Features of nonoha

nonoha is a node-based framework for signal processing and applications, made for real-time signal processing in Python. Developers build signal-processing graphs together with AI, and the whole application for their end users around them.

Pure Python and NumPy

nonoha processes signals in real time with pure Python and NumPy, without intricate acceleration tricks or dependencies in other languages.

Writing nodes and plugins
from scipy.signal import welch

def _on_input(self, values):
    # A plain NumPy array, no wrapper
    _, power = welch(values[0], fs=self.fs)
    self._output.dispatch([power])

A lightweight core

nonoha-core, the core nonoha is built on, is designed to process signals lightly and fast, with an optimized hot path. It depends only on NumPy: a small layer that plugins and other extensions build on to add features.

Real-time signal processing

NumPy is fast at vector operations over whole arrays. nonoha processes signals in batches, one NumPy operation per batch, and keeps up in real time.

A node-based editor that runs in your browser

In the standard browser that most platforms come with, place nodes, set their properties, and watch the signals flowing through the graph in live plots.

Concepts
The web-based editor with a graph: a clock drives a synthesizer, whose waveform goes through an EMA filter and a spectrum analyzer into four live plots

Your own web UI

Besides nonoha editor for editing and monitoring graphs, you can build a custom UI made for each workspace. It is web-based, so the libraries of the web ecosystem (WebGL, Web Audio, and more) let you build full-featured apps. nonoha engine hosts the UI you build, with no extra UI framework needed.

Your own web UI
Signal Lab, the web UI of a new workspace: the waveform, the spectrum, and the band energy, with controls for the frequency, the harmonics, and the waveform

Remote control over REST and WebSocket

The engine is an API server. Run it on one machine, and edit and monitor its graphs from a browser or your own app on another, over REST and WebSocket.

Built by people and AI together

People see and check in the GUI; AI coding agents read and write code. Both work on the same workspace, and Build with AI in nonoha Hub hands it to your agent.

People, in the GUI
AI agents, in code
Graphs
People, in the GUISee and edit the nodes in the editor
AI agents, in codeRead and write the graph's JSON file
Web UI
People, in the GUIUse the app in a browser
AI agents, in codeWrite the web page, then open it in a browser to check it
Workspaces and nonoha Cloud
People, in the GUInonoha Hub: create, run, download, and publish
AI agents, in codeThe nonoha CLI: the same operations as commands with JSON output, included with nonoha Hub

Share across people and organizations

nonoha Cloud is a platform that gathers workspaces made by many people and lets you share them: openly in the Community, or only with specific people or organizations.

From desktop to device

nonoha is designed to run across a wide range of environments, from desktops to mobile and embedded devices such as the Raspberry Pi. A workspace built on the desktop runs as it is on mobile and embedded devices, as long as its plugins support the platform. On desktops and embedded devices nonoha engine runs it directly; inside apps, on mobile or on a server, nonoha embedded does.

Windows
macOS
Raspberry Pi
iOS
Android