Hi guys, I have a project on the go. Basically I need to identify the qualities of audio input (e.g. is it low pitch, is it a short duration e.t.c) and then store them as recordings in a database. So the recording is classified and saved. something like Input 1 is high pitched and has a short duration, so save file in ‘high pitch and short duration’ folder. ( would this be more simple with embedding metdata?)
So my main question is what objects do I need in my MAX patch to identify audio like this?
Hi, do you need this identification in real time ? Or from an audio file ?
In realtime, you could start with analyzer~ e.g. (http://cnmat.berkeley.edu/downloads) based on Miller Puckette’s famous fiddle~. With it you can easily determine if pitches are low or high (regarding your definition of what is low and what is high and where is the limit between them).
Duration is another story, the easiest way is to divide the events regarding their attacks. But if the attacks are too soft you will miss some events. You will have to combine objects to considerer when a certain activity is on or off regarding the qualities determined by the previous analysis.