Splitwave

Name: Splitwave — Download free. Audio router
Availability: InStock
Rating: 5 (2 reviews)
Author: Horuse

Splitwave is a node-based audio routing application for macOS, Linux, and Windows that lets you build custom audio pipelines visually. You connect microphones, system audio, individual applications, and audio files into a graph, process the signal through effects such as equalization, compression, reverb, and limiting, host third-party CLAP, VST3, and AU plugins, and then direct the result to speakers, recording files, or virtual audio devices that other applications can use as inputs. The graph operates in real time, and you can rewire connections while audio continues to play. On macOS and Linux, Splitwave installs a driver and creates named virtual audio cables that the operating system treats as real hardware. On Windows, you connect Splitwave to VB-Cable or a comparable virtual audio cable. Any application can play into a virtual device, and any application can select one as its microphone, with both ends appearing inside your graph.

★★★★★

5.0(2 ratings)

Download Splitwave (Official links)

File size: 19.8 MB

The latest version of Splitwave is: 1.4.0

Operating system: Windows, MacOS, Linux

Languages: English

Developer: Horuse

Price: $0.00 USD (Open Source (MIT))

Node-based pipeline construction. Splitwave replaces the fixed channel strip layout of a traditional mixer with a canvas on which you place nodes and draw connections between them. Input nodes represent audio sources, effect nodes process the signal, and output nodes define where the audio goes. Each node exposes handles for incoming and outgoing connections, and you join an output handle of one node to the input handle of another. The graph executes continuously, and you can add, remove, or reconnect nodes while audio is running. Pressing Activate starts the pipeline, and pressing Stop tears it down.
Microphone capture. The Microphone node records audio from any input device the operating system exposes, including built-in microphones, USB interfaces, and headsets. The node requests microphone permission through the standard system prompt on each platform. Once captured, the signal enters the graph and can be routed through any combination of effects, merged with other sources, or split to multiple destinations at the same time.
System audio capture. The System Audio node captures everything the computer is currently playing. On macOS 13 Ventura and later it uses ScreenCaptureKit, which Apple introduced for this purpose. On Windows and Linux it uses the system's own loopback capture mechanism. The captured stream becomes a source inside the graph, so you can process the entire system output, record it, or forward it to another application through a virtual device.
Per-application audio capture. The App Audio node targets the sound of a single running application instead of the whole system. This separation lets you treat one program differently from the rest: route a browser through noise suppression, send a media player into a recording file, or pass a game through equalization while leaving other audio untouched. Each captured application becomes its own source node with independent wiring.
File playback input. A File node reads an audio file from disk and feeds it into the graph as a source. This makes it possible to process prerecorded material through the same effect chain used for live audio, mix a backing track with a microphone, or run a file through a plugin chain and record the result without involving any external player.
Virtual audio devices. A virtual device is a software sound card that other applications can select as an input or an output. On macOS and Linux, Splitwave installs a userspace audio driver and creates as many virtual devices as required, each with a name you choose. On Windows, Splitwave does not create these devices itself; you install VB-Cable or a similar virtual audio cable and Splitwave uses it like any other device. Sending audio into a virtual output and selecting that same device as a microphone in Zoom, Discord, OBS, or a DAW delivers your processed signal to those programs as though it came from real hardware.
Built-in effect nodes. Splitwave includes a set of processing nodes that cover common audio tasks. The equalizer shapes frequency response, the compressor reduces dynamic range, the gate silences quiet passages below a threshold, the limiter prevents peaks from exceeding a ceiling, the reverb simulates acoustic space, the delay produces echo effects, and the saturator adds harmonic coloration. Each of these nodes stores named parameter presets, and Splitwave ships with built-in presets that can be selected but not overwritten.
Noise suppression. A dedicated noise suppression node uses the DeepFilterNet model to remove background noise from a signal. The model operates internally at 48 kHz, and Splitwave resamples audio in and out automatically when the pipeline runs at a different sample rate. This node is placed in the chain like any other effect, so it can be applied to a microphone before the signal reaches a virtual device, a recording file, or a broadcast application.
Third-party plugin hosting. Plugin nodes load installed CLAP, AU, and VST3 plugins and insert them into the graph as effects. You point a Plugin node at a plugin on your system and wire it into the chain alongside built-in effects. The plugin's own editor window opens from the node, and its parameters can be exposed and automated directly on the node without opening the editor.
