This week's releases: Arturia's first dedicated bass synthesizer, a broadcast-grade voice isolator from CEDAR Audio, AudioThing's Euclidean drum machine with cross-platform reach, and a free four-knob character dynamics plugin from an independent developer that has been circulating quietly in producer communities.
Pure SUB, Arturia
$49 intro (regular $79, intro ends September 15)
Most synthesizers designed for bass are general-purpose instruments adapted for low-end work. Pure SUB, released August 27, is built specifically for it. The architecture separates bass generation into three independent layers that process in parallel before merging at the filter stage: Sub, Harmonics, and Texture.
The Sub layer handles the fundamental. You control level, phase, and octave independently, and Arturia's Sub Regen technology keeps the fundamental phase-stable after processing, which matters once you start driving the signal into the downstream filter or effects chain. The Harmonics layer offers two synthesis modes: Pressure for lower-density waveform editing, and Reese for high-density unison with up to several detuned voices stacked for that characteristic wide bass texture. Texture adds the rougher material: ring modulation, amplitude modulation, noise sources, or sample-based layers that bring grit and movement without touching the Sub or Harmonics signals.
After the layers merge, 40+ filter modes handle tonal shaping, and a 12-slot effects rack (18 effects) covers distortion, modulation, reverb, and compression downstream. Three envelopes and six modulator slots handle both standard velocity-to-amplitude routing and more experimental signal paths.
This layered design lets you dial in a clean sub without harmonics muddying it, add Reese spread only to the midrange partials, and place texture only in the upper content, then blend all three through the same filter stage. It is a more surgical way to build 808s, sub patches, and stacked bass tones than adapting a Moog-style signal path after the fact.
Runs on Windows 10+ and macOS 11+ in VST, VST3, AU, and AAX formats. No Linux support. 300 factory presets included.
For whom: producers building electronic bass sounds who want explicit control over what happens at each frequency register.
Voxis, CEDAR Audio
$269
CEDAR Audio has been building audio restoration tools for broadcast and archival work since the 1990s. Their hardware and software appear in post-production facilities and broadcast chains where noise reduction has to work reliably and without artifacts. Voxis, released August 27, brings that approach into a DAW plugin format with one focus: real-time voice isolation.
The neural network inside Voxis is trained to recognize speech and sung voice and separate them from background noise, air conditioning rumble, traffic, audience wash, and excess room reverb. Its key differentiator from tools like RX Dialog Isolation is latency. Voxis specifies under 10 milliseconds, which makes it usable in live environments, front-of-house vocal chains, live broadcast, and conference AV where latency is a constraint. Offline or DAW-buffered approaches cannot handle those applications.
The interface is minimal: three frequency-band controls (low, mid, high) for adjusting processing intensity per range, a Link button that locks all three bands together for consistent tonal balance, and a Voice control that raises the isolated vocal in the output. The band controls adjust how aggressively the network targets low rumble, midrange noise, and high-frequency hiss separately, which is useful when noise character varies across the spectrum.
A 14-day free trial is available from the CEDAR website. Runs on macOS and Windows. Price is $269 before local tax.
For whom: live sound engineers processing vocals in noisy environments, broadcast and podcast engineers dealing with persistent background noise, anyone needing real-time voice isolation rather than offline processing.
Orbita, AudioThing
$39 intro (regular $69, intro ends September 30)
Euclidean sequencing works by distributing a chosen number of hits as evenly as possible across a pattern of steps. Twelve steps with five hits produces a different spread than twelve steps with seven. The mathematical distribution creates rhythms that feel irregular without being random, because they are symmetrical in a way the ear reads as groove rather than programmed pattern.
Orbita, released August 28, builds an eight-track drum machine around this algorithm. Each track gets independent control over Steps (pattern length), Hits (how many triggers to distribute), Accents (emphasis on specific hits), Offset (rotating the pattern around the cycle), Swing (timing nudge), and Reset (when the pattern snaps back into alignment). When tracks run at different lengths, they drift in and out of phase with each other, generating polyrhythmic relationships automatically without requiring a separate composite pattern.
Each track loads samples from 36 included factory banks or from custom banks built from your own files. Per-track filtering covers low, high, band, and notch types in 2-pole and 4-pole variants, and an LFO per track can modulate pitch, pan, volume, filter cutoff, or resonance. A pad mode lets you trigger tracks manually via MIDI.
Platform support is unusually broad. Orbita runs on macOS (AU, VST2, VST3, AAX, CLAP), Windows (VST2, VST3, AAX, CLAP), and Ubuntu 20.04+ (VST2, VST3, CLAP). An iOS and iPadOS version is available on the App Store as AUv3 and standalone.
For whom: producers exploring polyrhythmic drum programming, live performers who want controlled rhythmic variation between patterns without programming every combination by hand.
PUNCHDUST, mfkne (free)
mfkne has been releasing small, technically specific free plugins over the last year or so. PUNCHDUST, released in late August, combines three functions into four knobs: a transient shaper (PUNCH), a reactive lo-fi texture generator (DUST), and an upward compressor (SQUEEZE).
PUNCH runs through a dual-stage transient shaper into a soft clipper at 2x oversampling, using a 97-tap linear-phase half-band filter. The oversampling and linear-phase filter keep boosted transient content clean, avoiding the aliasing artifacts that cheaper transient tools introduce on high-frequency percussion.
DUST blends 5-bit rasterized digital crackle with warm analog hiss. Unlike a static noise layer, the texture reacts to the input signal: it rises as the source falls away, so the crackle and hiss appear in the decay tail of each hit rather than as a constant blanket over the signal.
SQUEEZE applies upward compression to the sustain, lifting quieter content by up to 26 dB while explicitly excluding transients from the compression detector, so the transient punch stays intact as sustain density increases. A mono sum detector maintains stereo imaging through the transient shaping stage.
Latency is 1.0 ms at 48 kHz. The plugin includes a built-in noise gate, a 15 Hz subsonic filter, and a master safety soft clipper. Community reception has been consistently positive, particularly for drum bus and room mic applications where adding character and punch without a separate processing chain is the goal.
VST3 only, macOS (Universal Binary, Apple Silicon and Intel) and Windows 64-bit. Free with no restrictions.