Terms of Service
- Application: ToneSphere
- Publisher: Neural Nexus Studios, an individual developer, Kolkata, West Bengal, India
- Document version: 1.4
- Effective date: 1 October 2026
1. What this document covers
This document describes what ToneSphere offers you as a working application — the capabilities it provides, the ones it does not, how it reaches you, how it is updated, and what support and availability you can expect.
It is the service-facing half of the agreement. The other half, the Terms and Conditions, is the agreement governing your use of the software itself: the grant of use, acceptable use, third-party plugins, warranties, liability and governing law. Those subjects are not repeated here. The Privacy Policy covers information handling.
By using ToneSphere you accept this document together with the Terms and Conditions.
2. What ToneSphere provides
ToneSphere is a local desktop application for routing, mixing and processing audio on your own computer. Provided you have working audio hardware and drivers, it offers:
- Routing and mixing between the audio devices on your machine. On Windows this runs on a native real-time engine: one device’s clock drives the mix, and every other device crosses into it through a drift-corrected buffer. On Linux and macOS it runs on a PortAudio host that mixes inside each driver callback, which is not real-time safe in the strict sense.
- Mixer controls — pan on routes (constant power for mono sources, balance for stereo), balance on stereo channel strips, polarity inversion, per-channel trim, mute and solo, channel swap, smoothed gain changes, and a limiter on each output.
- Built-in effects — biquad equalisation (peaking and shelving filters, high-pass and low-pass), a compressor with attack, release, knee and makeup gain, a limiter, and a delay with feedback; on Windows, in each device’s and bus’s insert chain alongside plugins.
- A patchbay in which routes are made by dragging a port to a port, and feedback loops are refused rather than created.
- Presets saved as plain YAML and keyed on device names rather than device indices, so that they survive hardware being plugged in or removed. A preset saved with an interface attached will load without it and report what was missing.
- Latency reporting that says what each figure is. The application shows the latency the audio driver reports next to the figure that is only buffer arithmetic, labels each for what it is, and shows a measurement it did not take as unavailable rather than as zero. On Windows it can also measure the round trip, by playing a test sweep on an output and timing its return on an input; that needs a physical path between the two (a loopback cable, or a microphone that hears the speaker), and without one the result is shown as unavailable.
- Multiple audio backends: on Windows, WASAPI (shared and exclusive) and ASIO on the native engine, with the PortAudio backends (WDM-KS, DirectSound, MME) available as a fallback; ALSA and JACK on Linux; CoreAudio on macOS. Exclusive mode falls back to shared mode per device when it is refused, and the application tells you which it obtained. ASIO support is a separately licensed component; see section 3.
- VST3 effect plugins on Windows that you already own: a browser that scans the standard folders and yours, each module in a separate process so that a broken one cannot take the application down; insert chains on each device’s input and output; parameters, bypass and the plugin’s own editor; and presets that carry the chain and each plugin’s state; the same from the REST API and the command line. VST3 instruments, played by MIDI from a MIDI input port, an on-screen keyboard, the REST API or the command line. Proven with the project’s own test plugin and two third-party plugins (Surge XT and Dexed); see section 4.
- Per-application audio capture on Windows, capturing one process’s output through the operating system’s process-loopback interface, on Windows 10 build 20348 and later, and whole-system capture of everything an output plays. No driver and no virtual cable is required for either.
- Detection of which applications are currently producing sound.
- A Linux audio sink that other applications can select, created through PulseAudio or PipeWire and bridged into the audio engine as an ordinary output.
- A macOS CoreAudio plug-in publishing a loopback device, on the terms set out in section 4 below — which are narrower than the others in this list, and deliberately so.
- A local REST API with a WebSocket statistics feed, and an interactive command-line interface, in addition to the desktop interface.
- An optional network streaming feature, described in section 5.
The project’s README.md and its roadmap table are the authoritative statement of which
features are complete and which are in progress at any given version. Where this document
and that table disagree, the table is newer and the table is right.
3. What ToneSphere does not provide
This section exists so that you do not have to discover these by trying them.
- No accounts, no cloud, no hosted service. There is no sign-in, no online workspace, no server-side storage, no cross-device sync, and no online licence check. ToneSphere runs entirely on your machine.
- No virtual audio device that other applications can select, on Windows, in any build we distribute. The application’s buses are summing points inside its own process; nothing outside ToneSphere can see them. A kernel-mode driver publishing such devices (“virtual cables”, which the application can add, rename, disable and uninstall) exists in the project for development and testing only: it has been shown to work in a test virtual machine, not on an installed system, and it cannot be offered to users until Microsoft signs it. The virtual audio driver notes say exactly where it stands.
