Exploring spectrum, understanding the importance of the parameters. Hi ! in this video, let’s explore the spectrum of a sound through different analysis parameters. Here is a bell sound: Let’s start and zoom on a part of the spectrum, one with the most activity. Then, pick the Magic Wand tool in the tool bar (upper left) to explore the sonagram. Just like the magic wand in Photoshop, the part you click on is a color zone with close colors. It means here : a spectrum zone where amplitudes are almost the same. Listen by pressing the space bar. Then select different zones by clicking and holding shift. and listen by pressing the space bar. What you hear is a resynthesis of the part of the sound. Resynthesis comes from an analysis made before, at the moment the sonagram was calculated, when the soft was launched. Now, let’s observe the difference of representation when changing the analysis parameters. Open the parameter window of the FFT sonagram with cmd-d. The main parameter to modify is the Window size. We had previously made an analysis with the default parameter : 4000 samples. With a shorter window (2000 samples), let’s cut the sound in more frequent slices. and detect more rythmic life inside the sound. Now notice that the frequency representation is quite blurry. Listen to the resynthesis of this zone of the sound with this representation: select this zone around 700 Hz with shift held down, and then press the space bar. here, you hear the sound as a vibration. With a quite long window (16000 samples), you slice the sound less often: and consequently represent the rhythmic variations with a lesser precision, but a much better way for the frequencies. Listen to the resynthesis of the same sound zone with this parameters: you cleary hear a rhythmic difference : the vibration is no more represented – at the same speed, at least. Anyway, you have a much better precision in the representation of the partials. So we see here that the analysis window size as a crucial importance for the sound representation. So how to get an ideal representation? It’s impossible to have temporal and spectral precision at the same time. It will depend on your musical use. Yet, on a sound like this one: it may be relevant to use the adaptative sonagram called 3AS. This algorithm allows to choose the window size automatically, depending on the local features of the sound signal. The sound is represented in a more optimal way in its rhythmic and resonant parts.