How WIDI works

WIDI is a complex system and it contains different algorithms for different steps of music recognition process.

There are two types of algorithms - Note Detection algorithms and Supplementary algorithms.

Current WIDI contains three different Note Detection algorithms - they are Euclid, Drum Detection and Advanced sensors. Their task is to extract actual note data. Euclid and Advanced Sensors are general-purpose algorithms, and Drum Recognition performs only recognition of percussion instruments.

Different algorithms perform differently, and every algorithm has its own advantages and disadvantages. The performance of the algorithms is mainly affected by Wave->Pitch, Pitch->MIDI Recognition Settings pages and Equalizer.

Supplementary algorithms allow to identify additional features of the composition, and this helps Note Detection algorithms to work better.

Currently Supplementary algorithms include Tuning Determination and Tonality Determination. These algorithms are controlled from the Tuning and MIDI Options tabs of the Recognition Settings and from Scale Mask Equalizer tab.

Some properties of the resulting MIDI file that are not detected automatically are set based on the MIDI Options tab.

Supplementary algorithms works together with Note Detection algorithms, and resulting recognition sequence can be very complex.

Here is the list of undertaken operations if Try to detect Global Tuning, Determine Key Automatically and Use Recognized Key features are turned on:

•	Determine Global Tuning (this will be used by Note Detection algorithms)

•	Recognize preliminary with selected Note Detection algorithm using flat Scale Mask

•	Determine Tonality

•	Set Scale Mask according to detected Tonality

•	Re-recognize with selected Note Detection algorithm using defined Scale Mask

•	Launch TrueTone Editor or MIDI Window with recognition result.

In less complex situations, the sequence is shorter, for example, if Determine Key Automatically is off, steps 2-4 have no meaning.

Recognition of a particular piece is an iterative process, usually you start from some predefined recognition settings and then adjust them evaluating previous recognition attempts. After several attempts you find some good setup when you cannot achieve better quality by modifying parameters. At this point you take the best recognized variant and then export it or edit it manually. Probably, the best way to do editing and correction is the internal TrueTone Editor available in WIDI Professional.