- No plugin hosting outside Windows, and no AU plugins. No commercial plugin has been tested, and none is represented as compatible. Plugins run inside ToneSphere’s own process: a plugin that faults while processing audio is bypassed, but a plugin can still bring the application down.
- No audio interface’s own ASIO driver tested. ASIO on Windows is provided by a component built from Steinberg’s ASIO SDK under the GNU GPL version 3, and a build that includes it is distributed under that licence (section 3 of the Terms and Conditions). It has been verified with FlexASIO, a genuine but software ASIO driver, and with ASIO4ALL, a third-party driver, both driving a USB audio interface; no interface manufacturer’s ASIO driver has been tested. The Microsoft Store package does not include ASIO at all; it runs on WASAPI, shared and exclusive.
- No recording to file. ToneSphere routes and monitors audio; it does not capture it to disk. If you want to record, record in your DAW or recorder of choice, downstream of ToneSphere.
- No guaranteed latency, dropout-free operation, or plugin compatibility. These depend on your hardware, drivers, buffer size, backend and plugins. See section 8 of the Terms and Conditions.
4. Platform features and their evidence
ToneSphere is offered for Windows, Linux and macOS, and the platform-specific features differ in how well they are established. We would rather label that difference than average it away:
| Feature | Platform | Status |
|---|---|---|
| Audio routing and mixing between devices | Windows, Linux, macOS | Working and covered by automated tests |
| Routing through an internal bus from one device to another | Windows | Working; covered by automated signal tests of the native engine |
| Routing through an internal bus from one device to another | Linux, macOS | Works; tested on Linux through a PulseAudio server. On macOS it runs the same code but has not been tested with real devices |
| Built-in effects | Windows | Working, from the interface; proven by tests that drive the interface’s own controls and measure what the effect did to a known signal |
| Built-in effects | Linux, macOS | Covered by automated tests; not reachable from the interface |
| Per-application capture | Windows 10 build 20348 and later | Working, proven by a test that captures a known tone and measures it |
| Whole-system capture of an output | Windows | Working, proven by capturing two known tones played by other programs and measuring their levels |
| VST3 effect plugins | Windows | Working with the project’s test plugin and with Surge XT, proven by tests that measure what the plugin did to a known signal; other plugins untested |
| VST3 instruments played by MIDI | Windows | Working with Surge XT and Dexed, proven by tests that play a note and measure its pitch; MIDI from a hardware port untested (no device) |
| ASIO | Windows | Verified against FlexASIO and ASIO4ALL, third-party ASIO drivers; no interface manufacturer’s driver |
| Measured round-trip latency | Windows | Proven on a digital path, through an interface’s cable and acoustically on the project’s own hardware; a figure for your hardware depends on a physical path existing between the output and input you choose |
| Selectable virtual sink | Linux, with PulseAudio or PipeWire running | Working |
| Selectable virtual device | Windows | Not provided — a development-only driver exists, proven in a test virtual machine only; see section 3 |
| “ToneSphere Audio” loopback device | macOS | Proven in continuous integration only. The plug-in builds, installs and round-trips a measured tone on a macOS CI runner. Its behaviour in day-to-day use — appearing in System Settings and in other applications’ device pickers, surviving sleep and wake, and installing past Gatekeeper on an end user’s machine — is unverified. Treat it as untested in real use until the project says otherwise. |
Features in that last row are provided on that basis, and no representation is made that they work outside the conditions described.
5. The optional network streaming feature
ToneSphere can send audio to, and receive audio from, another machine over a network. This feature is part of the application, not a service we operate, and these terms apply to it:
- It is off by default and entirely user-initiated. Nothing is sent until you enable the feature and enter the address and port of the peer you want to reach.
- We operate no server, relay or intermediary. The connection is between your machine and the peer you nominate. Your audio does not pass through us, is not stored by us, and cannot be observed by us.
- The transport is unencrypted and unauthenticated. Audio is carried as PCM, or compressed with Opus, in UDP datagrams with a binary header, sequence numbers and a jitter buffer that sizes itself to the measured network jitter; or over TCP in either direction. There is no cipher and no verification of the peer. Anyone who can observe the network path could listen to the audio, and anyone who can reach the port you are listening on could send audio into it.
- Use it on networks you trust. It is designed for monitoring across machines you control, on a network you control. It is not a hardened remote-collaboration service, and it should not be exposed to the public internet.
