Take Settings

The Take parameter keeps your most recently recorded or imported MIDI data ready for Figure Recognition, which will convert it to a Figure. After adjusting your settings, press Apply to start the recognition process. The result will be stored in the Figure parameter of the same phrase.

Figure Recognition

Figure Recognition turns flat MIDI data in a Take into a highly structured Figure. This is a complex procedure that involves a lot of estimation. If there are enough notes, the Auto button can estimate the settings and provide a good starting point. Although sometimes you need to experiment a bit until you find the settings that work best.

Both Harmony and Playing Ranges have a big impact on figure recognition. Be sure to re-harmonize an imported phrase if you suspect harmony does not match the take. Also, when you are recording in an arrangement, make sure you play in the current key.

Tipp: You can select a span in a Take and run figure recognition on it selectively.
Preset
Pick a preset that best suits the Take. Don't take the label "Keyboards" literally. It can be any polyphonic instrument, actually.
Algorithm
Select a method and strategy for recognizing figures.
  1. Auto: Estimates an appropriate algorithm automatically.
  2. Linear: Examines the input from left to right, detecting reoccurring patterns and characteristic movements on the way. Works best for monophonic melodic voices.
  3. Pattern Recognition: By looking at the take as a whole, this method identifies areas on interest and compares millions of potential segments with each other. A time-intensive but effective method that does a good job recognizing polyphonic phrases.

  4. Rhythm Input: Every single recorded note is turned in to a chord or bass note. This is great if you simply want to tap-in a rhythm that is supposed to be played as chords or bass.

  5. Simplified Bass: Converts the input to plain isolated bass symbols only, which works fine for many popular music styles. For contrapuntal and melodic bass voices, you should use Linear.
  6. Static Pitch: Converts the take to absolute pitch symbols. This is extremely fast and good for drums and percussion, or other instruments that don't follow Harmony. You can also use this to import large MIDI files more quickly and postpone figure recognition until after the import on a track by track basis.
Segment Types
Select the desired symbol types you want to generate. Some algorithms allow only a single type. The Linear algorithm supports multiple types only when voices are separated.
Accidentals
Create symbols with chromatic components where they are needed to faithfully recreate the input. Disable this to get more portable phrases that are easier to edit.
Ranges
When enabled, this assigns a specific playing range (lower, upper) to segments if they are far off the middle range. This moves more segments towards the center, visually, but also helps a phrase translate better to other instruments that might have very different ranges. When this is disabled however, the resulting Figure is more easily editable.
Resolution
Notes starting within this window are considered simultaneous. This setting merely affects the grouping of symbols into segments and does not alter the timing of your result.
Number of Voices
How many parallel voices are assumed to be in the Take. This is for information only unless you choose to separate them.
Separate
Separate voices from each other before processing them. This can much improve results when voices are sufficiently apart from each other. Where the input is an entangled mess however, it can also make things worse. If in doubt, try both options and see what works best.
Anmerkung: Voice separation works better on isolated phrases than entire tracks worth of imported MIDI data, which may contain multiple sections with contradicting properties.
Transpose
Transpose the input by octaves in order to best match the target instrument's playing range.
Sustained
The sound is sustained while a note remains depressed. Consecutive notes that overlap each other are meant to be independent voices. Disable this to better handle decaying or percussive sounds like guitar and piano.
Simplify Bass
Place bass segments around the zero line, so they will translate to whatever is the current bass, rather than attempt to recreate the original input. Makes a phrase more portable.
Simplify Chords
Generalize chords such that a phrase translates better to different Harmony.
Apply
Start Figure Recognition. Depending on complexity this can take a while. A log console will open if the process turns out to be running longer than expected. Results are saved o the Figure parameter. After this you can optimize the settings and repeat figure recognition until you are satisfied with the result.
Terminate
Prematurely end the recognition process if it takes too long, or if you want to try different settings.

Tags

You can help Figure Recognition achieve better results by grouping symbols into segments or tagging their supposed types beforehand. In the Parameter View of Take, simply group segments and change their symbol types as you would do with a Figure.

Recording Options

Under the tab with a red circle, you'll find several options that will alter your input before it is processed by Figure Recognition.

Quantize
Select a grid to destructively snap all notes to immediately after recording.
Monophonic
Reduce recorded input to a single voice before processing it.
Overdub
Don't clear the phrase before recording, that is, add whatever is recorded to the existing Take.