The Important Role Of The Attack/ Decay Transients In Music Instrument Recognition

Intro
It has long been known that the onset of a note, the attack, plays an important role in our perception of timbre. The attack transient components in music instruments have been known to contain a vast amount of information about the music instrument.

There are hundreds of different types of music instrument on the surface of planet earth today. Each of those music instruments has its own unique characteristic that distinguishes it from another. Looking at the music instruments signal from a temporal perspective we can divide the signal into four main elements; attack, decay, sustain, and retard (release). The main area of focus for instrument recognition has always been the so called ‘steady-state’ (sustain). The reason for this is that’s easy to analyze.

It needs to be addressed that doing a study about the attack transient is very difficult due to its short period of time and its rapid changes. Therefore it comes as no surprise that attack transients are not well understood nor well represented within any of the available analysis-resynthesis models.

Frequency Analyzers
The first step in understanding your sound as a graphical representation is having the right frequency analyzer. As we all know from my earlier post about the frequency analyzer is that it truncates the data (audio) coming in. Not only is this a problem when we try to analyze attack transients, the other important part is the response time of the frequency analyzer which is at a fixed rate also named ‘hop-distance’. It is the analyzers primary task to detect and locate the so called ‘partials‘ or in better understanding language: the sinusoidal elements that compose the harmonic structure of a pitched sound. Therefore to gain the highest possible frequency resolution it is better to have a long analysis window, but for time resolution this is the complete opposite.

Because of the rapid changes of the attack transients –the note’s onset, and the analyzer’s relative slow response time (frames), it will create distortion or in the worst case a drop-out on synthesis as its algorithm can’t handle the amount of incoming energy from the percussive attack transients and therefore create artifacts upon synthesis such as phase discontinuities; which happens in the high frequency spectrum.

Transient Designer As Re-Tuning Device
The resynthesis of the attack transient results in a better perception of pitch. Unfortunate enough NO ONE has adapted a study into this, but you can use this little bit of knowledge in advantage during mixdown. Think about it.
Having that said it is thus possible to ‘re-tune’ audio programs with a dedicated transient designer (Imagine applying (pre-)EQ to this).

Outro
However during the mixdown phase it’s quite uncommon to see a mixing engineer grab for a transient designer to resynthesis the audio program. The reason for this can be found in our tools, starting off with our frequency analyzers. Another important part is the role of synthesizer programmers in this issue. As many engineers still see transient designing or envelope generating as a task for the ‘MIDI guys’.

Signing off,

Yours truly