File recording output. The File Recording node writes the incoming signal to disk while the rest of the graph continues playing. It supports lossless WAV, FLAC, and AIFF output as well as encoded Opus, MP3, and AAC streams. Because the node accepts any connection, you can record a processed microphone, a system audio capture, a mix of several sources, or the final output of the chain. Recording can run at the same time as monitoring through speakers, headphones, or virtual devices.
Network streaming nodes. Splitwave includes a WebRTC node and a UDP sender node for transmitting audio over a network. Both require you to configure the destination yourself, and no audio leaves the machine unless one of these nodes is deliberately added to the graph and pointed at a peer. No account, analytics service, or cloud rendering is involved in any part of the processing.
Sample rate conversion. Splitwave resamples audio between devices that run at different sample rates. A source operating at 44.1 kHz can feed an output running at 48 kHz without manual configuration. The same conversion applies inside the graph, including around the DeepFilterNet noise suppressor, which runs its model at 48 kHz regardless of the pipeline's rate.
Pipeline and preset management. Named pipelines store complete graphs, including every node, connection, and setting, so you can keep separate configurations for different tasks and switch between them. Effect nodes additionally maintain their own named presets, which are independent of the pipeline. This two-level arrangement means a compressor setting can be reused across different graphs without rebuilding the surrounding chain.
Platform support. Splitwave runs on macOS 13 Ventura and newer on both Apple Silicon and Intel processors, on Linux with an x86_64 architecture and a PipeWire audio session, and on Windows 10 and 11 in 64-bit versions. The macOS installation uses a small userspace audio driver rather than a kernel extension, so no System Integrity Protection workaround is required. Administrator approval is requested once during installation and again whenever a virtual device is added or removed, because macOS requires it for changes to the system audio configuration.
Permissions. Splitwave requests only the permissions associated with the sources you actually use. On macOS this means microphone access for microphone capture, screen recording for system and per-application audio capture, and disk access when a recording destination is chosen. Windows and Linux request microphone access through their own system prompts in the same manner. Permissions for unused sources are never requested.
Local processing and privacy. All audio processing takes place on the local machine. There is no account system, no analytics collection, and no cloud rendering. Audio leaves the computer only when a network node is added to the graph and configured with a destination. The application contacts GitHub to check for new releases, which is the only routine outbound request it makes.
Cross-application routing. None of the three supported operating systems provides a direct way to send one application's sound into another. Splitwave addresses this by capturing a single application with the App Audio node, passing the signal through any effects, and ending the chain at a virtual output device. The receiving application selects that device as its input, so the audio remains on the local machine and no additional hardware is required. Splitwave creates the virtual devices itself on macOS and Linux, while on Windows it works alongside VB-Cable or a comparable virtual audio cable.
DAW integration. A virtual output device created by Splitwave, or VB-Cable on Windows, can be selected as the input device in Logic Pro, Ableton Live, Reaper, GarageBand, or any other digital audio workstation. A DAW master output can also be routed back through Splitwave for live effects processing, which allows the graph to act as an effects insert for the entire mix.
Comparison with other routing tools. Utilities such as BlackHole and VB-Cable provide a plain virtual audio cable without processing. Loopback and VoiceMeeter add a mixer on top of the cable concept. Splitwave provides a full node graph across macOS, Linux, and Windows, with noise suppression, equalization, compression, third-party plugin hosting, recording, and network streaming included. It is distributed free under the MIT license and creates unlimited named virtual devices of its own on macOS and Linux.

Splitwave was created as an open source project and is distributed under the MIT license, with its source code hosted on GitHub. The application is written primarily in C++ with a Rust component for the audio engine and platform-specific driver code, while the node graph interface is built with modern web technologies embedded in a native shell. Development began in the early 2020s and the project has since expanded to cover macOS, Linux, and Windows with a unified node-based workflow. The development team consists of audio software engineers and contributors from the open source community who maintain the codebase, review pull requests, and publish regular releases.