Short, surface-level, but ultimately accurate.
Tuesday, March 10, 2015
Tuesday, February 17, 2015
Sunday, February 8, 2015
I just realized I don't use musical instruments anymore. Not only are my guitars collecting dust, but even in my electronic music I'm not using virtual instruments much. All I do is edit audio and apply effects. I have become a fleshy granular synthesizer.
I only understand the extreme basics of microsound and granular synthesis, but the concept of a sonic "grain" and the microsonic (AKA, shorter-than-one-note) time scale is the best way to explain what I have been doing. I've known about microsound and used granular techniques for a number of years, but in recent months I have used them to the exclusion of almost everything else I know how to do, including drum programming. Interestingly, I'm also more productive. Check the dates on my Soundcloud posts.
Here is what I actually do.
This is the spoken sample I will start with. It is
6 beats long. You can see some of it's other vital statistics in the screen shot.
Now I am going to select a few very small slices of this sample from which to create new sounds. First I cut the sample up (Ardour's "split" command in this case--mouse over where I want to cut and press "s").
Now I will spread them out so I can hear what I have isolated in each slice.
I want the first one for "s" and "e" sounds, and the second to last one for "tuch." I will now move them into a different measure where I can work with them by themselves.
The amount of sound I am working with now totals one third of one second.
I'm demonstrating this with a real Ardour project and these are sounds I really did just finish a composition with. You can here it here. The following steps are what I would have done if I wanted to make that piece slightly longer (or add slightly more variety). I am going to compose an additional 7-beat bar using the two sounds selected above.
I am going to start by cutting these sounds up even smaller, then copying the most interesting of the smaller slices. I am going to do this experimentally. I will loop over the bar I copied them into and listen as I cut and move things around.
Looped, that beat sounds like this.
I want it to sound brighter, so I am going to turn off the fades at the beginning and end of each clip. Now it sounds like this.
If you ever wondered why some audio software puts tiny little fades-ins and fade-outs on audio clips, now you know. It gets rid of the clicks I just deliberately re-introduced to my composition.
Now I am going to experiment with the spacing and number of these sounds within the beat. That process sounds like this. My apologies if this just sounds like a bad minimalist composition, but you can hear each change I make.
Now the beat looks like this. What you heard at the end of the audio clip, played back at 242.1 bpm, looks like this. I moved the first four instances of the sound closer together and added two more, spaced out, later in the beat.
Now I am going to alter the volume of individual instances of the sound. Those changes sounds like this.
You probably clearly heard me make two of the sounds louder. Turning down some of the others had a subtler effect. Even when making "loud" music subtle changes can matter.
This is what the beat looks like now, with the gain/volume of the individual clips adjusted.
Back to beat 1, where I isolated and copied "er."
Right now it sounds like this.
What I really want is to emphasize the vowel, "e." One of the first things I need to do is get back down to one instance of the sound so I can really hear it. It turns out I cut it too short on the front. Because Ardour is a non-destructive editor, I just pull the front edge of the clip back to the left to get the earlier part of the sound.
Here is what trying to find and isolate it sounds like.
Still not as much "e" as I want. I could try editing together multiple instances of the vowel, but I don't think I can make it as smooth and "singing" as I want. I want a "note" to pop up out of the percussive, clicking and buzzing sounds I tend to create. I am going use Ardour's time stretching tool to actually make this sound longer. The way Ardour handles time stretching has so many possibilities I could do a whole essay on this process alone, but in the interest of time (wink, wink) just trust me. I stretched this sound.
It now looks like this.
...and sounds like this.
So, I got what I wanted, an extended vowel with slightly "sung" quality and a bonus percussive noise at the end. Woo! Hoo! New sound!
I am going to separate the "new" sound from the end and move it up to the front of the beat, repeat it, then fill up the rest of the beat with the "e." I will also adjust the volumes of the sounds. Here is what some parts of that process sounded like, including the end result.
I like my "e" so much I'm going to put a longer (second stretch process) version in another beat. Here I will also play with it's volume envelope. It looks like this, with it's volume envelope displayed.
If I stop now, this is what this bar would look
and sound like.
It needs more gravitas. Right now the "heaviest" sound I have is at the beginning of beat 1. I'm going try reusing that sound later in the bar to give the bar more weight.
Here is another process I am going to describe briefly but not spell out. I have found that I can create interesting sounds by stretching combined grains. Here I have three instances of the sound from the front of beat one, one of them reversed.
I am going to consolidate them into a single region and then stretch that region. After that I can use it whole, or cut it up further. It's just a kind of resampling. Here is what I got in this case.
The track is in 7/4, so a measure is a lot longer than a 4/4 measure at the same tempo, but each beat is the same length.
I should probably do a proper screen capture of this process but screen cap doesn't work well on my music rig. Maybe later, and with different sounds.
Ardour uses Rubber Band. If you are not getting the results you want from within Ardour you can also run Rubber Band on an individual files from the command line.
Ardour's term for a clip, in this case the visual representation of an audio file, or part of an audio file.