My main works revolve around exploring the past and discovering thier languages, how they lived and what were the events that led to thier destruction I find these concepts really facinating though I have a certain way of doing things,
I'm sometimes not quite sure what
I'm doing with my music I feel like I have too many ideas and I'm too strict on what
I want to work on, part of the reason
I think is because I don't recieve any kind of constructive criticism or feedback
on my work in short I'm looking for bad reviews because that way I will atleast have a different point of view on my music other than my own. This next track is called 'The Fog Of War' which is coming out as a part of a compilation Ambient Layers Vol.II on 7K.
I always wanted to play my Harmonium with my two hands so a few weeks earlier an idea very spontaniously came down that if I could install my spare 8000 rpm cpu fan at the back of my harmonium I could play it with both
my hands without having to vello it.
Wrapping this up mix or whatever you may call this, another unreleased track called 'Forgotten Waters' it was composed only using the harmonium and running the cpu through very high resitance wires and will be out via another compilation on Swiss noise label Czarnagora I don't know how
it's pronouced pardon my language.
Artistically, this is probably our most exciting conception of I/R and convolution. We can have an infinity of spectral combination in a very quick process – drag and drop an audio in a plugin – which encourages experimentation. It is a way to enlarge sound textures without any synthesis. But, as we will now see, both the convolution and the Impulse/Response technics have problems and limits.
⒮⒪⒨⒠ ⒤⒮⒮⒰⒠⒮
The multiplicative nature of the convolution process makes it unable to create spectral content that is not already present in the incoming source. Convolution is thus a linear process, in opposition to the non-linearities added by processing like saturation, distortion or wave-shaping. Let say we have a 440Hz sine wave that gets convolved with a 3-second-long pink noise, what will result is the sine wave multiplied by the amplitude of the 440Hz present in the pink noise. All the other frequencies of the pink noise are multiplied by nothing (0) and are thus null. This means that the spectral complexity of the sounds used for convolution is an important parameter to keep in mind. Convolution is usually a more “substrative” process, ending up with less spectral diversity than the source before convolution. This multiplicative nature also brings another common issue: the overamplification of the low-end and resonant frequencies that exist in both the source and the offline sample. Since equilibrated sounds usually contains more low-end energy than high end, multiplying two equilibrated sounds tends to over accentuate the low end. It creates an undefined rumbling effect than is rarely pertinent. Sometimes, resonant frequencies are overwhelming, and, in that case, I tend to use a dynamic filter to attenuate them only when they exceed a chosen threshold. In general, there are a few technics to overcome those issue; either EQ the incoming source, EQ the output, use damping effects embedded in the convolution plugin or use a spectral compressor. To use a multiband compressor on the output is also another option. Another creative solution if to change the reading sample rate of one of the sounds to pitch it up or down, searching for a sweet spot where no frequencies are adding up too much. In Ableton’s Convolution reverb pro, this process (offline, rewriting the offline sample in a buffer at a different sample rate) is available by the parameter “Size”.
The last observation is the effect of convolution on the time domain. Because each input’s FFT windows source goes through each and all the offline sample’s window, it results in a high number of superposed layers that emulate what reflections could be. It creates this chorus-like effect and the sensation of deepness. It also tends (if the offline sample is more than one second long) to eliminate the dynamics of the incoming source. We have a result that tends to have the average of the dynamic and spectral repartition of both samples. In a way, we have a spectral and dynamic compression. For this very reason, it is harder to use this convolution process with rhythmical elements if we want to keep them rhythmical. They will be blended in themselves, often becoming lush, durational, and resonant. If the offline sample is short (less than 1 second), our rhythmical element can keep their dynamic, but will acquire the spectral resonance of the convolution, close to how a comb-filter would sound. It can definitively be used as an aesthetical treatment, but I admit that I prefer the sound of convolution with longer lasting sounds. Moreover, transient elements often contain very short and noisy parts, which the deconvolution process tends to destroy. In general, the linear nature of convolution/deconvolution doesn’t grasp easily sounds with a lot of non-linearities, saturation, distortion, and transients.