Characters — the call-centre cast¶

← Not-Happy-Jan docs

When your coding agent wants your attention, you'll hear from Jan, Bazza or Karren — depending on the severity. Be warned: if Karren picks up the call, you know something's gone wrong. She'll be wanting to speak to a manager. That's you.

It's a three-tier escalation ladder — like a support line where the problem keeps getting bumped up the chain. Each character has its own voice (zero-shot cloned from a reference sample), a phrase bank, and dials you can tune live.

A severity ladder rising bottom to top. Jan (reception) handles routine work and all-done — fires on ok, step, celebrate. Bazza (middle management) handles something that needs a look — fires on warn. Karren (the manager) takes over when it broke or is urgent — fires on err and attn.

The dials¶

Each character has an intensity dial and (for Jan and Bazza) a competence dial — 1 = useless, 10 = sharp:

Jan — reception (default). ockerism 1 (professional) → 10 (broad bogan) → 11 (swearing); competence 1 (mangles everything, trails off) → 10 (clean relay).

"Beauty, all sorted." / "Er, in-fur-france needs help, or whatever…"

Bazza — middle management. stress 1 (apathetic) → 10 (anxious, talking fast); plus the same competence scale.

"Yeah, might wanna look at that when you get a sec."

Karren — the manager. A single karren scale: 1 (curt) → 5 (rude) → 10 (full-Karen meltdown, demanding your manager — i.e. you).

"Oh, Jan. Jan. Keep calm, count to ten… NOT — HAPPY — JAN!"

Tune any of them live, without restarting — just ask Claude ("make Jan more bogan", "set Jan's competence to 3"), or use the CLI:

nhj set jan.ockerism 11    # full swearing bogan (swears bleeped — see dynamic-voices.md)
nhj set jan.competence 3   # muddled — mangles the message
nhj set bazza.stress 9     # frantic middle-manager
nhj set karren.karren 10   # full meltdown
nhj status                 # show current dials

The swearing/bleep behaviour at ockerism 11 is covered in Dynamic voices.

Adding a voice character¶

voices/
  myvoice/
    ref.wav        # 5-10s clean audio sample (no reverb)
    ref.txt        # verbatim transcript of ref.wav
    character.yaml # name, model_tier, preamble, phrases

Add your own characters by dropping a ref.wav, ref.txt, and character.yaml into voices/<name>/. Each character is a standalone voice with its own reference clip. See voices/jan/character.yaml for a full example.