ToneSphere

Low-latency audio routing and mixing for Windows, Linux and macOS. Published by Neural Nexus Studios, Kolkata, West Bengal, India.

Plug a guitar into an audio interface, run it through the plugins you already own, and hear it in your headphones — without a chain of virtual cables, a donation timer, or a driver panel from 2006. That is the whole idea.

What it does

Routes and mixes audio between the devices on your own machine. On Windows the audio runs on a native real-time engine with WASAPI and ASIO backends; on Linux and macOS, on a PortAudio host.
Hosts VST3 effect plugins on Windows — a browser that scans each plugin in a separate process, insert chains on any device’s input or output, parameters, bypass and the plugin’s own editor. Proven with the project’s own test plugin, with Surge XT, and with Native Instruments’ Guitar Rig 7.
Gives you a mixer and a patchbay: pan and balance, polarity, trim, mute, solo, smoothed gain, a limiter on every output, per-channel meters, and cables you make by dragging a port onto a port.
Ships real effects: biquad EQ, a compressor with actual attack, release, knee and makeup, and a delay with feedback.
Captures one application’s audio on Windows (Windows 10 build 20348 and later, no driver and no virtual cable).
Creates a Linux sink other applications can select, and publishes a CoreAudio loopback device on macOS that is proven in continuous integration and unverified in day-to-day use — stated that way on purpose.
Comes back as you left it: the session — routes, mixer, plugins and their settings — is saved as you work and restored when you open it. Presets are the same plain YAML, keyed on stable device identifiers, and say what was missing when your interface is not attached.
Keeps latency figures apart: ToneSphere’s own buffer arithmetic, what the drivers and plugins report, and — on Windows — a round trip it measures by playing a sweep and timing its return. Until that measurement has been taken on your own output and input, at your current settings, it is shown as --.

There is one rule in this project, and its test suite enforces it:

A feature does not exist until a test proves it moves audio. And a measurement you did not take is reported as --, never as 0.

The repository’s README is the authoritative list of what works, what does not work yet, and why — including the Windows virtual device, whose driver carries audio between applications in a test VM but is not yet signed for anyone else’s machine, and exactly where it stands in the virtual audio driver notes.

Download

Latest release downloads — a Windows installer (per user, no administrator prompt) or portable zip, a Linux AppImage and a macOS disk image, each launched by continuous integration the way you would launch it before it is published. A new release is published on every change that passes.

They are not code-signed yet: on Windows choose More info → Run anyway when SmartScreen asks; on macOS right-click the app → Open the first time. Then press Monitor Input: it picks your interface and its first input, in both ears. How that was checked with a guitar, a commercial amp simulator and the release build is in the first-run verification.

The Windows build on GitHub includes ASIO support built from Steinberg’s ASIO SDK under the GNU GPL version 3, and is distributed under that licence; its complete source, SDKs included, is published with every release in the gpl-source release and linked from the release notes. The Microsoft Store package leaves ASIO out, and it and the Linux and macOS builds are MIT-licensed.

Engineering documentation

How ToneSphere is built, and the evidence for every claim above. Each capability carries one evidence level, from NOT IMPLEMENTED to HARDWARE VERIFIED, and a figure nobody measured is shown as --.

Engineering report — what the Windows-native rebuild delivered, with every figure’s source.
Implementation status — the record of what is done, capability by capability.
First-run verification — the release build installed and used through its own window, with a guitar plugged in, and the output measured.
Architecture — the Python control plane, the native real-time engine, and the C ABI between them.
Real-time rules — what the audio thread may and may not do, and the measured callback timings.
Windows audio, ASIO, VST3 and the virtual audio driver — each backend, and exactly what has been verified against what.
Testing — how “it works” is established, with real signals.
Building on Windows and dependencies — the toolchain, and why each dependency is there.

ToneSphere is free today, and every feature in it is available without paying anything. Nothing is time-limited, nothing nags, and no function waits behind a donation prompt — that experience is the reason this project exists at all.

If it saved you the price of a routing utility, or a re-purchase of software you already owned, you can put that toward its development:

GitHub Sponsors — github.com/sponsors/avishakeadhikary
Patreon — patreon.com/avishakeadhikary
Ko-fi — ko-fi.com/avishakeadhikary
Buy Me a Coffee — buymeacoffee.com/avishake69

Sponsorship is voluntary. It buys no feature, no priority and no private support channel — see section 10 of the Terms of Service — it just funds the time that goes into the next release.

Legal

Terms and Conditions — the agreement governing your use of the application: grant of use, acceptable use, third-party plugins, warranties, liability, and governing law.
Terms of Service — what the application provides and does not provide, availability and support, updates, the optional network feature, and distribution through the Microsoft Store.
Privacy Policy — no accounts and no collected information, what stays on your own machine, and the one optional feature that sends audio off it.

Project and contact

Source, issues and releases: github.com/AvishakeAdhikary/tone-sphere
Questions, bugs and anything about the documents above: open an issue. Issues are public, so keep confidential details out of them.