How can I reduce CPU use from vocal plug-ins on iPad?
Use fewer plug-ins, raise the buffer size when you aren't singing through the chain, and switch off stages you don't need. Each AUv3 runs in its own process with its own render call and interface, so a chain of separate plug-ins pays that overhead several times. OneMove Vocal Console pays it once: its full 14-module chain uses about 6 % of one core on an iPad Air M3. We make OneMove Vocal Console.
What to try, in order
- Remove stages you don't need. A clean recording may not need noise reduction at all.
- Check the reverb. In our Mac tests, the choice of reverb made most of the difference between desktop stacks.
- Raise the buffer size while mixing. Lower it only while you sing through the chain.
- Freeze or bounce finished tracks, if your host offers it.
- Replace several plug-ins with one if you can.
OneMove Vocal Console's measured CPU
| configuration, iPad Air M3, 48 kHz, 256-frame blocks | percent of one core |
|---|---|
| Full chain, Studio, Classic Clean | 5.7 % |
| Full chain, Live, Spectral Clean | 5.2 % |
| Full chain, Live, Zero-latency Clean | 6.5 % |
| Full chain, Studio, Neural Clean | 5.6 % audio thread, 10.7 % total |
| One module on its own | 1.0–2.4 %, including about 1 % of fixed overhead |
| Everything off | 0.6–1.0 % |
These figures were measured before Space was added. On the Mac, Space adds about 0.3–0.4 % of one core. Method and raw report.
Questions
Will fewer plug-ins stop my iPad getting hot?
We don't know. We have not measured heat, throttling or battery, so we make no claim about them.
How does this compare with a stack of separate plug-ins?
On a Mac (Apple M4 Pro), the full chain used 6 % of one core, against 9–22 % for desktop stacks doing the same jobs. These are Mac numbers, not iPad numbers. Plug-ins, settings and method.
What we have not measured
Older iPads, the iPhone, heat and battery.