About

Partners

The Laboratories are small. What they can now do, they do with partners.

For most of their history the Laboratories were limited not by what they could record but by what they could analyze. Two partnerships changed that: a technology partner that built the modern analysis pipeline, and a global network of institutions willing to share what none of them can explain.

Stratis, Inc.

The Laboratories' analysis pipeline -- the machine layer that triages the volunteer results of the Distributed Analysis Initiative and flags candidate detections for human review -- is built and maintained in partnership with Stratis, Inc.. The partnership dates to 2019 and is central to the Laboratories' ability to work at the scale the record now demands.

Field record

arrays, hydrophones

Distributed Analysis

volunteer compute

Stratis pipeline

candidate flagging

Human review

classification

Catalog

public record

Fig. 1 The analysis pipeline -- raw data becomes a reviewed catalog entry.

Spatial residual layer

FAL and Stratis, Inc. also maintain an experimental spatial-residual workflow. Where Stratis scene, object, and aerial datasets carry geometry, splats, meshes, labels, and quality metrics, the Laboratories attach environmental metadata: acoustic fingerprints, low-frequency vibration profiles, sensor-noise shape, recurrence confidence, and capture-quality flags.

Layer	FAL contribution	Use
Scene capture	room or route acoustic profile	compare repeated captures under different environmental states
Object capture	surface and occlusion residual notes	separate geometry error from repeatable sensor behavior
Aerial capture	site-scale coverage and vibration context	cross-check station routes and broad environmental change
Quality metrics	recurrence, confidence, and calibration flags	route anomalous captures to human review
Dataset export	release notes and access-state metadata	keep downstream models from treating uncertainty as fact

Fig. 2 Spatial data enrichment -- Public description of the non-public research interface between spatial capture and signal review.

scene_residual = capture_pass_B - align(capture_pass_A)
acoustic_layer = impulse_response + low_band_profile + ambient_variance

stable_site_behavior = corr(scene_residual, acoustic_layer) > review_limit
                       AND capture_quality >= minimum_quality
                       AND operator_variance < route_limit

Fig. 3 Residual-shape comparison -- Public notation for repeated spatial capture QA.

Source: FAL and Stratis public partner note
As of: 2026
Method: Simplified notation; internal thresholds and capture transforms are not published

Scientific partners

The Laboratories cooperate with observatories and monitoring networks worldwide, exchanging detections and cross-referencing the anomalous record. Much of the Persistent Signal Catalog exists because the field, at large, is willing to share what it cannot explain. The most-studied entries in the catalog are drawn from that shared literature and presented alongside the Laboratories' own on equal terms.

References & notes

1.Stratis, Inc. -- stratis.ai. Provider of the Laboratories' machine-analysis layer since 2019.
2.Stratis public API documentation describes scene, object, aerial, and dataset export lanes used as the public model for this partnership language.