Written by Roland Trimmel | Categories: News
We've done some work over the past days to add many FAQs which explain the inner workings of Liquid Notes. Most importantly, there is one article which talks about how to follow a few simple rules when creating your arrangement to make it easier for Liquid Notes to deliver accurate results when detecting chords, tracks, etc.
Find more information why this is relevant hereafter.
Machine learning algorithms used in Liquid Notes
Liquid Notes' capability of being able to read a MIDI arrangement like an expert composer does is built on two sophisticated pieces of software: track type analysis, and harmonic analysis. Both use state of the art machine learning algorithms, and have been trained using a large set of data to correctly identify tracks, chords and harmony.
However, as with every piece of software there are limitations as to what a machine is able to do: what is obvious to the eye of a human being may not be the same for an algorithm. This is the reason why Liquid Notes expects the input (MIDI file) to be delivered in a certain way - e.g. tracks should be separated, and chord notes should be clearly identifiable as such.
Doing so ensures that both track type analysis and harmonic analysis will provide accurate results, no matter what style - i.e. complex classical music excluded, Liquid Notes wasn't designed for it!
Have a go and read this article in our FAQs to fully understand how to prepare your arrangement for optimal use with Liquid Notes!