OneMove Vocal Console: technical specifications
Fourteen modules. One move each. A complete vocal chain in one AUv3 plug-in for iPad and iPhone. Every figure on this page was measured. The method and raw reports are in our research notes.
Format
| Plug-in | AUv3 Music Effect: audio in and out, MIDI in (Tune's target notes and reference tone), host side-chain |
| Devices | iPad and iPhone (a layout for each: portrait and landscape) |
| Channels | mono and stereo |
| Sample rates | 44.1 to 96 kHz, every engine including neural Clean (tested at 44.1, 48, 88.2 and 96 kHz) |
| Processing | 32-bit floating point throughout |
| Side-chain | the host's side-chain bus, or Key Link: music from another track, in hosts without side-chain routing |
| Host automation | every control is a host parameter (207 in all) |
| In-app console | try it without a host: two demo vocals, or your microphone with headphones; A/B bypass; chain presets |
Signal path
| Everything off | bit-exact: the output is the input, delayed by exactly the reported latency |
| Added noise | none. Digital silence in gives digital silence out with all 14 modules on, in both modes |
| Bypass | the host's bypass (and the console's) passes the dry signal at the same latency: switching never jumps in time |
| Latency reporting | exact to the sample, so your host lines the vocal up perfectly |
| Analogue figures (impedance, converter noise, headroom) | set by your audio interface: the plug-in has no analogue stage |
Modes and latency
This is the plug-in's own latency, reported to the host. Round-trip latency also includes your host's buffer and your audio interface.
| mode | Clean engine | plug-in latency (48 kHz) | use it for |
|---|---|---|---|
| Live | Spectral (default) | 191 samples, 4.0 ms | tracking and performing with monitoring |
| Live | Zero-latency or Neural 0 ms | 63 samples, 1.3 ms | the lowest possible delay |
| Studio | Neural (default) | 2,958 samples, 61.6 ms (host-compensated) | mixing: the best noise reduction and pitch correction |
| Studio | Classic | 1,550 samples, 32.3 ms | mixing with lower delay |
CPU on iPad Air M3 (13")
Percent of one core, at 48 kHz with 256-frame buffers, measured 7 November 2026. Method and raw report. The full-chain figures were measured before Space was added; Space adds about 0.3–0.4 % of one core on the Mac (5.2 → 5.6 % there); the iPad will be re-measured.
| configuration | audio thread | total |
|---|---|---|
| Full 14-module chain, Studio, Classic Clean | 5.7 % | 5.7 % |
| Full 14-module chain, Live, Spectral Clean | 5.2 % | 5.2 % |
| Full 14-module chain, Studio, Neural Clean | 5.6 % | 10.7 % |
| Full 14-module chain, Live, Neural 0 ms Clean | 6.6 % | 11.6 % |
| Default chain (8 modules), Studio, Neural Clean | 3.3 % | 8.3 % |
| One module on its own | 1.0–2.4 % | |
| Everything switched off | 0.6–1.0 % |
Built to stay cool
- The neural model runs on its own thread. Your host's DSP meter sees only the light filtering on the audio thread, and a late frame never glitches: it plays dry.
- Idle modules cost almost nothing. A module that's off keeps only its fixed delay, so switching it never changes the latency.
- The interface draws only what you can see. Hidden panels and off-screen strips are taken out of the view tree, and meters refresh only while the plug-in is on screen, at a rate you choose (15, 30 or 60 frames per second). Opening the plug-in on an iPhone 15 Plus takes under a second.
Not measured yet: heat and battery over long sessions, and older devices. We'll publish them when we have them.
The 14 modules
| # | module | One Move knob | in short |
|---|---|---|---|
| 1 | Clean | Suppression | noise reduction: neural (Studio), spectral or zero-latency (Live) |
| 2 | Tune | Natural ↔ Hard | real-time pitch correction: key and scale, Auto Key, Humanize, MIDI target notes |
| 3 | Tame | Tame | four-band dynamic EQ for mud, box, harshness and sibilance |
| 4 | Shape | Shape | corrective EQ: high-pass, mud, box and harshness cuts |
| 5 | Level | Peak Reduction | 2A-style opto leveller |
| 6 | Punch | Drive | 76-style FET limiter |
| 7 | Tape | Tape | tape saturation and compression |
| 8 | Floor | Floor | downward expander that keeps the gaps between phrases quiet |
| 9 | Ride | Ride Range | automatic vocal rider that follows the music (side-chain or Key Link) |
| 10 | Tube | Tube | valve harmonics |
| 11 | Character | Character | voice transformation: gender, age, giant, robot |
| 12 | Air | Air | dynamic high-frequency lift |
| 13 | Loud | Loud | perceived loudness with peak shaving |
| 14 | Finish | Finish | final tone EQ, plus Space: a plate reverb and a doubler (zero latency; on in most chain presets) |
Studio closes with a true-peak limiter. The order is free: move any module with ◂ ▸.
Pitch correction (Tune)
| Studio | Live | |
|---|---|---|
| Hard: settles on a new note | 15 ms | 28 ms |
| Hard: note-to-note change | 35 ms | 46 ms |
| Steady pitch error on held notes | 0.1 cent | 0.1 cent |
| First usable pitch reading after a note starts | 15.5 ms median (537 real onsets) |
Natural keeps the singer's vibrato (80–96 % at 50–100 cents depth) and never exaggerates it.
Noise reduction (Clean)
| engine | noise removed in pauses (our test set) |
|---|---|
| Neural (Studio) | 37–40 dB on hiss, fan, clicks and traffic; 21 dB on background voices |
| Neural 0 ms (Live) | 8–28 dB |
| Classic Studio | 1.5–10 dB |
The neural engine uses RNNoise (BSD-3, Xiph.Org), running on the device, with OneMove's attack protection.
Compressors (Level and Punch)
Static curves are within 0.19–0.43 dB RMS of measured reference curves at every ratio and mode, and neither adds latency. Details: Level and Punch.
Measured 7–8 November 2026 on iPad Air M3 13" (iPadOS 27.2), iPhone 15 Plus and the development Mac. "2A-style" and "76-style" describe circuit behaviour; OneMove is not affiliated with the owners of any hardware trademarks.