MAT File Viewer: Open .mat Files Without MATLAB

Where your .mat files come from

The viewer finds the signals in the file for you, whichever tool wrote it:

MATLAB save

Workspaces saved with save or from the Workspace browser: vectors, matrices, structs (also nested) and struct arrays of signals. A time vector such as t, time or tout becomes the x axis of the signals of the same length.

Simulink To Workspace and logged outputs

Structure with time, Structure and Array formats, plus tout and yout. Each block’s signals become one table with their labels as column names, multi-element signals included.

dSPACE ControlDesk

Measurement exports with X and Y channels. Every raster (for example a 1 ms and a 10 ms task) becomes its own table with its own time axis, and channel names and units are kept.

CANape and INCA style exports

Files where every channel is a vector, with one or more time vectors such as t_eng or time_brake. Signals are paired with the matching time vector by name, and matrices with a list of channel names get those names as column headers.

Python

Files written by scipy.io.savemat (v4, v6, v7) and hdf5storage (v7.3) open the same way as files from MATLAB.

What isn’t supported yet

• MATLAB table, timetable and datetime objects. These objects (and string arrays, categorical arrays and Simulink Dataset objects) can only be read by MATLAB itself today. The viewer lists them in the notes and opens everything else in the file.
• Function handles. Skipped, with a note.
• Very large sparse matrices. Sparse matrices with more than a million elements are listed in the notes instead of being expanded.
• Editing and saving. The viewer reads .mat files; it never changes them.

Tip: to open a MATLAB table here, turn it into a struct of columns before saving: s = table2struct(T, 'ToScalar', true); save('data.mat', '-struct', 's'). For a timetable, call timetable2table first. Every column then appears as a signal.

Built for big files

Long test runs and high-rate loggers easily produce .mat files of hundreds of megabytes. The viewer opens them in seconds, right in your browser. In our tests in Chrome on a recent laptop, a file with a time vector and four channels of 10 million samples each opened in:

• about 1 second saved with -v6 (400 MB);
• about 2 seconds saved with -v7 (330 MB, compressed);
• under 3 seconds saved with -v7.3 (330 MB).

A one-sample spike in 10 million samples is still visible in the overview, and zooming in reloads the exact values of the visible range instantly. Slower machines take longer, and very large files need enough free memory.

Private by design: nothing is uploaded

Measurement data is often confidential: unreleased prototypes, customer projects, supplier test results. Online converters that upload your .mat file put a copy on someone else’s server. MegaRows does not: the viewer runs entirely in your browser, and the file and its name are never sent anywhere. Read the privacy statement for details.

Need to plot CSV, Excel or text logs as well? The same viewer opens them: see the plotter.