Qune

Opinion

A tuner that can be heard doing its job or kept out of earshot. With Glide off and Amount at 100 % it is the hard, stepped correction everyone knows; with a glide of a couple of hundred milliseconds and Amount at 60 % it nudges a singer toward the key and leaves the vibrato. Switching Take notes from to MIDI turns it into something else: hold a chord and the voice lands on the nearest chord tone, or play a melody and the voice sings it.

The quality tiers are worth trying on the same take. Fast is a few milliseconds and sounds like a resampler; Good keeps formants and is clean on a single voice; Best is smoother on breath and noise and costs 43 ms. The pitch trace on the panel shows what was heard and where it was sent, which is the quickest way to tell whether Stickiness and Voicing are set for your source.

It follows one pitch. Feed it a voice, a bass or a lead line, not a chord.

Functionality

A normalised-square-difference pitch tracker (decimated search, full-rate refinement, octave-error checks) reads the sum of the two channels every 4 ms. The target note comes from the scale or from held MIDI notes; Amount, Glide, Stickiness and Transpose turn the difference into a shift, and one of three shifters applies it to both channels:

Unvoiced sound passes through unchanged. The dry path is delayed to match, and the latency is reported to the host and shown in the header.

MIDI. MIDI chord snaps to the nearest held pitch class in any octave; MIDI note sings the last note played. Pitch bend (range set on the panel) moves the target, CC 64 holds released notes, CC 120/123 clear them. When no key is held chooses between leaving the voice alone and falling back to the scale. No other CCs are mapped; automate the host parameters.

MIDI out. While a pitch is being corrected, Qune also sends the note it is moving it to as MIDI note-on/note-off, one note at a time and legato (the new note-on goes out before the old note-off). That makes it a voice-to-MIDI converter that snaps to the scale: route it to a synth. The note is the whole-semitone target including Transpose; pitch bend is not sent. It is timed when the pitch is detected, so it leads the delayed audio by the reported latency.

Parameters

Status

Core DSP, 190 000 deterministic checks, the VST3 target and the catalogue screenshot are complete. It has not been auditioned on real voices and has not been run in a host, so latency reporting, quality switching under automation and MIDI input are unverified outside the test rig.