Overview

Signal Studio is a workspace for analysing neural recordings where every step of the analysis can be opened, adjusted and checked.

The idea

Most analysis pipelines are scripts. A script works, but its decisions are invisible: which window function, which transition width, whether the filter ran before or after epoching. Those decisions change results, and they are buried in function arguments spread across several files.

Here, each step is a card on a canvas. Every parameter that affects the output is on the card, visible, with its resolved value. Open the gears view and you see the operations inside.

A pipeline is a file

Wiring cards together produces a graph, and the graph is a file you can save, send and re-open. Handing it to a colleague hands them the analysis, not a description of it. Archive it next to your data and the run is reproducible.

Cards are checked

Every built-in card is compared against an independent implementation computed separately in Python, on a real recording, before it is accepted. The agreement figures are published on the validation dashboard. A card that merely runs is not validated.

It runs on your machine

No account is needed to install or use it. Recordings never leave the machine; there is no code path that uploads one. See data location if your institution needs that in writing.

Where to go next

Install it, then follow your first pipeline. If you would rather read first, concepts explains the vocabulary.