Zita-at2 Autotuner LV2
----------------------

Zita-at2 is an 'autotuner' normally used to correct small pitch
errors of a vocal track. The main differences with the older at1
version is that at2 will try to preserve formants while changing
the pitch of a voice signal, and that it is now a LV2 plugin
rather than a Jack client.

Modes
-----

At2 can be used in four different 'modes', selected by the 'Auto'
and 'Midi' buttons. Note that there is no full 'bypass' mode as
that will normally be provided by the plugin host.

With both 'Auto' and 'Midi' off, the input signal is resynthesized
using the detected pitch without any correction apart from the
static pitch and formant offsets (see below). This can occasionally
be useful as an effect, in particular on speech. 

With 'Auto' on and 'Midi' off, the retuning logic will select the
nearest note enabled in the GUI. Octaves are ignored, so the largest
correction will be 6 semitones up or down. With 'Midi' enabled, the
logic is the same, but any MIDI input will override the static
selection in the GUI.

With 'Auto' off and 'Midi' enabled, the retuning will be to the
pitch of the most recent MIDI 'Note On' event, as long as that
note is active. Octaves are not ignored in this mode. 

The GUI keyboard
----------------

Click-L or -R to enable individual keys. Shift-Click-L to select
a major scale, -R  to select a minor one. When the retuner is
active, the selected note is displayed in blue, and the pitch
error is shown on the bargraph meter below (in cents).

The controls
------------

Rotary controls can be set by dragging left-right or up-down,
or by using the mouse wheel. In the latter case keeping Shift
pressed will provide smaller steps.

All controls can be automated if the host supports this.
Automation 'touch' is supported as well.

From left to right:

* MIDI channel selection: click-L or -R to cycle throught channel
  1-16 or 'Any'

* MIDI reset: clears any 'hanging' midi notes.

* Range: normally the lowest pitch that will be reliably detected
  will be around Eb2 (~78 Hz). Selecting the 'High' range increases
  this to Bb2 (~116 Hz) while reducing latency. Use this for female
  voices.

* Mode: see above.

* Threshold: any input below this level (dBFS) will be considered
  'unvoiced' and not be processed. This can be useful to remove
  erratic retuning of signals that leak into a vocal track.

* Tuning: sets the frequency of A4, normally 440 Hz. 

* Bias: in 'Auto' apply some preference to the current note
  by allowing it to deviate more then half the distance to
  the nearest enabled note.

* Speed: controls a lowpass filter on the retuned pitch.
  This emulates the finite speed at which a singer can
  control his/her vocal cords. The filter is reset after
  100 ms of unvoiced signal.

* Modulation: controls the amount of pitch variation on
  continuous long notes. Any such modulation is faded in
  250 ms after a unvoiced to voiced transition.

* Pitch: add a static pitch change, up to one octave up
  or down.

* Formant: add a static formant shift, up to 4 semitones
  up or down.


  
 
