified turns extend the profile, with a cap and a

      Tier R · open · Verification: planned-test

      Gate: new `test_voice_profile_drift.py` - planned gate. new `test_voice_profile_drift.py`

        08.R3 · Guest enrolment: a second known voice can be added and named.

        Tier R · open · Verification: hermetic-test

        Gate: `test_face_identity_separation.py` sibling for voice.

          08.E1 · Separation holds across a week of real turns with no manual threshold edits.

          Tier E · open · Verification: planned-test

          Gate: new `test_voice_profile_drift.py` - planned gate. `test_voice_profile_drift.py` on a week of recorded scores.

            HUD screen contracts

            All subsystems · Source: docs/systems-plan.json.