- You are responsible for what you send. Sending audio to a peer is a transmission you are making. Section 4 of the Terms and Conditions applies.
The local REST API and WebSocket feed are subject to the same caution: they bind to
127.0.0.1 by default and have no authentication, so changing that bind address exposes
control of your audio engine to anyone who can reach the port.
6. Availability
There is nothing to be up or down. ToneSphere provides no hosted service, so we make no uptime, availability or continuity commitment, and none is needed: an installed copy keeps working whether or not we are reachable, whether or not the project’s repository exists, and whether or not development continues.
The only network activity is what you configure yourself under section 5, and its availability depends on your network and your peer, not on us.
7. Support
Support is provided on a best-effort, no-commitment basis through the project’s issue tracker:
https://github.com/AvishakeAdhikary/tone-sphere/issues
- There is no service-level agreement, no guaranteed response time, and no undertaking that a given issue will be investigated, reproduced or fixed.
- There is no private or paid support channel, and no support channel operated by us other than the issue tracker. Sponsorship does not create one — see section 10.
- Issues are public. Do not include confidential information, credentials or licence keys in
one. Diagnostic output from
main.py testand your ToneSphere logs is the useful thing to include, and both are local files you can review before posting.
8. Updates and versions
- Releases are published to GitHub Releases when a version tag is pushed, and are built and smoke-tested on Windows, Linux and macOS before publication. The current downloads are at https://github.com/AvishakeAdhikary/tone-sphere/releases/latest.
- A copy obtained from the Microsoft Store, once ToneSphere is available there, is updated through the Store according to your Windows update settings. A copy downloaded from GitHub Releases does not update itself, and the application performs no update check of its own.
- We may add, change, deprecate or remove features between versions, and are under no
obligation to maintain any particular feature indefinitely. Changes that affect what the
application claims to do are reflected in the
README.mdroadmap. - Configuration files and presets are plain YAML and are read tolerantly, but compatibility of a preset across versions is not guaranteed. Keep a copy of a preset that matters to you.
- We may discontinue distribution or development at any time. If that happens, copies already installed continue to work under the Terms and Conditions.
9. Distribution through the Microsoft Store and GitHub
ToneSphere is distributed for Windows, Linux and macOS through GitHub Releases, and is being prepared for distribution for Windows through the Microsoft Store. The terms in this section apply to a copy acquired from the Store once it is available there.
- Acquisition, installation, updates, ratings, refunds and any billing relating to a copy obtained from the Microsoft Store are handled by Microsoft under Microsoft’s own terms, including the Microsoft Store Terms of Sale and the Standard Application License Terms. Those terms apply to the acquisition in addition to this document, and where they conflict with this document in respect of the acquisition itself, Microsoft’s terms govern it.
- Microsoft is not a party to this agreement, does not endorse the Application, and has no obligation to support it. Support is as described in section 7.
- Downloading from GitHub is subject to GitHub’s terms. GitHub is likewise not a party to this agreement.
- Information handling by both channels is described in section 6 of the Privacy Policy.
10. Price, sponsorship and donations
ToneSphere is currently offered at no charge, and every feature it has is available without paying for it. No function is withheld, delayed, time-limited, nagged for, or gated behind a payment or a donation prompt.
Future versions or editions may be offered for a price, through the Microsoft Store or another channel. That does not change the terms of a copy you already have, and a price is always shown before anything is charged; section 3 of the Terms and Conditions sets this out.
You can support the project’s continued development through GitHub Sponsors, Patreon, Ko-fi or Buy Me a Coffee. If you do:
- The contribution is voluntary and is a contribution to development, not the purchase of a product, a licence, a subscription or a support contract.
- It grants no additional rights, features, priority, warranty or service commitment, and creates no obligation on us to deliver anything specific.
- Payments are processed by that platform under its own terms, and refunds, cancellations and billing questions are handled by that platform, not by us.
11. Changes to this document
We may update these Terms of Service, in particular as features move from in progress to working, or as platform support changes. Every version carries a version number and an effective date at the top of the document, and the current version is published at the address where you are reading it. Continued use of the Application after an updated version takes effect is acceptance of it. Superseded versions remain in the project repository’s history.
12. Governing law
These Terms of Service are governed by the laws of India, and the courts at Kolkata, West Bengal, India have exclusive jurisdiction, on the terms set out in section 13 of the Terms and Conditions.
13. Contact
https://github.com/AvishakeAdhikary/tone-sphere/issues
That is the project’s contact channel for everything in this document. Issues are public, so do not include anything confidential in one.