Research
How OneMove Vocal Console works, and how we know. Every number on this site comes from a measurement, with its device, settings and date. These notes show the method, so you can check them.
Pitch correction against a measured contract
We measured what a pitch corrector must do on test signals and real takes, wrote it down as a pass/fail contract, and rebuilt OneMove Vocal Console's Tune until it met it. Method and results.
Pitch detector benchmark
How quickly can a pitch corrector know which note you are singing? We benchmarked YIN layouts, McLeod, aubio, CREPE and PESTO on 537 real note onsets. Our detector gets a usable reading in 15.5 ms.
Neural noise reduction on singing
A small neural noise-reduction model against classic spectral suppression on real vocals with five kinds of noise: 37–40 dB removed in pauses. Method, results and limits.
Latency explained
What plug-in latency is, why OneMove Vocal Console has a Live mode at 1.3–4 ms and a Studio mode at 32–62 ms, and why pitch correction can never be instant.
A full vocal chain on an iPad Air M3 at ~6 % of one core
How much CPU each OneMove Vocal Console module uses on iPad Air M3, how we measured it, and why one plug-in with 14 modules costs less than a stack of separate ones.
Level and Punch: how we built an opto leveller and a FET limiter for iPad
How OneMove Vocal Console's Level (2A-style opto) and Punch (76-style FET) work: what the original circuits do, what we modelled, and how close they measure.
Vocal chain order
A practical order for vocal plug-ins on iPad and iPhone: clean first, then pitch, EQ and dynamics, colour, air, loudness and space, with the reason for each step and when to break the rule.
How we test claims
- Measured, or not said. If we haven't measured it (heat, battery, older devices), we say so instead of guessing.
- Dated and specific. Each figure names the device, the sample rate and the settings.
- Fair references. When we compare with another product, we measure it the same way, name it only for comparison, and say where it does better.
More in our post how we test claims.