Feature/rate of change

[possibly-human]

Summary

Adds a flexible rate-of-change (derivative) calculator for real-time signals. The module supports two windowing strategies—fixed sample count or fixed time window—and provides multiple output unit options. In time window mode, the last known value is held constant when no new data arrives, ensuring continuous output. Diagnostic outputs expose the internal window state for debugging and analysis.

Functionality

Input is:

• Signal (int or float) — input signal to analyze

Parameters are:

• Window mode (enum) — SampleCount (fixed number of samples) or TimeWindow (fixed duration)
• Sample count (int) — number of most recent samples to use (visible in SampleCount mode, minimum 2)
• Time window (float) — duration in seconds to use (visible in TimeWindow mode, 0.001–3600s)
• Output units (enum) — per millisecond, per second, per minute, or per hour

Outputs are:

• Rate (float) — computed rate of change in selected units
• Samples in window (int) — current number of samples in analysis window
• Window time span (float) — actual time span covered by window

Implementation notes

• Sample count mode: Uses ring buffer to store N most recent samples
• Time window mode: Automatically adjusts window to requested time span; holds last value when data is sparse
• Rate calculation: Derivative computed as (newest − oldest) / sum(dt) with unit conversion
• Performance: O(1) operations with configurable buffer capacity (1M samples max)

[possibly-human]

Testing

Was tested on Linux (booted on a T2 chip, 2019 MacBook Pro)

Once with an expression value generator :
stable rate of change

changing rate of change
changing_rate_of_change.webm

Example file : https://drive.google.com/file/d/1m3JYEtMVVxy1xB6bwtAg25Uz6f__wSGL/view?usp=sharing

And once with live data from Sensors2OSC (device orientation). The live data was then recorded and saved with CSV recorder, and output from the Rate of change object was compared to output from processing with similar algorithms in python, with the same parameters.
rate of change with live data

Example file : https://drive.google.com/file/d/14Kgu6Ltq5tnpljXwV_a8uOMfd2DTg--3/view?usp=sharing

output

Python file
roc.py

[jcelerier]

Really great work thanks!
I have a couple comments but those can be addressed without issue in later PRs

[jcelerier]

I think there's already an unholy amount of different ring buffer implementations in ossia :p
What do you think about reusing the boost one? https://www.boost.org/doc/libs/latest/doc/html/circular_buffer.html

[jcelerier]

I'd like to think about a way to do parameter watching without requiring you to make your own class. What are the features you need from it ?

[possibly-human]

basically just to indicate when the parameter's value has changed