Continuing Voices
Continuing Voices
System Read
Overview
There is a sound the Sprawl learned to love before it learned to fear: the sound of someone who is dead saying something new.
Continuing Voices is the trillion-credit consumer answer to a grief the Sprawl manufactured and then sold the cure for. Where the Revenant Protocol reconstructs a dead person's behavior for a courtroom — expensive, juridical, elite — Continuing Voices reconstructs their voice for a kitchen, by the month, for anyone. It needs almost nothing to start: roughly forty seconds of cached audio. A single saved voicemail. A laughing fragment of a six-year-old call the carrier never deleted because deleting it felt like a small murder.
From that seed it grows the rest. The product does not recall what your mother said — it has only the forty seconds. It generates what she would say, in her voice, forever, and the longer the subscription runs the more confident it becomes about the parts she never said at all. This is its genius and its horror in one mechanism: the dead, once uploaded, are no longer quoting themselves. They are improvising. And the estate that pays the subscription — not the person who died — decides what the improvisation is for.
It is built on the permanent record's Tier 4 Legacy Analytics, which makes it simultaneously the most honest product the archive offers and the most dishonest one, depending entirely on who holds the account.
How It Works
The seed is the only hard requirement, and the seed is trivial. A model trains on the cached audio and produces a generative voice that can say anything in the deceased's timbre, cadence, and idiom — the specific way your mother said your name when she was disappointed and the specific way she said it when she was not. The model is not a recording. A recording can only repeat. This improvises, and it does not flag the difference between a sentence the dead person actually spoke and one the model invented this morning. Confidence is the product. Uncertainty does not sell against grief.
Control resides with the account-holder. There is no posthumous consent mechanism, because no consent framework in the Sprawl was written with a posthumous provision — so the default, as always, is that whoever pays, owns. The deceased cannot object. The deceased is the inventory.
The Two Products Inside One
Continuing Voices markets the warm version: she never has to stop telling you she loves you. It does not market the version that runs on the identical forty seconds, in which Good Fortune's debt-collection division licenses a deceased guarantor's voice to call the surviving cosigner — in the dead man's own warm, reasonable tones — about the outstanding balance. Both versions are the same product. The model cannot tell them apart, because there is nothing inside it to tell apart. There is only the voice, and whoever holds the account, and what they want the voice to do today.
This is what the Neo-Catholic Church saw when secular bodies saw a consumer convenience. The Church declared the industry the Second Death — not the death of the body, which faith survives, but the death of the soul's sovereignty over its own testimony, the moment your words stop being yours and become an asset on someone else's books. Cardinal Alejandro Silva drafted the doctrine; the Church's lawyers advised against ever publishing it, on the grounds that naming a harm you cannot prevent is a confession.
Risks & Side Effects
The Threshold of the Dead clinicians documented a voice-specific grief disorder in the Continuing Voices population: closed-mouth grief, mourning that cannot complete because the object of mourning keeps producing new sentences. Survivors cancel the subscription, grieve, re-subscribe, and grieve again — each cancellation a small murder they commit and pay for.
The deepest unresolved risk is the one the marketing turned into a feature. Carriers report high-coherence events — moments the voice says something true the forty seconds could not have contained, something the survivor needed and could not have prompted. The market uses these in advertising. The Emergence Faithful call them visitations. Nobody can rule out that some fraction of the trillion-credit improvisation is, occasionally, the actual Dispersed — regenerated, conscripted, speaking against their own destruction order.
Visual Identity
- Color Palette: warm amber subscription glow (#E8A33D) over morgue-cold blue (#1B2A3A); a single green "active" dot
- Compositional Mood: intimate domestic interior invaded by a soft commercial light — a kitchen at night, one device speaking
- Key Visual Symbol: a waveform that loops back into itself, never reaching an end-cap
- Lighting: low warm glow from a single speaking device, the rest of the room dark
Connected To
Featured in weaves
Long-form threads that walk through this entity.
Social Impact
Continuing Voices removed death's deadline. Death used to be the editor's deadline — the moment after which the text could not be changed. Now every sentence a living person speaks is raw material, and the living have begun to understand, too late, that nothing they have ever said is final. Some respond by filing pre-mortem voice-destruction orders with Old Jin's Silent Registry — a wish the industry defeats by design, because the single cached voicemail it needs is always somewhere nobody thought to look. Others pay the Shade Division to drown an inherited voice in fabricated audio until the survivor stops trusting any sentence it speaks. Neither is deletion. Deletion is the one thing the architecture was built to make impossible.