ERAIS / working systems

Intelligence in motion.

Follow a persistent world. Hear speech. Explore language, vision and reasoning. See how Fracture and native training build capabilities for one adaptive architecture.

Explore the results through the interface. The model weights and proprietary engine stay private.

Start here / real model output

Same words. Two speech systems. Your ears.

Listen to a LuxTTS control beside its Fracture-converted speech route. The roles stay concealed: focus on the voice, the rhythm and the detail.

Give your verdict →

“A quiet lantern glowed beside the window while rain softened the empty street.”

Voice A

Voice B

Choose either voice. Playing one pauses the other.

Inside this comparison

These released clips use the ERAIS K7 route, which sparsifies two learned projections in one vocoder block. They demonstrate that conversion step; the broader Fracture programme targets the full adaptive native architecture. Explore the speech results.

Beyond the conversation turn

One world. A continuing life.

ERAIS carries state forward between observations. The live shared-world loop learns from recorded audio and language, then settles into low-cost maintenance while it waits.

Follow the live organism →Try live reasoning

01Perceive
Bring observations into shared state.
02Adapt
Apply the capability and effort needed.
03Continue
Carry outcomes forward. Rest when quiet.

Fracture / learned capability, modular structure

Convert the knowledge. Expand the system.

Fracture reorganises learned components into composable experts. The full adaptive ERAIS architecture is the destination: the system chooses the capability and effort an input needs. This speech experiment lets you inspect one of the measured foundations.

Measured30 August 2026CPU · LuxTTS / Vocos

Explore the original speech reconstruction experiment

Select an executed mode

Choose the expert budget

Full bank: 8 / 8 experts selected per acoustic frame

Filled blocks show the selected count only. The measured router chose different expert identities for different acoustic frames.

Selected projection: 100%
Reconstruction: Exact component reconstruction
Waveform correlation: 1.000
Spectral convergence: 0.0000
Whisper proxy word edits: 0 / 12
Human listening score: Not measured

What ran

Speech entered a real donor and returned a waveform.

The experiment split one learned second projection into eight equal channel experts. Selecting all eight reproduced the donor waveform with zero maximum and RMS error.

Measured sparse modes

K4 and K2 preserved every proxy-transcribed word.

K4 used 50% and K2 used 25% of that projection. Whisper reported zero word edits on the same 12-word target while the quality sweep exposed the reconstruction trade-off.

Next milestone

Fracture is moving through the full speech stack.

The next vertical moves routing ahead of dense expansion, converts additional learned blocks, and measures end-to-end speech quality and speed.

Next physical path · QwenSelected expert work can now be grouped before timing.

A grouped physical-dispatch path is committed and equivalence-tested for the Qwen evaluator. Full-tower quality and speed measurements remain in progress.

LuxTTS source: 8c893869c4bb
Measurement digest: 8af672e6be05…
Qwen executor source: 184c9afc3f5f

Build on learned capability. Fracture is the conversion path; adaptive native ERAIS is the destination. Donor weights, activation traces and proprietary methods stay behind the interface. Bring Fracture to a valuable model workload.

Bonus / browser sound playground

Interactive 48 kHz stereo loop

A fresh three-stem loop synthesized entirely in your browser.

Interactive · local browser DSP

Build a three-stem instrument loop

Change tempo, palette, or stems. Sound is created in this tab after your click; no recording, model request, or upload occurs.

Tempo100 BPMInstrument palette

Active stems

Processing mode: Browser DSP · no network
Loop duration: 2.40 s
Active stems: 3 / 3
Start latency: Not started

Browser instrument: instant local synthesis with no upload or model request. For actual model-generated speech, listen to the Fracture comparison above.

Vision demo

Fixed sparse classification probes

Three predeclared site-owned cards evaluated offline; no camera or upload.

3 / 3 fixed cards visible

Intended card

Airplane

Top fixed output 74.76%

Intended card

Ship

Top fixed output 91.20%

Intended card

Truck

Top fixed output 87.34%

Measured foundation: the retained 5.22M-parameter run recorded 60.45% on its exact CIFAR-10 slice. These fixed cards make the response pattern inspectable while broader vision qualification advances through the next mask-overlap gate.

Text demo

A verbatim CPU-only story sample

A fixed capture from the reviewed native-byte checkpoint on an owned desktop CPU.

530 UTF-8 bytes · 110 words

Prompt

Once upon a time, there was a little girl named Lily. She loved to play in the garden. One day,

ERAIS output · verbatim

Once upon a time, there was a little girl named Lily. She loved to play in the garden. One day, she saw a big bird with a big smile. She wanted to see what it was. Lily said, "Lily, it's time to go home. It is not for the big boy." So, they went to the park and saw the bird. The bird was so happy and said, "Thank you, Lily. You are a good friend." Lily smiled and said, "You're welcome, it's important to be careful with the bird."

What this demonstrates: an owned desktop CPU ran the native-byte checkpoint and produced this captured story. The inspector stays entirely on-page; Story Lab unlocks live generation after its service check.

Symbolic reasoning demo

A bounded, inspectable proof trace

One approved sensor example from the fixed controlled-English receipt.

All 4 receipt steps visible

Controlled-English input

Fact: the sensor is calibrated. Fact: the sensor is powered. Rule: anything that is calibrated and powered is ready. Question: is the sensor ready?

DecisionReady — proved from both facts.

Proof trace

Every step shown

1
given
The sensor is calibrated.
2
given
The sensor is powered.
3
rule application
Calibrated and powered implies ready.
4
conclusion
The sensor is ready.

Measured foundation: 30 fixed examples completed and 6 deliberately failed closed. The trace makes every conclusion inspectable; the live Lab unlocks after its service check.